{"id":5693,"date":"2026-09-23T13:42:17","date_gmt":"2026-09-23T13:42:17","guid":{"rendered":"https:\/\/brandmentions.com\/blog\/?p=5693&#038;preview=true&#038;preview_id=5693"},"modified":"2026-09-23T13:55:11","modified_gmt":"2026-09-23T13:55:11","slug":"find-subreddits-target-audience","status":"publish","type":"post","link":"https:\/\/brandmentions.com\/blog\/find-subreddits-target-audience\/","title":{"rendered":"How to Find Subreddits for Your Target Audience (2026): 4 Methods + Fit Score"},"content":{"rendered":"<p><em><strong>To find subreddits where your target audience is active, start with four discovery methods: search Reddit using your customers\u2019 problem language, run Google <code dir=\"ltr\">site:reddit.com<\/code> searches, map commenter overlap, and ask existing customers which communities they read. Then validate every candidate against the same 28-day sample.<\/strong><\/em><\/p>\n<p>That process usually produces 30 to 60 candidate communities. The Subreddit Fit Score then narrows them to the eight to twelve worth acting on by combining buyer fit, qualified intent, activity consistency, discussion quality, search visibility, and competitive room. Use weekly visitors and contributions to understand current activity, but use the buyer-fit-gated Fit Score to rank the final shortlist. Every community that survives should have a clear role, a named owner, and a date for re-scoring.<\/p>\n<p><strong>What is subreddit audience research?<\/strong> Subreddit audience research is the process of identifying and evaluating Reddit communities by who participates, which needs recur, how discussions develop, and what the rules permit. Its output is a ranked shortlist where every community has an assigned job, not a longer list of related names.<\/p>\n<p><img decoding=\"async\" src=\"https:\/\/brandmentions.com\/blog\/wp-content\/uploads\/2026\/09\/image_02_3b6c7791.webp\" alt=\"Six-step workflow for finding subreddits where target buyers participate\" \/><\/p>\n<h2 id=\"how-to-find-subreddits-for-your-target-audience-6-steps\">How to Find Subreddits for Your Target Audience: 6 Steps<\/h2>\n<ol>\n<li><strong>Write a one-page audience brief.<\/strong> Name who owns the problem, who champions the purchase, who controls budget, and who should not count.<\/li>\n<li><strong>Search Reddit in your customers' problem language.<\/strong> Collect seed communities plus the exact phrasing your buyers use, because every later method depends on that vocabulary.<\/li>\n<li><strong>Run Google site-search on 20 buyer questions.<\/strong> Use <code>site:reddit.com\/r\/ \"your phrase\"<\/code> to see which Reddit threads the outside world is actually shown.<\/li>\n<li><strong>Map commenter overlap from your smallest credible seed.<\/strong> Shared contributors surface adjacent communities your keyword list can never reach.<\/li>\n<li><strong>Interview five to fifteen customers with consent.<\/strong> Include the question no tool can answer: which communities do you read but never post in?<\/li>\n<li><strong>Score each candidate on a 28-day sample.<\/strong> Apply the Subreddit Fit Score and the Buyer-Fit Gate, then assign a tier, an action label, an owner and a re-score date.<\/li>\n<\/ol>\n<h2 id=\"summary-key-takeaways\">Summary: Key Takeaways<\/h2>\n<p><em>The direct answer to \"how do I find subreddits for my target audience\" is the bolded paragraph at the top of this page; the list below is the key-takeaways digest of the full guide.<\/em><\/p>\n<ul>\n<li>Reddit announced on <strong>September 9, 2025<\/strong> that it was replacing the member count in subreddit headers with activity-based indicators. The platform said it would replace the \"Member count\" with two new metrics based on a rolling 28-day average. Reddit's documentation defines Weekly Visitors as unique users who visited in the past 7 days based on a 28-day rolling average, and Weekly Contributions as the number of posts and comments made in the community in the past 7 days, excluding removed content.<\/li>\n<li>Subscriber counts did not become worthless, and they did not vanish everywhere. Reddit states that member count is currently available on the Mod Insights page and that this information will be removed in the future as it shifts focus to weekly visitors and contributions. Treat membership as historical context, never as your ranking column.<\/li>\n<li>Activity does not equal citation. A Semrush analysis of 217,000 unique prompts across Google AI Mode, Perplexity and ChatGPT Search identified 248,000 unique Reddit URLs cited or mentioned in those responses, and found that 80% of cited posts had fewer than 20 upvotes, 70% had fewer than 20 comments, cited posts averaged around 900 days old, and the median length was about 80 words. Small, old, specific threads get cited.<\/li>\n<li>Reddit's own scale keeps growing: global daily active uniques rose 18% year over year to 130.3 million in Q2 2026, and weekly active uniques increased 24% year over year to 514.6 million. Platform scale tells you the audience is probably there. It tells you nothing about which room they sit in.<\/li>\n<li>The reliable process is four discovery methods feeding one scored sample, then two separate decisions per community: whether to monitor it and whether to participate in it. A community can be your best research asset and still prohibit the post you wanted to write.<\/li>\n<li>Discovery decays. Overlap datasets are historical, moderators change rules, and activity shifts inside a 28-day window, so a shortlist without a re-score date is a guess with an expiry you have not noticed.<\/li>\n<\/ul>\n<h2 id=\"at-a-glance-which-method-fits-your-scenario\">At a Glance: Which Method Fits Your Scenario?<\/h2>\n<ul>\n<li><strong>Best for zero budget and a same-day shortlist:<\/strong> Reddit keyword search using customer problem language. No tools, no API access, and it gives you the vocabulary every other method depends on.<\/li>\n<li><strong>Best for search and AI-answer visibility:<\/strong> Google site-search across Reddit. It shows you which communities' threads are actually surfaced for buyer queries, which is a different question from which communities are busy.<\/li>\n<li><strong>Best for finding communities you would never guess by name:<\/strong> Commenter-overlap mapping. It connects communities by shared contributors rather than shared vocabulary, so it surfaces the adjacent professional rooms your keyword list cannot reach.<\/li>\n<li><strong>Best for teams with existing customers:<\/strong> Consent-based customer research. Highest confidence per hour, because you start from people who already bought instead of people who happen to use your words.<\/li>\n<li><strong>Best for deciding where the hours go:<\/strong> The Subreddit Fit Score with a buyer-fit gate. It stops a big, busy, search-visible community from outranking a small community full of your actual buyers.<\/li>\n<li><strong>Best for a program rather than a one-off audit:<\/strong> A monitoring layer keyed to buyer-intent phrases, with a monthly re-score of the core tier. Discovery is a project. Staying right is a routine.<\/li>\n<\/ul>\n<h2 id=\"whats-in-this-guide\">What's in This Guide<\/h2>\n<ol>\n<li><a href=\"#how-is-this-different-from-the-old-search-keywords-and-sort-by-size-method\">How Is This Different From the Old \"Search Keywords and Sort by Size\" Method?<\/a><\/li>\n<li><a href=\"#how-do-you-tell-a-buyer-community-from-a-topic-community\">How Do You Tell a Buyer Community From a Topic Community?<\/a><\/li>\n<li><a href=\"#how-to-find-relevant-subreddits-4-discovery-methods\">How to Find Relevant Subreddits: 4 Discovery Methods<\/a><\/li>\n<li><a href=\"#reddit-weekly-visitors-vs-subscriber-count-what-changed-and-what-it-means-for-audience-research\">Reddit Weekly Visitors vs. Subscriber Count: What Changed and What It Means for Audience Research<\/a><\/li>\n<li><a href=\"#should-you-prioritize-subreddits-that-show-up-in-google-and-ai-answers\">Should You Prioritize Subreddits That Show Up in Google and AI Answers?<\/a><\/li>\n<li><a href=\"#can-this-work-when-most-relevant-communities-ban-promotion\">Can This Work When Most Relevant Communities Ban Promotion?<\/a><\/li>\n<li><a href=\"#the-subreddit-fit-score-how-to-rank-subreddits-on-a-0-100-scale\">The Subreddit Fit Score: How to Rank Subreddits on a 0-100 Scale<\/a><\/li>\n<li><a href=\"#subreddit-tiers-and-action-labels-three-tiers-two-decisions-one-owner\">Subreddit Tiers and Action Labels: Three Tiers, Two Decisions, One Owner<\/a><\/li>\n<li><a href=\"#worked-example-a-social-listening-platform-finds-its-rooms\">Worked Example: A Social Listening Platform Finds Its Rooms<\/a><\/li>\n<li><a href=\"#what-changes-if-your-buyers-are-not-english-speaking-or-not-in-your-market\">What Changes If Your Buyers Are Not English-Speaking or Not in Your Market<\/a><\/li>\n<li><a href=\"#free-subreddit-finders-vs-paid-reddit-monitoring-platforms\">Free Subreddit Finders vs. Paid Reddit Monitoring Platforms<\/a><\/li>\n<li><a href=\"#turn-the-shortlist-into-a-living-dashboard\">Turn the Shortlist Into a Living Dashboard<\/a><\/li>\n<li><a href=\"#from-shortlist-to-paid-test\">From Shortlist to Paid Test<\/a><\/li>\n<li><a href=\"#the-30-day-action-plan\">The 30-Day Action Plan<\/a><\/li>\n<li><a href=\"#framework-glossary-the-named-rules-in-this-guide\">Framework Glossary: The Named Rules in This Guide<\/a><\/li>\n<li><a href=\"#frequently-asked-questions\">Frequently Asked Questions<\/a><\/li>\n<\/ol>\n<h2 id=\"how-is-this-different-from-the-old-search-keywords-and-sort-by-size-method\">How Is This Different From the Old \"Search Keywords and Sort by Size\" Method?<\/h2>\n<p><strong>The old method ranked communities by subscriber count; the current method ranks them by recent, first-person problem evidence from your target roles, measured on a consistent 28-day sample.<\/strong><\/p>\n<p>The old method treated discovery as a search problem. Type the category, read the results, pick the biggest.<\/p>\n<p>The current method treats it as an audience-mapping problem with an evidence requirement. You are not looking for communities <em>about<\/em> your category. You are looking for communities where people who own your problem describe it in their own words, repeatedly, recently.<\/p>\n<blockquote><p>A subreddit can be relevant to your category and still be irrelevant to your buyer.<\/p><\/blockquote>\n<p>That distinction matters because buyers rarely self-identify by product category. A communications lead does not spend her evenings in a community named after your software segment. She is asking about client reporting, crisis timing, agency staffing and approval workflows, in communities whose descriptions never mention your category.<\/p>\n<p>The second difference is cadence. The old method produced a list once. Moderator teams change, rules tighten, activity migrates, and overlap datasets age. Build the review into the process or the shortlist quietly rots while you keep posting into it.<\/p>\n<h2 id=\"how-do-you-tell-a-buyer-community-from-a-topic-community\">How Do You Tell a Buyer Community From a Topic Community?<\/h2>\n<p><strong>Look for responsibility, constraints and decisions, in that order.<\/strong><\/p>\n<p>\"Social media is changing\" is topic relevance. \"I handle reporting for six clients and I cannot separate campaign mentions from support complaints\" is a job, a constraint and a trigger. The second sentence is what you are hunting.<\/p>\n<p>Before you evaluate a single community, write a one-page audience brief with four answers:<\/p>\n<ul>\n<li><strong>Who owns the problem?<\/strong> The practitioner who feels the pain weekly.<\/li>\n<li><strong>Who champions the purchase internally?<\/strong> Often the same person, often not.<\/li>\n<li><strong>Who controls budget and who runs procurement?<\/strong> In B2B these are frequently different people in different communities, and a community full of hands-on users may contain almost none of them.<\/li>\n<li><strong>Who should not count?<\/strong> Vendors promoting products, students doing coursework, job seekers, and people using your keyword in an unrelated sense.<\/li>\n<\/ul>\n<p>Then separate three kinds of evidence when you read threads, because collapsing them is the most common scoring error I see:<\/p>\n<ol>\n<li><strong>Problem evidence:<\/strong> an unresolved task, described in the first person.<\/li>\n<li><strong>Evaluation evidence:<\/strong> comparisons, shortlists, \"what do you use for\" requests.<\/li>\n<li><strong>Purchase evidence:<\/strong> budget approval, procurement, migration, renewal deadlines.<\/li>\n<\/ol>\n<p>A community rich in problem evidence and empty of purchase evidence is a product-research asset. That is a real outcome, not a failure. Give it a research objective instead of judging it against a pipeline target.<\/p>\n<blockquote><p>Not every valuable community needs to generate leads. Some are valuable because they show you how buyers think.<\/p><\/blockquote>\n<h2 id=\"how-to-find-relevant-subreddits-4-discovery-methods\">How to Find Relevant Subreddits: 4 Discovery Methods<\/h2>\n<p>The four discovery methods are <strong>(1) Reddit keyword and problem-language search, (2) Google site-search across Reddit, (3) commenter-overlap mapping, and (4) consent-based customer research.<\/strong> This is the order I recommend running them, not a claim that one outperforms the others on conversions. Each answers a different question, and the fourth one exists to catch the errors of the first three.<\/p>\n<p>The session lengths below are planning allocations from my own audits, not measured completion times.<\/p>\n<table>\n<thead>\n<tr>\n<th>Method<\/th>\n<th>What it answers<\/th>\n<th>Complexity<\/th>\n<th>Suggested first session<\/th>\n<th>First useful output<\/th>\n<th>Main constraint<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Reddit keyword search<\/td>\n<td>Where is this problem discussed in customer language?<\/td>\n<td>Low<\/td>\n<td>60 minutes<\/td>\n<td>Seed communities plus a customer-phrase dictionary<\/td>\n<td>Limited by your starting vocabulary<\/td>\n<\/tr>\n<tr>\n<td>Google site-search<\/td>\n<td>Which Reddit discussions get surfaced for buyer queries?<\/td>\n<td>Low<\/td>\n<td>45 minutes<\/td>\n<td>Search-visible threads and extra candidates<\/td>\n<td>Results are not exhaustive<\/td>\n<\/tr>\n<tr>\n<td>Commenter-overlap mapping<\/td>\n<td>Where else do these contributors participate?<\/td>\n<td>Low<\/td>\n<td>45 minutes<\/td>\n<td>Adjacent, non-obvious candidates<\/td>\n<td>Historical datasets, commenter bias<\/td>\n<\/tr>\n<tr>\n<td>Consent-based customer research<\/td>\n<td>Where do real buyers actually read and ask?<\/td>\n<td>Moderate<\/td>\n<td>One interview round<\/td>\n<td>Customer-confirmed communities<\/td>\n<td>Small, self-selected sample<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h3 id=\"method-1-reddit-keyword-and-problem-language-search-best-for-a-free-first-pass\">Method 1: Reddit Keyword and Problem-Language Search (Best for a Free First Pass)<\/h3>\n<p>Start inside the platform, using <a href=\"https:\/\/redditinc.com\/news\/introducing-reddit-answers\" target=\"_blank\" rel=\"noopener noreferrer\">Reddit's own search and Answers experience<\/a>, which Reddit merged into a single search surface in a May 26, 2026 update. Native search is not a weak method, it is a differently scoped one: it reads the conversation as members see it.<\/p>\n<p>Scale justifies starting here. Reddit last disclosed that over 80 million people searched directly on Reddit every week as of Q4 2025. Your buyers are using this search box too, and the phrasing they use is the asset you are collecting.<\/p>\n<p><strong>How to execute it<\/strong><\/p>\n<ol>\n<li>Build a six-angle phrase matrix before you search anything: category terms, task terms, failure terms, decision terms, role terms, constraint terms.<\/li>\n<li>Run each phrase <strong>separately<\/strong> first. Combine later. If you start with a stacked query, you cannot tell which word killed the good result.<\/li>\n<li>Check the Communities tab, then the Posts tab sorted by New with a recent time filter. Relevance sort will hand you 2019 threads and quietly distort your read on current activity.<\/li>\n<li>Search comments as well as posts. Switching language lives in replies. \"We moved off X because\" is almost never a post title.<\/li>\n<li>Search <em>within<\/em> each promising community using its own vocabulary, and add any new phrases to your dictionary.<\/li>\n<li>Save one source thread beside every candidate. A row that says \"three discussions about client reporting in the last 28 days\" is evidence. \"Looks relevant\" is not.<\/li>\n<\/ol>\n<p>Structured queries help here. Apply <a href=\"https:\/\/brandmentions.com\/blog\/boolean-search\/\">Boolean search operators<\/a> consistently across your six angles, and test syntax per platform rather than assuming a query transfers unchanged from Google.<\/p>\n<p><strong>Two supplementary discovery leads inside Reddit itself.<\/strong> First, <strong>r\/findareddit<\/strong> exists precisely for \"which community should I ask this in\" requests, and searching its archive for your problem phrasing often surfaces community names your keyword matrix never produced; asking there yourself is also permitted, provided you describe the need rather than the product. Second, many communities maintain moderator-curated <strong>\"Related Communities\"<\/strong> sidebar modules, wiki pages or rule-page link lists. Those are human-selected neighbours rather than algorithmic ones, which makes them a useful cross-check on overlap maps in Method 3. Treat both as lead generators only: every name they produce still has to survive the 28-day sample.<\/p>\n<p><strong>Decision thresholds.<\/strong> Promote a community to the validation sample when it appears for at least two different phrase angles, ideally one problem angle and one decision angle. Category-only matches are often vendor echo chambers. Problem-only matches are often general help forums.<\/p>\n<p><strong>Failure modes.<\/strong> The Communities tab matches names and descriptions, so a community whose sidebar never mentions your category can discuss it daily and stay invisible. Result depth is capped per query, so long-tail communities fall off the bottom. Methods 2 and 3 exist to patch both gaps.<\/p>\n<p><strong>The non-obvious part.<\/strong> Search competitor names paired with \"alternative\" and \"vs,\" then record the <em>community<\/em>, not the thread. Switching conversations concentrate in a small number of rooms, and those rooms are disproportionately where evaluation happens. In several audits, that single query set pulled attention away from the largest generalist communities and toward smaller professional ones where people name tools, prices and constraints.<\/p>\n<h3 id=\"method-2-google-site-search-best-for-externally-visible-buying-conversations\">Method 2: Google Site-Search (Best for Externally Visible Buying Conversations)<\/h3>\n<p>Use <a href=\"https:\/\/developers.google.com\/search\/docs\/monitor-debug\/search-operators\/all-search-site\" target=\"_blank\" rel=\"noopener noreferrer\">Google's documented site: operator<\/a> to restrict results to Reddit, or to one subreddit, and see which discussions the outside world is actually shown.<\/p>\n<p><strong>How to execute it<\/strong><\/p>\n<ol>\n<li>Write 20 buyer questions in buyer phrasing, not category phrasing.<\/li>\n<li>Run each as <code>site:reddit.com\/r\/ \"your phrase\"<\/code>. Record subreddit, thread date, the question asked, and the query that found it.<\/li>\n<li>Open the threads. Snippets lie about recency and about whether the thread still exists.<\/li>\n<li>Re-run the same buyer question without the <code data-start=\"407\" data-end=\"414\">site:<\/code> restriction to see whether Reddit - and which community - surfaces in normal results. Restricted search finds conversations; unrestricted search provides point-in-time evidence of search visibility. Record the market, date and search conditions where practical, because results can vary by location, language and device.<\/li>\n<li>Freeze your query panel. Reuse the identical queries next month, or a change in your wording will look like a change in visibility.<\/li>\n<\/ol>\n<p><strong>Decision thresholds.<\/strong> Treat a community as a visibility anchor when it appears in unrestricted results for at least three of your 20 buyer questions. Restricted-only appearances are discovery leads, not visibility evidence.<\/p>\n<p><strong>Failure modes.<\/strong> Google states plainly that site: results are not exhaustive, so absence proves nothing about the index and even less about demand. The operator also cannot tell you whether a surfaced thread is current or abandoned.<\/p>\n<p><strong>The non-obvious part.<\/strong> Restricted and unrestricted search measure two different things, and conflating them is the most common measurement error in this entire workflow. Restricted search finds conversations. Unrestricted search shows whether that conversation appears for the query your buyer actually types at the time of the check. And never use Google's result count as a denominator for intent density. There is no valid population behind that number.<\/p>\n<p><strong>Reddit search vs. Google site-search, the core difference in execution:<\/strong> Reddit shows you the community as members experience it, including comment-level language. Google shows you the narrow subset it chooses to surface to outsiders. You need both readings, and you should record them in separate columns.<\/p>\n<h3 id=\"method-3-commenter-overlap-mapping-best-for-non-obvious-adjacent-communities\">Method 3: Commenter-Overlap Mapping (Best for Non-Obvious Adjacent Communities)<\/h3>\n<p>Overlap tools connect communities by shared contributors rather than shared words. The reference implementation is Andrei Kashcha's <a href=\"https:\/\/github.com\/anvaka\/sayit\" target=\"_blank\" rel=\"nofollow noopener noreferrer\">related-subreddit project<\/a>, which links communities using shared commenting activity, with documented manual overrides for some very large communities. The companion <a href=\"https:\/\/anvaka.org\/map-of-reddit\/\" target=\"_blank\" rel=\"nofollow noopener noreferrer\">Map of Reddit<\/a> positions roughly 116,000 communities so that distance reflects audience relationships.<\/p>\n<blockquote><p>Search finds communities by language. Overlap finds them by people.<\/p><\/blockquote>\n<p>Dates matter more than dataset size here. The sayit project documentation describes comments from <strong>August and September 2018<\/strong> across roughly 38 million user-subreddit records. Third-party finders advertise larger corpora, including one that claims more than 150 million comments analyzed without publishing a collection period. Any of these is a historical discovery aid. None is a current audience census.<\/p>\n<p><strong>How to execute it<\/strong><\/p>\n<ol>\n<li>Enter a seed community you have already supported with recent threads, not a community you assume is relevant.<\/li>\n<li>Record the connected communities in both directions: where these contributors also comment, and which communities point back.<\/li>\n<li>Cut connections with no plausible buyer, role or workflow relationship. Hobby and regional clusters usually fall here unless your brief says otherwise.<\/li>\n<li>Open each survivor and check the last 28 days of discussions. A 2018 relationship is a hypothesis, not a finding.<\/li>\n<li>Repeat from a seed representing a different buying role, then keep the union with provenance noted.<\/li>\n<li>Cross-check the machine-generated neighbours against the moderator-curated \"Related Communities\" sidebars of your seeds. Where an algorithmic link and a human link agree, you have a stronger hypothesis.<\/li>\n<\/ol>\n<p>The user-overlap technique has a research pedigree well beyond marketing tools, and it is worth understanding the assumptions behind it if you want defensible <a href=\"https:\/\/brandmentions.com\/blog\/academic-research-social-listening\/\">audience research methods<\/a> rather than a screenshot of a map.<\/p>\n<p><strong>Decision thresholds.<\/strong> Promote a candidate when two independent seeds point to it, or when one seed points to it and you can find at least three relevant recent discussions inside it. One weak link and no current evidence means Monitor tier at best.<\/p>\n<p><strong>Failure modes.<\/strong> Shared commenting observes people who post. It says nothing directly about silent readers, customer status or purchasing authority. Overlap similarity also degrades for the largest communities, which is documented in the sayit project itself, where popular subreddits connected to nearly everything and required manual overrides.<\/p>\n<p><strong>The non-obvious part.<\/strong> Start overlap mapping from your <strong>smallest credible seed<\/strong>, not your largest. Seed with a giant generalist community and you get noise shaped like insight. Seed with a narrow professional community and the neighborhood is legible. Also watch for a vendor conflict in this tool category: some free \"similar subreddit\" finders are operated by services that also sell upvotes, comments and accounts. Use the free utility if it helps, credit it honestly, and do not let convenience turn into an endorsement of vote manipulation.<\/p>\n<p><strong>The Crosspost Discount: count a repeated need once when you estimate total opportunity.<\/strong><\/p>\n<p>When shared participants ask the same question in three connected communities, you have three placement options and one underlying need. Keep all three community observations. Count one opportunity. Otherwise adjacent communities inflate your demand estimate exactly where overlap is highest.<\/p>\n<blockquote><p>Three conversations do not always mean three opportunities.<\/p><\/blockquote>\n<h3 id=\"method-4-consent-based-customer-research-best-for-correcting-the-other-three\">Method 4: Consent-Based Customer Research (Best for Correcting the Other Three)<\/h3>\n<p>Set the boundary before you start, using Reddit's <a href=\"https:\/\/support.reddithelp.com\/hc\/en-us\/articles\/42728983564564-Responsible-Builder-Policy\" target=\"_blank\" rel=\"noopener noreferrer\">Responsible Builder Policy<\/a>, which prohibits attempts to re-identify or de-anonymize Redditors by matching off-platform identifiers, and prohibits inferring sensitive characteristics.<\/p>\n<p>That rules out the version of \"reverse lookup\" that circulates in growth-hacking threads. You do not take a customer email list and hunt for pseudonymous accounts. You ask.<\/p>\n<p><strong>How to execute it<\/strong><\/p>\n<ol>\n<li>Recruit five to fifteen customers across different roles and use cases. Include one churned customer if you can.<\/li>\n<li>Ask five questions: Which communities do you read for work? Where did you last ask for a recommendation? Which do you read but never post in? Would you share a thread that was useful? Would you consent to us reviewing your public, work-related activity?<\/li>\n<li>Review only what they agreed to share.<\/li>\n<li>Aggregate by community and by task. Record community-level counts, not per-person dossiers.<\/li>\n<li>Strip account identifiers from the working shortlist once aggregation is done.<\/li>\n<\/ol>\n<p><strong>Decision thresholds.<\/strong> A community named independently by three or more customers in different roles goes straight into the validation sample regardless of its size. A community named once, with no supporting threads, does not.<\/p>\n<p><strong>Failure modes.<\/strong> Your customers are a selected group. They bought, they answered, and they are comfortable talking about tools. Their habits will over-represent enthusiast communities and under-represent the quiet buyer who never posts. Keep this method as one input, not as the ranking.<\/p>\n<p><strong>The non-obvious part.<\/strong> The highest-value question is the one no tool can answer: <em>which communities do you read but never post in?<\/em> Overlap graphs are built from contributors, so lurker-heavy communities are structurally invisible to them. In my experience, interview answers to that single question surface at least one community that no keyword query and no overlap map produced, and it is often where the buying committee reads rather than where practitioners argue.<\/p>\n<p><img decoding=\"async\" src=\"https:\/\/brandmentions.com\/blog\/wp-content\/uploads\/2026\/09\/image_03_04bdb175.webp\" alt=\"Bar chart comparing subreddit activity using contributions per thousand visitors\" \/><\/p>\n<h2 id=\"reddit-weekly-visitors-vs-subscriber-count-what-changed-and-what-it-means-for-audience-research\">Reddit Weekly Visitors vs. Subscriber Count: What Changed and What It Means for Audience Research<\/h2>\n<p><strong>Weekly Visitors counts unique users who visited a community in the past 7 days on a 28-day rolling average; Weekly Contributions counts posts and comments made in the community in the past 7 days, excluding removed content; historical subscriber count measures accumulated joins since launch and says nothing about whether anyone still reads or replies.<\/strong> Use the first two to assess current activity. Use historical member count only as context, and rank the final shortlist with the Buyer-Fit-gated Fit Score.<\/p>\n<p>The change everyone half-remembers is the metrics swap. The change that matters is the question you are now forced to answer.<\/p>\n<p>Reddit replaced total subscribers, online now, and rank size with weekly visitors, and the reason was blunt. Reddit said the member count does not fully represent true engagement, because users can post and comment without being a member of a community. That is the part marketers skipped for a decade.<\/p>\n<table>\n<thead>\n<tr>\n<th>Signal<\/th>\n<th>What Reddit measures<\/th>\n<th>What it does not establish<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Weekly Visitors<\/td>\n<td>Unique users who visited in the past 7 days, on a 28-day rolling average, excluding anonymous browsing on the native apps and bots<\/td>\n<td>Buyers, subscribers, or addressable ad reach<\/td>\n<\/tr>\n<tr>\n<td>Weekly Contributions<\/td>\n<td>Posts and comments made in the past 7 days, excluding removed content<\/td>\n<td>Unique contributors, or any purchase intent<\/td>\n<\/tr>\n<tr>\n<td>Historical member count<\/td>\n<td>Accumulated joins since the community launched<\/td>\n<td>Whether anyone still reads or replies<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Two practical constraints before you build a spreadsheet on this.<\/p>\n<p>First, availability is uneven. Reddit says these stats appear on community homepages and post pages on the native iOS and Android apps and the updated reddit.com site on desktop and mobile. Old Reddit shows neither set of numbers, so pull every candidate from the same interface on the same day or your comparisons are noise.<\/p>\n<p>Second, this is now governance infrastructure, not a cosmetic header change. As of March 31, 2026, Reddit actively enforces moderation limits based on Weekly Visitors: one person can moderate up to five high-traffic communities with more than 100,000 weekly visitors, subject to Reddit's stated exemptions. A metric wired into moderation policy is a metric worth taking seriously.<\/p>\n<h3 id=\"where-subscriber-counts-still-earn-their-place\">Where Subscriber Counts Still Earn Their Place<\/h3>\n<p>Do not throw the number away. Use it for three narrow jobs.<\/p>\n<p>Historical growth context, because archived membership curves tell you whether a community grew, plateaued or stalled, which no seven-day number can. Moderator diagnostics, since member count is currently available on the Mod Insights page. And sanity checks on very old communities, where a huge roster next to thin contributions is itself the finding.<\/p>\n<blockquote><p>Membership reflects who joined. Weekly contributions show how much posting and commenting activity happened this week. Use both as context, not as substitutes for buyer fit.<\/p><\/blockquote>\n<h3 id=\"a-real-comparison-same-roster-different-room\">A Real Comparison: Same Roster, Different Room<\/h3>\n<p>Here is the cleanest public observation I have seen of the gap, recorded days after the new metrics appeared. On September 16, 2025, data engineer Felipe Hoffa published the <a href=\"https:\/\/hoffa.medium.com\/bigquery-vs-snowflake-vs-databricks-which-subreddit-community-beats-385315515494\" target=\"_blank\" rel=\"nofollow noopener noreferrer\">subreddit activity numbers he pulled for competing data-platform communities<\/a>.<\/p>\n<table>\n<thead>\n<tr>\n<th>Subreddit (observed Sept 16, 2025)<\/th>\n<th>Subscribers<\/th>\n<th>Weekly visitors<\/th>\n<th>Weekly contributions<\/th>\n<th>Contributions per 1,000 visitors<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>r\/bigquery<\/td>\n<td>~19.6k<\/td>\n<td>7.5k<\/td>\n<td>33<\/td>\n<td>4.4<\/td>\n<\/tr>\n<tr>\n<td>r\/snowflake<\/td>\n<td>~18.9k<\/td>\n<td>16k<\/td>\n<td>172<\/td>\n<td>10.8<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>By membership, these two communities looked interchangeable, and r\/bigquery looked marginally bigger. By weekly visitors, one was more than twice as busy. By weekly contributions, more than five times as busy. A subscriber-sorted shortlist would have put them in the same tier and been wrong.<\/p>\n<p>That last column is worth adopting, with one caveat that most guides get wrong.<\/p>\n<p><strong>The Two-Clock Rule: contributions per 1,000 visitors is a descriptive activity indicator used to compare candidate subreddits against each other, never an engagement rate - because Weekly Visitors is a 28-day smoothed count of people and Weekly Contributions is a 7-day count of actions.<\/strong><\/p>\n<p>Weekly Visitors is smoothed over 28 days. Weekly Contributions counts the last seven. One is people, the other is actions, and one person can contribute fifty times. So use contributions per 1,000 visitors to compare candidates against each other, and never label it \"the percentage of members who engage.\"<\/p>\n<p>This is a two-community observation from a single date, not a benchmark. Pull your own. The pattern repeats in every niche I have audited: the roster and the room are different things, and Reddit now shows you the room.<\/p>\n<h2 id=\"should-you-prioritize-subreddits-that-show-up-in-google-and-ai-answers\">Should You Prioritize Subreddits That Show Up in Google and AI Answers?<\/h2>\n<p>Include search visibility. Keep it below buyer fit.<\/p>\n<p>Google's documentation is specific about eligibility, not guarantees: pages need to be indexed and eligible to appear in a search result snippet to show up as supporting links in <a href=\"https:\/\/developers.google.com\/search\/docs\/appearance\/ai-features\" target=\"_blank\" rel=\"noopener noreferrer\">AI Overviews and AI Mode<\/a>. There is no additional technical lever that buys inclusion.<\/p>\n<p>The citation pattern itself is counter-intuitive, and it should change which threads you value. In that 248,000-URL analysis, the most frequently cited Reddit threads were not the viral ones, they were often older and brief with limited activity, with median upvotes ranging from 5 to 8 and median comments from 11 to 19. Among the Reddit URLs cited in the study, Q&amp;A threads were disproportionately common, accounting for more than half of cited content, followed by comparison and discussion posts.<\/p>\n<p>Read that again if you have been chasing large subreddits. The study shows that high engagement was not required for citation: many cited URLs were older, relatively small threads, and Q&amp;A formats were common among them. This is also the mechanism behind <a href=\"https:\/\/brandmentions.com\/blog\/why-ai-cites-reddit\/\">why AI cites Reddit<\/a> so often for commercial-intent questions: the thread structure often resembles the question-and-answer formats that were common among cited Reddit URLs in the study.<\/p>\n<p>Be careful with headline citation percentages, because the denominators conflict. Studies that report Reddit as a top-ranked source among the most-cited domains are measuring share of citations. A separate sample of 2,010 Google AI Overview responses collected between June 24 and July 23, 2026 by the citation tracker AmICited found Reddit cited in 0.9% of answers while still ranking eighth out of 1,037 domains. Both can be true. \"Most-cited domain\" and \"appears in most answers\" are different claims, and I would not build a budget on the second one.<\/p>\n<p>Finally, do not let an AI answer stand in for a sample. The September 13, 2026 preprint <a href=\"https:\/\/arxiv.org\/abs\/2609.14575\" target=\"_blank\" rel=\"noopener noreferrer\">The Wisdom of the Loudest<\/a> audited Reddit Answers with 10,000 queries drawn from 20 advice and support communities, repeated three times, and found retrieval favored already-visible top-level comments and varied between runs. That finding is about Reddit Answers specifically. It suggests a caution worth testing on other generative-search systems, but the study does not establish the same retrieval behavior elsewhere. In practice, use generated answers to find source threads, then read the threads yourself.<\/p>\n<p>Log three separate observations for every buyer query you test: the thread appears in ordinary search, an AI answer cites the thread, and the answer mentions your brand. Teams routinely report the first and claim the third. If you track <a href=\"https:\/\/brandmentions.com\/blog\/ai-visibility-guide\/\">AI visibility<\/a> seriously, keep those columns apart.<\/p>\n<h2 id=\"can-this-work-when-most-relevant-communities-ban-promotion\">Can This Work When Most Relevant Communities Ban Promotion?<\/h2>\n<p><strong>The Permission Ceiling is the rule that each community's own rules cap what you are allowed to do there, so a high audience score can never override a prohibited action.<\/strong> Yes, the workflow still works - and the restriction is partly why the signal is clean. When vendors cannot pitch, buyers ask each other, and you get unfiltered requirement language.<\/p>\n<p>Reddit's Reddiquette has long carried a 9:1 rule of thumb about submissions, meaning roughly one in ten of your posts should be your own content. Read it as a submission etiquette guideline, not as a permission slip that unlocks promotion after nine helpful comments. Individual communities set stricter rules, and those rules are what actually govern you.<\/p>\n<p>So run permissions as a separate column, not as points inside a score.<\/p>\n<p><strong>The Permission Ceiling: a high audience score cannot override a prohibited action.<\/strong><\/p>\n<p>A community can score 90 for research value and still prohibit the exact post you had planned. That is not a downgrade of the community. It is a change to your intended action. The failure mode I see most often is a team averaging a rules violation into an attractive total and calling the result a strategy.<\/p>\n<p>One more warning that costs people accounts: do not infer permission from a surviving promotional comment. Moderation queues are uneven, and a post that has not been removed yet is not a precedent.<\/p>\n<p><img decoding=\"async\" src=\"https:\/\/brandmentions.com\/blog\/wp-content\/uploads\/2026\/09\/image_04_c78ffa98.webp\" alt=\"Buyer-Fit Gate scorecard for ranking subreddits by audience relevance\" \/><\/p>\n<h2 id=\"the-subreddit-fit-score-how-to-rank-subreddits-on-a-0-100-scale\">The Subreddit Fit Score: How to Rank Subreddits on a 0-100 Scale<\/h2>\n<p><strong>The Subreddit Fit Score is a 0-100 rating that weights six dimensions - buyer fit, qualified intent, activity consistency, discussion quality, search visibility and competitive room - and gates the result on buyer fit.<\/strong> It is a scored sample, not a vibe check.<\/p>\n<p>Discovery produces 30 to 60 candidates. This is how you get to the eight to twelve you will act on, in a way you can defend to a skeptical CFO.<\/p>\n<h3 id=\"step-1-collect-a-comparable-sample\">Step 1: Collect a comparable sample<\/h3>\n<p>Consistency here matters more than sophistication. For each candidate:<\/p>\n<ol>\n<li>Review up to <strong>30 consecutive non-pinned discussions from the previous 28 days<\/strong>, newest first. Consecutive, not Top. Sorting by Top over-represents popular threads and will bias every number downstream.<\/li>\n<li>Separate recurring automated or scheduled threads from member-started discussions.<\/li>\n<li>Read the post plus the relevant replies, not just titles. Intent language concentrates in comments.<\/li>\n<li>Assess reply depth only on threads that have had at least <strong>48 hours<\/strong> to develop, so a fresh thread does not look like a dead one.<\/li>\n<li>Record sampling date, how many you reviewed, and what you excluded.<\/li>\n<li>Mark the assessment <strong>provisional<\/strong> if you cannot collect the full sample, and never promote a provisional community to your core tier.<\/li>\n<\/ol>\n<p>Use keyword searches to <em>find<\/em> evidence. Use the chronological sample to <em>measure<\/em> how common it is. Searching \"recommend a tool\" and then reporting that the community is full of recommendation requests is circular, and it is the single most common flaw in published subreddit shortlists.<\/p>\n<blockquote><p>Discovery gives you possibilities. Sampling gives you evidence.<\/p><\/blockquote>\n<h3 id=\"step-2-record-activity-and-intent-separately\">Step 2: Record activity and intent separately<\/h3>\n<ul>\n<li><strong>Daily post velocity:<\/strong> eligible member-started posts divided by days observed.<\/li>\n<li><strong>Median comments per post:<\/strong> the middle value across sufficiently mature threads.<\/li>\n<li><strong>Substantive reply depth:<\/strong> replies adding experience, constraints, numbers or a useful follow-up question.<\/li>\n<li><strong>Independent buyer breadth:<\/strong> distinct participants describing a relevant need, which is a better signal than raw thread count.<\/li>\n<li><strong>Qualified-intent density:<\/strong> sampled threads containing explicit evaluation or purchase evidence, divided by total sampled threads.<\/li>\n<\/ul>\n<h3 id=\"step-3-run-the-intent-language-checklist\">Step 3: Run the intent-language checklist<\/h3>\n<p>These are classification examples, not automatic lead labels. Read the surrounding conversation, because a vendor can phrase an advertisement as a question.<\/p>\n<table>\n<thead>\n<tr>\n<th>Signal<\/th>\n<th>Example language<\/th>\n<th>Classify as<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Unresolved task<\/td>\n<td>\"We still compile this by hand every Monday\"<\/td>\n<td>Problem evidence<\/td>\n<\/tr>\n<tr>\n<td>Recommendation request<\/td>\n<td>\"What do you use for tracking this?\"<\/td>\n<td>Evaluation<\/td>\n<\/tr>\n<tr>\n<td>Comparison or switching<\/td>\n<td>\"Has anyone moved from X to Y?\"<\/td>\n<td>Evaluation or migration<\/td>\n<\/tr>\n<tr>\n<td>Budget<\/td>\n<td>\"We have sign-off for next quarter\"<\/td>\n<td>Purchase evidence<\/td>\n<\/tr>\n<tr>\n<td>Deadline<\/td>\n<td>\"We need this before the launch\"<\/td>\n<td>Urgency<\/td>\n<\/tr>\n<tr>\n<td>Requirement<\/td>\n<td>\"Must handle multiple client workspaces\"<\/td>\n<td>Qualification context<\/td>\n<\/tr>\n<tr>\n<td>Supplier promotion<\/td>\n<td>\"I built a tool that does this\"<\/td>\n<td>Supply, not demand<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Two detection notes on authenticity. Coordinated promotion tends to show shallow account history, a single-category comment pattern, and identically framed recommendations across unrelated communities. When a positive recommendation carries no constraints, no trade-offs and no follow-up questions, treat it as unverified rather than as buyer demand.<\/p>\n<h3 id=\"step-4-score-six-dimensions-then-gate-on-fit\">Step 4: Score six dimensions, then gate on fit<\/h3>\n<p>Rate each dimension <strong>0, 1 or 2<\/strong>. Total score equals the sum of each dimension's weight multiplied by its rating, divided by two. Maximum 100.<\/p>\n<table>\n<thead>\n<tr>\n<th>Dimension<\/th>\n<th align=\"right\">Weight<\/th>\n<th>0<\/th>\n<th>1<\/th>\n<th>2<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Buyer fit<\/td>\n<td align=\"right\">20<\/td>\n<td>No target-role context in the sample<\/td>\n<td>Occasional or mixed fit<\/td>\n<td>Target-role context in most sampled threads<\/td>\n<\/tr>\n<tr>\n<td>Qualified intent<\/td>\n<td align=\"right\">30<\/td>\n<td>Fewer than 2 qualifying threads<\/td>\n<td>2 to 4 qualifying threads<\/td>\n<td>5 or more qualifying threads<\/td>\n<\/tr>\n<tr>\n<td>Activity consistency<\/td>\n<td align=\"right\">20<\/td>\n<td>No recent member-started discussions<\/td>\n<td>Activity in one observed week<\/td>\n<td>Activity across at least two observed weeks<\/td>\n<\/tr>\n<tr>\n<td>Discussion quality<\/td>\n<td align=\"right\">10<\/td>\n<td>Median of 0 substantive outside replies<\/td>\n<td>Median of 1 to 2<\/td>\n<td>Median of 3 or more<\/td>\n<\/tr>\n<tr>\n<td>Search visibility<\/td>\n<td align=\"right\">10<\/td>\n<td>Not found in your frozen query panel<\/td>\n<td>Found only via restricted search<\/td>\n<td>Found via unrestricted buyer queries<\/td>\n<\/tr>\n<tr>\n<td>Competitive room<\/td>\n<td align=\"right\">10<\/td>\n<td>Promotional replies dominate<\/td>\n<td>Mixed independent and promotional<\/td>\n<td>Independent discussion plus identifiable unanswered needs<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Worked calculation. Ratings of 2, 2, 2, 1, 1, 1 produce 20 + 30 + 20 + 5 + 5 + 5 = <strong>85 points<\/strong>.<\/p>\n<p><strong>The Buyer-Fit Gate: a community that scores 0 on buyer fit cannot exceed Monitor tier, whatever its total.<\/strong><\/p>\n<p>This is the correction most scoring frameworks need. Without the gate, a huge, busy, search-visible community full of the wrong people can mathematically outrank a small community full of your buyers. Volume must never be allowed to rescue fit. For the same reason, an empty competitive field does not earn a 2 by default: require an identifiable unanswered need, because sometimes nobody is selling there for a good reason.<\/p>\n<h3 id=\"step-5-keep-an-evidence-packet-and-calibrate-the-scorer\">Step 5: Keep an evidence packet, and calibrate the scorer<\/h3>\n<p>For every scored community, store: observation dates, number of threads sampled and excluded, three to five source thread links, coded intent counts, activity figures with the interface you pulled them from, the current rule text, component ratings, the total, and the assigned action.<\/p>\n<p>Then run one calibration check that almost nobody does. Have a second person independently score two communities from the same sample. If their component ratings differ by more than one point, your definitions are too loose to be reused next quarter. Fix the definitions, not the scores.<\/p>\n<p>Be equally clear about what this model does <strong>not<\/strong> do. It does not predict conversion rates, ad performance or revenue. It ranks evidence quality and audience fit at a point in time. Change the weights and the order can shift, so publish your weights alongside your shortlist and re-check after your first month of real outcomes.<\/p>\n<blockquote><p>The score ranks evidence and audience fit. It does not predict business outcomes.<\/p><\/blockquote>\n<h3 id=\"step-6-handle-what-you-cannot-observe\">Step 6: Handle what you cannot observe<\/h3>\n<p>Missing data is not evidence of absence. Private and restricted communities, removed threads, deleted accounts and incomplete tool coverage all produce blanks that look like silence.<\/p>\n<p>Label those rows <strong>unobserved<\/strong>, not \"no demand.\" If two of your three discovery methods cannot see inside a community, the honest score is provisional and the honest action is a manual read next month.<\/p>\n<blockquote><p>An absent signal usually means your instrument could not reach it. Write \"unobserved\" in the cell and move on, instead of quietly downgrading a community you never actually sampled.<\/p><\/blockquote>\n<p><img decoding=\"async\" src=\"https:\/\/brandmentions.com\/blog\/wp-content\/uploads\/2026\/09\/image_05_54ca93d2.webp\" alt=\"Three-tier subreddit shortlist model with Core Secondary and Monitor actions\" \/><\/p>\n<h2 id=\"subreddit-tiers-and-action-labels-three-tiers-two-decisions-one-owner\">Subreddit Tiers and Action Labels: Three Tiers, Two Decisions, One Owner<\/h2>\n<p><strong>Subreddit tiers are Core, Secondary and Monitor, assigned by Fit Score and buyer fit; action labels are research, participation or paid-test candidate.<\/strong><\/p>\n<p>Tiers allocate attention. Action labels prevent the rule violations that get accounts banned.<\/p>\n<table>\n<thead>\n<tr>\n<th>Tier<\/th>\n<th align=\"right\">Starting threshold<\/th>\n<th>What belongs here<\/th>\n<th>Operating approach<\/th>\n<th>Re-score<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Core<\/td>\n<td align=\"right\">75 to 100, buyer fit above 0<\/td>\n<td>Strong fit, recurring qualified needs, complete sample<\/td>\n<td>Read on working days, one named owner<\/td>\n<td>Monthly<\/td>\n<\/tr>\n<tr>\n<td>Secondary<\/td>\n<td align=\"right\">50 to 74<\/td>\n<td>Relevant but narrower, less consistent or less direct<\/td>\n<td>Weekly review, contribute only where you add something<\/td>\n<td>Quarterly<\/td>\n<\/tr>\n<tr>\n<td>Monitor<\/td>\n<td align=\"right\">Below 50, or provisional<\/td>\n<td>Emerging, occasional or unobserved<\/td>\n<td>Selective keyword alerts only<\/td>\n<td>Every 6 months<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Then attach one of three action labels to every community: <strong>research<\/strong>, <strong>participation<\/strong>, or <strong>paid-test candidate<\/strong>. These thresholds are my operating recommendations, not platform facts. Narrow B2B categories usually need a lower Core cutoff, because the best communities in a specialist niche are small by design.<\/p>\n<p><strong>The Capacity Rule: Never carry more Core communities than your team can consistently review on working days; every Core community should have clear ownership.<\/strong>. As a starting point, one dedicated marketer can usually own up to three Core communities; larger teams should assign clear ownership rather than multiplying communities automatically.<\/p>\n<p>As a practical starting point, one dedicated marketer can manage up to three Core communities. For larger teams, set the number based on actual review capacity and named ownership rather than simply multiplying three communities by every person on the team.<\/p>\n<h2 id=\"worked-example-a-social-listening-platform-finds-its-rooms\">Worked Example: A Social Listening Platform Finds Its Rooms<\/h2>\n<p>Buyers here are social media managers, PR and communications professionals, agency owners, and growth teams. This is the exact scenario I know best, and it is instructive precisely because the most relevant communities are the most restrictive.<\/p>\n<p><strong>Seeds (Methods 1 and 2).<\/strong> Problem queries around missed complaints, client reporting, competitor tracking and monitoring alternatives, plus competitor-name and \"alternative to\" queries, consistently return r\/marketing, r\/socialmedia, r\/PublicRelations and r\/DigitalMarketing, with r\/SaaS appearing on comparison and pricing queries.<\/p>\n<p><strong>Adjacent candidates (Method 3).<\/strong> Overlap exploration from the narrowest seed, not the broadest, points toward agency-operations, community-management and small-business communities. Treat every one of these as a hypothesis until the 28-day sample confirms it.<\/p>\n<p><strong>Customer input (Method 4).<\/strong> Interviews reliably add at least one read-only community that no keyword or overlap method produced, and they tell you which room the budget owner reads rather than where practitioners argue.<\/p>\n<p><strong>Then the rule check changes everything.<\/strong> r\/marketing currently publishes a <a href=\"https:\/\/www.reddit.com\/r\/marketing\/\" target=\"_blank\" rel=\"noopener noreferrer\">zero-tolerance policy on advertising, self-promotion and spam<\/a> and also prohibits research surveys. r\/PublicRelations currently prohibits self-promotion and states that it does not accept posts discussing the products or services PR practitioners use. Read that second one carefully: it restricts <em>tool discussion<\/em>, not only pitching. Professional relevance created no permission at all.<\/p>\n<table>\n<thead>\n<tr>\n<th>Candidate<\/th>\n<th align=\"right\">Illustrative score<\/th>\n<th>Tier<\/th>\n<th>Research hypothesis<\/th>\n<th>Permitted action today<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>r\/socialmedia<\/td>\n<td align=\"right\">85<\/td>\n<td>Core<\/td>\n<td>Social-team monitoring and reporting workload<\/td>\n<td>Verify current rules, then assess participation<\/td>\n<\/tr>\n<tr>\n<td>r\/PublicRelations<\/td>\n<td align=\"right\">85<\/td>\n<td>Core<\/td>\n<td>Reputation workflows, crisis timing, approvals<\/td>\n<td>Observe only; do not initiate tool posts<\/td>\n<\/tr>\n<tr>\n<td>r\/marketing<\/td>\n<td align=\"right\">80<\/td>\n<td>Core<\/td>\n<td>Marketing operations and reporting requirements<\/td>\n<td>Research only; no promotion, no surveys<\/td>\n<\/tr>\n<tr>\n<td>r\/DigitalMarketing<\/td>\n<td align=\"right\">60<\/td>\n<td>Secondary<\/td>\n<td>Campaign measurement, agency workflows<\/td>\n<td>Investigate recurring needs weekly<\/td>\n<\/tr>\n<tr>\n<td>r\/SaaS<\/td>\n<td align=\"right\">60<\/td>\n<td>Secondary<\/td>\n<td>Founder-led evaluation and pricing questions<\/td>\n<td>Separate buyer questions from vendor promotion<\/td>\n<\/tr>\n<tr>\n<td>r\/smallbusiness<\/td>\n<td align=\"right\">45<\/td>\n<td>Monitor<\/td>\n<td>Occasional reputation and review complaints<\/td>\n<td>Narrow phrase alerts<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Scores above are illustrative calculations to show how the model behaves, not measured assessments of these communities. The two rule findings are current as of this guide's verification date, and rules change, so re-read them before you act.<\/p>\n<p>The decision is not \"join all six.\" It is: validate one participation candidate, run two as research communities, inspect two secondary candidates, monitor one broadly. That program fits one person's week, which is why it survives past month two.<\/p>\n<h2 id=\"what-changes-if-your-buyers-are-not-english-speaking-or-not-in-your-market\">What Changes If Your Buyers Are Not English-Speaking or Not in Your Market<\/h2>\n<p>This is the branch most guides skip entirely, and it breaks shortlists quietly.<\/p>\n<p>Run the six-angle phrase matrix <strong>in the buyer's language first<\/strong>, then translate back, not the other way round. Machine-translating your category term usually produces a phrase nobody types. Ask two native-speaking customers to correct your failure-language column before you search anything.<\/p>\n<p>Then apply three filters that only matter outside your home market:<\/p>\n<ul>\n<li><strong>Product availability.<\/strong> A community full of qualified need in a market where you cannot sell, price or support is a research community by definition. Label it that way.<\/li>\n<li><strong>Community structure.<\/strong> Some language markets concentrate professional discussion in one national community plus one or two global English ones. Expect fewer, larger rooms and adjust your Core count down, not up.<\/li>\n<li><strong>Company-size mismatch.<\/strong> Enterprise and procurement conversations migrate off Reddit in many markets. If a full discovery run produces very few relevant communities, treat that as a signal to test whether your buyers concentrate elsewhere rather than forcing Reddit into the channel mix.<\/li>\n<\/ul>\n<h2 id=\"free-subreddit-finders-vs-paid-reddit-monitoring-platforms\">Free Subreddit Finders vs. Paid Reddit Monitoring Platforms<\/h2>\n<p>A finder tells you where to look. A monitoring platform keeps watching after you have decided which conversations matter. Compare the job before you compare feature lists.<\/p>\n<table>\n<thead>\n<tr>\n<th>Option<\/th>\n<th>Cost model<\/th>\n<th>Appropriate use<\/th>\n<th>What you must still validate<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Reddit search<\/td>\n<td>Free<\/td>\n<td>Conversation-level discovery in customer language<\/td>\n<td>Buyer fit, query coverage, current rules<\/td>\n<\/tr>\n<tr>\n<td>Google site-search<\/td>\n<td>Free<\/td>\n<td>Externally visible thread discovery<\/td>\n<td>Recency, completeness, non-exhaustive results<\/td>\n<\/tr>\n<tr>\n<td>Map of Reddit \/ related-subreddit finder<\/td>\n<td>Free, open source<\/td>\n<td>Historical commenter-overlap exploration<\/td>\n<td>Dataset date, present-day activity<\/td>\n<\/tr>\n<tr>\n<td>r\/findareddit and moderator-curated \"Related Communities\" sidebars<\/td>\n<td>Free<\/td>\n<td>Human-suggested community names for a described need<\/td>\n<td>Whether the suggestion reflects current activity and rules<\/td>\n<\/tr>\n<tr>\n<td>Third-party \"similar subreddit\" finders<\/td>\n<td>Usually free<\/td>\n<td>Extra overlap candidates<\/td>\n<td>Collection period, and whether the operator also sells engagement<\/td>\n<\/tr>\n<tr>\n<td>F5Bot<\/td>\n<td>Free tier plus paid plans<\/td>\n<td>Keyword alerts on new posts and comments<\/td>\n<td>Plan limits and whether your use case is permitted<\/td>\n<\/tr>\n<tr>\n<td>BrandMentions<\/td>\n<td>Paid<\/td>\n<td>Best for pairing subreddit-level mention capture with wider-web and AI-answer reputation reporting in one place<\/td>\n<td>Query coverage, community-level grouping, report configuration<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Two licensing details that catch marketing teams out. F5Bot's own <a href=\"https:\/\/f5bot.com\/faq\" target=\"_blank\" rel=\"nofollow noopener noreferrer\">FAQ<\/a> states that the free tier exists for personal use and that free accounts cannot monitor very common words or phrases, with a paid tier removing that restriction. Keyword allowances depend on your account tier, and each tier also caps how many alerts a single keyword and a whole account can generate in 24 hours. Useful to know before you build a commercial workflow on a personal-use plan. On the plus side, it scans both submissions and comments, and you can restrict alerts to specific subreddits, which is genuinely handy for a validated shortlist.<\/p>\n<p>Second, check that your recommended tool still accepts customers. GummySearch, for years the default answer to this question, <a href=\"https:\/\/gummysearch.com\/closing-time\/\" target=\"_blank\" rel=\"nofollow noopener noreferrer\">closed new signups on November 30, 2025<\/a> while keeping limited arrangements for existing users. A curated comparison of <a href=\"https:\/\/brandmentions.com\/blog\/best-free-reddit-monitoring-tools\/\">free Reddit monitoring<\/a> options is worth a read before you pay for anything, because the free layer covers more of the discovery job than vendors like to admit.<\/p>\n<p>Before you buy any monitoring platform, run an acceptance test with your own queries rather than a demo dataset. Feed it three posts and three comments you already know exist. Measure false positives on your noisiest keyword. Time delivery. Ask how deleted content is handled, how far back history goes, and whether you can export source URLs. A polished dashboard cannot answer those questions for you.<\/p>\n<p><img decoding=\"async\" src=\"https:\/\/brandmentions.com\/blog\/wp-content\/uploads\/2026\/09\/image_06_f07c49b5.webp\" alt=\"Living Reddit monitoring dashboard for buyer intent and subreddit sentiment\" \/><\/p>\n<h2 id=\"turn-the-shortlist-into-a-living-dashboard\">Turn the Shortlist Into a Living Dashboard<\/h2>\n<p>A shortlist is a snapshot. The job after discovery is different: catch the few conversations that change a decision.<\/p>\n<blockquote><p>Discovery is a project. Staying right is a routine.<\/p><\/blockquote>\n<p>Build three monitoring groups, not one.<\/p>\n<ol>\n<li><strong>Category and problem phrases<\/strong>, to catch unmet needs before any brand enters the thread.<\/li>\n<li><strong>Brand and competitor names<\/strong>, for experiences, comparisons and switching.<\/li>\n<li><strong>Decision language<\/strong>, for budget, migration and requirement talk.<\/li>\n<\/ol>\n<p>For teams that need Reddit conversations alongside wider-web reputation reporting and AI-answer mentions in the same review, BrandMentions is one option designed to support that workflow. Its documentation covers public Reddit posts, comments and discussion threads as mention types, with saved report filters for keywords, sources, sentiment, language and country. Verify subreddit-level grouping during your trial, or maintain it with tags and exported source URLs, rather than assuming a native per-community view exists.<\/p>\n<p>Keep sentiment and intent in separate columns, always. \"This tool is frustrating\" is a support or reputation signal. \"What should we replace it with?\" is a buying signal. Collapsing them is how teams end up with a <a href=\"https:\/\/brandmentions.com\/blog\/sentiment-analysis\/\">sentiment analysis<\/a> chart that moves for reasons nobody can explain. Make <a href=\"https:\/\/brandmentions.com\/blog\/reddit-brand-mentions-tracking\/\">tracking Reddit mentions<\/a> a reviewed routine with an owner, not an unattended collection exercise.<\/p>\n<p>Your dashboard row should carry: subreddit, observation period, relevant mention volume, independent buyer needs, intent category, sentiment toward the specific issue, current rule status, permitted next action, source thread, owner, outcome.<\/p>\n<h3 id=\"data-governance-for-your-research-files\">Data governance for your research files<\/h3>\n<p>This part is boring and it protects you. Decide, in writing, before the first export.<\/p>\n<p>Store community-level aggregates as the default, and account-level notes only while a specific analysis is open. Set a retention period for raw exports and honor it. Restrict access to the people running the program. Delete customer-provided thread links when the research question closes. And do not upload identifiable discussion content into third-party analysis tools without checking what that vendor does with it, which is the same question you would ask about your own customer data.<\/p>\n<h2 id=\"from-shortlist-to-paid-test\">From Shortlist to Paid Test<\/h2>\n<p>A research score is not an advertising forecast. If you plan to spend, treat the handoff as its own small project.<\/p>\n<p>Check four things before you brief a campaign. Whether your Core communities are actually available as targeting options, because eligibility differs from relevance. Whether you are buying community targeting or interest targeting, since the second is a much broader audience wearing your shortlist's name. Which communities you will exclude, and why. And how you will handle overlapping reach, because adjacent communities share contributors and your deduplicated reach will be smaller than the sum of the parts.<\/p>\n<p>Then write the stop condition <strong>before<\/strong> launch. One predefined success metric, one budget ceiling, one date. A test without a stop condition is a subscription.<\/p>\n<h2 id=\"the-30-day-action-plan\">The 30-Day Action Plan<\/h2>\n<p>These are work allocations, not performance guarantees.<\/p>\n<table>\n<thead>\n<tr>\n<th>Days<\/th>\n<th>Work<\/th>\n<th>Required output<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>1 to 3<\/td>\n<td>Write the audience brief: roles, buying committee, triggers, exclusions<\/td>\n<td>One page, agreed with sales<\/td>\n<\/tr>\n<tr>\n<td>4 to 8<\/td>\n<td>Run Reddit and Google discovery, freeze your query panel<\/td>\n<td>Candidate sheet with source threads and provenance<\/td>\n<\/tr>\n<tr>\n<td>9 to 12<\/td>\n<td>Expand via overlap mapping and customer interviews<\/td>\n<td>Adjacent candidates plus new customer phrases<\/td>\n<\/tr>\n<tr>\n<td>13 to 19<\/td>\n<td>Collect the 30-thread samples, read every rule page<\/td>\n<td>Completed scorecards with evidence packets<\/td>\n<\/tr>\n<tr>\n<td>20 to 22<\/td>\n<td>Apply the Buyer-Fit Gate, assign tiers, action labels and owners<\/td>\n<td>Operational shortlist with capacity check<\/td>\n<\/tr>\n<tr>\n<td>23 to 27<\/td>\n<td>Configure monitoring, run the vendor acceptance test, make permitted contributions<\/td>\n<td>Logged observations, false-positive count<\/td>\n<\/tr>\n<tr>\n<td>28 to 30<\/td>\n<td>Review missed mentions, noise and useful conversations<\/td>\n<td>Revised queries, next month's priorities, re-score dates<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Set up <a href=\"https:\/\/brandmentions.com\/blog\/mentions-alerts\/\">buyer-intent alerts<\/a> around a decision that somebody owns. A purchasing question, a reputation risk and a research curiosity should not all trigger the same notification, because when everything is urgent the alerts get muted by week three.<\/p>\n<p>At the end of the month, measure <strong>useful opportunities reviewed per research hour<\/strong>, and keep it strictly separate from conversions. Count conversions only when your attribution evidence actually supports the claim.<\/p>\n<blockquote><p>Do not re-score from memory. Re-score from the same sampling protocol you used the first time, or you are measuring your own changing attention instead of the community.<\/p><\/blockquote>\n<h2 id=\"framework-glossary-the-named-rules-in-this-guide\">Framework Glossary: The Named Rules in This Guide<\/h2>\n<table>\n<thead>\n<tr>\n<th>Framework<\/th>\n<th>One-sentence definition<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td><strong>Subreddit Fit Score<\/strong><\/td>\n<td>A 0-100 rating that weights six dimensions - buyer fit, qualified intent, activity consistency, discussion quality, search visibility and competitive room - and gates the result on buyer fit.<\/td>\n<\/tr>\n<tr>\n<td><strong>Buyer-Fit Gate<\/strong><\/td>\n<td>A community that scores 0 on buyer fit cannot exceed Monitor tier, whatever its total.<\/td>\n<\/tr>\n<tr>\n<td><strong>Two-Clock Rule<\/strong><\/td>\n<td>Contributions per 1,000 visitors is a descriptive activity indicator for comparing candidates, never an engagement rate, because visitors are smoothed over 28 days and contributions cover 7.<\/td>\n<\/tr>\n<tr>\n<td><strong>Permission Ceiling<\/strong><\/td>\n<td>A high audience score cannot override a prohibited action; community rules cap what you may do, not how valuable the community is.<\/td>\n<\/tr>\n<tr>\n<td><strong>Crosspost Discount<\/strong><\/td>\n<td>Count a repeated need once when estimating total opportunity, even when it appears in three connected communities.<\/td>\n<\/tr>\n<tr>\n<td><strong>Capacity Rule<\/strong><\/td>\n<td>Never carry more Core communities than you have people who will read them on working days.<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h2 id=\"frequently-asked-questions\">Frequently Asked Questions<\/h2>\n<h3 id=\"how-do-i-find-subreddits-for-my-target-audience-for-free\">How do I find subreddits for my target audience for free?<\/h3>\n<p>Combine four free methods: Reddit search using your customers' problem language, Google <code>site:reddit.com<\/code> queries for your 20 buyer questions, a commenter-overlap map to find adjacent communities, and short customer interviews asking where they read and where they last asked for a recommendation. Then validate each candidate against a consistent sample of its last 28 days of discussions before you spend money on tools or ads.<\/p>\n<h3 id=\"how-do-i-find-niche-subreddits-for-my-product-or-business\">How do I find niche subreddits for my product or business?<\/h3>\n<p>Stop searching your product category and start searching the failure language of the job your product does, then expand sideways: run commenter-overlap mapping from your <strong>smallest<\/strong> credible seed community, because narrow seeds produce legible neighbourhoods while giant generalist seeds produce noise. Cross-check the algorithmic neighbours against moderator-curated \"Related Communities\" sidebars and r\/findareddit suggestions, since those are human-selected. In specialist niches, expect the best communities to be small by design, so lower your Core cutoff rather than defaulting to the biggest room.<\/p>\n<h3 id=\"how-do-i-know-if-a-subreddit-is-right-for-my-target-audience\">How do I know if a subreddit is right for my target audience?<\/h3>\n<p>Score it on a 28-day sample rather than on how relevant its name sounds: review up to 30 consecutive non-pinned, member-started discussions from the previous 28 days and count how many contain target-role context and explicit evaluation or purchase evidence. Apply the Buyer-Fit Gate - a community that scores 0 on buyer fit cannot exceed Monitor tier, whatever its total score - so a big, busy, search-visible community full of the wrong people never outranks a small one full of your buyers. Then check the rules separately, because audience value and permission to post are two different columns.<\/p>\n<h3 id=\"what-are-the-best-free-tools-to-find-related-subreddits\">What are the best free tools to find related subreddits?<\/h3>\n<p>The free stack that covers most of the job is <a href=\"https:\/\/anvaka.org\/map-of-reddit\/\" target=\"_blank\" rel=\"nofollow noopener noreferrer\">Map of Reddit<\/a> and the open-source <a href=\"https:\/\/github.com\/anvaka\/sayit\" target=\"_blank\" rel=\"nofollow noopener noreferrer\">related-subreddit project<\/a> for commenter-overlap exploration, r\/findareddit for human suggestions when you can describe the need but not name the room, and the moderator-curated \"Related Communities\" sidebar modules and wiki pages inside your seed communities. Reddit's own search and Google's <code>site:<\/code> operator cost nothing and do the heaviest lifting. Remember that overlap datasets are historical - the sayit documentation describes comments from August and September 2018 - so every name they return still needs a current 28-day activity check.<\/p>\n<h3 id=\"how-do-i-check-if-a-subreddit-is-active\">How do I check if a subreddit is active?<\/h3>\n<p>Read the three activity numbers together on the community homepage: weekly visitors (unique users who visited in the past 7 days, on a 28-day rolling average), weekly contributions (posts and comments made in the past 7 days, excluding removed content), and contributions per 1,000 visitors, which you calculate yourself to compare candidates. Then confirm with a chronological read of the last 28 days of non-pinned, member-started posts, checking that threads receive substantive outside replies after at least 48 hours. Pull every candidate from the same interface on the same day, because Old Reddit does not display these metrics at all.<\/p>\n<h3 id=\"how-do-i-use-google-to-find-subreddits\">How do I use Google to find subreddits?<\/h3>\n<p>Write 20 buyer questions in your customers' phrasing, then run each one as <code>site:reddit.com\/r\/ \"your phrase\"<\/code> and record the subreddit, the thread date, the question asked and the query that found it. Re-run the same question without the <code data-start=\"4298\" data-end=\"4305\">site:<\/code> restriction to see whether Reddit - and which community - surfaces in normal results. Restricted search finds conversations; unrestricted search provides point-in-time evidence of search visibility. Google states that <code>site:<\/code> results are not exhaustive, so absence is never evidence of no demand, and result counts are not a valid denominator for anything.<\/p>\n<h3 id=\"did-reddit-remove-subscriber-counts-and-can-i-still-see-them\">Did Reddit remove subscriber counts, and can I still see them?<\/h3>\n<p>Reddit announced the change on September 9, 2025. Community pages now show weekly visitors, defined as unique users who visited in the past 7 days on a 28-day rolling average, and weekly contributions, defined as posts and comments made in the past 7 days. Member count is currently available to moderators on the Mod Insights page, and Reddit has said it will be removed in the future. Historical subscriber figures also survive in archived datasets, so treat any current member count you see as needing a date and a source.<\/p>\n<h3 id=\"how-many-subreddits-should-a-small-marketing-team-actively-participate-in\">How many subreddits should a small marketing team actively participate in?<\/h3>\n<p>Start with up to three Core communities for one dedicated marketer. Larger teams can cover more, but every Core community should have a named owner who can review it consistently on working days. Everything else belongs in a secondary tier reviewed weekly or a monitor tier with selective keyword alerts. Sporadic posting across ten communities reads as promotional to both moderators and members, which is the outcome you are trying to avoid.<\/p>\n<h3 id=\"can-a-subreddit-be-worth-tracking-if-i-am-not-allowed-to-promote-there\">Can a subreddit be worth tracking if I am not allowed to promote there?<\/h3>\n<p>Yes, and often those are the most useful ones. Assign it a research objective: requirement language, objections, workflow constraints, competitor complaints. Keep \"research value\" and \"permission to participate\" in separate columns, because a community can prohibit vendor posts, surveys or tool discussions while still being the best place to learn how your buyers describe the problem.<\/p>\n<h2 id=\"conclusion-build-a-shortlist-you-can-defend\">Conclusion: Build a Shortlist You Can Defend<\/h2>\n<p>Reddit made a quiet decision in September 2025 that most teams have still not absorbed. The numbers on the page now emphasize recent community activity rather than accumulated membership. That single change invalidated every subreddit list built by sorting on size, including the one sitting in your shared drive.<\/p>\n<p>The replacement is not a better tool. It is a discipline with three parts. Discover through more than one path so you find the rooms your keyword list cannot reach. Score a comparable sample instead of a feeling, and gate the score on buyer fit so volume never outranks relevance. Then give every community one job and one owner, and check the rules before you act, not after a removal.<\/p>\n<blockquote><p>Use search to discover. Use samples to measure. Use rules to decide what you can do.<\/p><\/blockquote>\n<p>The teams that win on Reddit are not the ones with the longest community list. They are the ones who can explain, with dates and source threads, why these six communities and not those twenty.<\/p>\n<p>Your next step takes twenty minutes. Pick your three most obvious communities, pull their weekly visitors and weekly contributions today from the same interface, calculate contributions per 1,000 visitors, and write those numbers beside whatever size figure you have been quoting internally. That one comparison usually ends the argument about which metric to rank on, and it starts the better conversation about who is actually in the room.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>A rigorous method for discovering, validating and ranking the subreddits where your buyers participate &#8211; four discovery methods, a Subreddit Fit Score, and a three-tier prioritization model.<\/p>\n","protected":false},"author":4,"featured_media":5687,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"_ef_editorial_meta_date_first-draft-date":"","_ef_editorial_meta_paragraph_assignment":"","_ef_editorial_meta_checkbox_needs-photo":"","_ef_editorial_meta_number_word-count":"","footnotes":""},"categories":[126],"tags":[290,293,292,291,288,95,289],"class_list":["post-5693","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-social-media-marketing","tag-audience-research","tag-b2b-marketing","tag-community-discovery","tag-reddit-2026","tag-reddit-marketing","tag-social-listening","tag-subreddit-research"],"acf":[],"_links":{"self":[{"href":"https:\/\/brandmentions.com\/blog\/wp-json\/wp\/v2\/posts\/5693","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/brandmentions.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/brandmentions.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/brandmentions.com\/blog\/wp-json\/wp\/v2\/users\/4"}],"replies":[{"embeddable":true,"href":"https:\/\/brandmentions.com\/blog\/wp-json\/wp\/v2\/comments?post=5693"}],"version-history":[{"count":8,"href":"https:\/\/brandmentions.com\/blog\/wp-json\/wp\/v2\/posts\/5693\/revisions"}],"predecessor-version":[{"id":5853,"href":"https:\/\/brandmentions.com\/blog\/wp-json\/wp\/v2\/posts\/5693\/revisions\/5853"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/brandmentions.com\/blog\/wp-json\/wp\/v2\/media\/5687"}],"wp:attachment":[{"href":"https:\/\/brandmentions.com\/blog\/wp-json\/wp\/v2\/media?parent=5693"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/brandmentions.com\/blog\/wp-json\/wp\/v2\/categories?post=5693"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/brandmentions.com\/blog\/wp-json\/wp\/v2\/tags?post=5693"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}