# TikTok SEO and the social search playbook by John Morabito · April 19, 2026 · 18 min read **TL;DR** - 74% of Gen Z uses TikTok as their go-to search tool. 64% as a search engine. 78% have purchased after seeing a product in a video. - TikTok SEO = keyword-rich captions, hashtags, on-screen text, voiceover keywords, metadata. - Social search feeds GEO. 58% correlation between TikTok views and Google branded search with a seven-day lag. - Facebook and Instagram are pay-to-play. TikTok and YouTube are still organic. TikTok is a search engine. Treat it like one. ## Why TikTok SEO matters now - Search behavior happens in-app. Users bypass Google. - Entertainment meets utility. Engaging content drives deeper discovery. - Low authority barrier. Algorithm favors relevance and engagement, not legacy domain authority. ## Core elements of TikTok SEO 1. **Hashtag and keyword optimization.** Intent-rich hashtags like `#best[product]`. Keywords in descriptions, on-screen text, captions, voiceovers. TikTok's index reads all of them including transcribed audio. 2. **Trend-aware, evergreen content.** Pair trending audio with keyword-optimized voiceover and on-screen text. 3. **Engagement signals.** Strong hooks in first two seconds, payoff by second ten, clear CTAs. 4. **Visual SEO.** Keyword overlay on thumbnails. Text overlays beat generic visuals. 5. **Platform-specific.** Shorter videos perform better. Attention spans roughly mirror age: 20-year-olds give you about 20 seconds. ## TikTok keyword research, step by step About 90 minutes per category: 1. **Search bar autosuggest mining.** Type the seed term one character at a time, log every autosuggest dropdown. Suggestions are ranked by real search volume. 2. **The "Others searched for" panel.** Mid-results module showing the related queries the algorithm wants to route. Log all of them. 3. **Creator Search Insights.** TikTok's official in-app tool. Shows searched topics by category and flags "content gap" topics where demand outstrips video supply. Content-gap topics are free rankings. 4. **Comment mining.** Recurring questions under the top five ranking videos are long-tail queries nobody has made videos for yet. Each one is a video brief. 5. **Competitor caption autopsy.** Pull captions, hashtags, and overlay text of the top ten videos per keyword. TikTok rewards literal phrase matches, so note exact phrasing. 6. **Map to a keyword sheet.** Territory, keyword, variants, content-gap flag, intent, video concept. Tactical note: TikTok rewards literal phrase matches far more than Google. Write the exact keyword into the voiceover, the overlay, and the caption. ### The weekly mining loop The 90-minute sprint is setup, not a system. A standing 30-minute loop every Monday: 1. **Re-run priority terms in the search bar.** Log autosuggest changes against last week. A new suggestion is new demand, weeks before any tool surfaces it. 2. **Pull fresh content-gap flags from Creator Search Insights.** Gaps get filled fast; the brand that checks weekly gets the free rankings. 3. **Harvest question-phrased comments off the past week's posts.** Every question under your own video is a query typed by a human who watched you and still could not find the answer. Tag each with the keyword it implies. 4. **Output: five to ten new video briefs into the production queue,** ranked content-gap first, then intent. This queue feeds the three-post weekly floor. No queue, no cadence, no rankings. ## Anatomy of a video that ranks - **Frame one is the thumbnail.** Keyword in large overlay text on frame one. Searchable beats clever. - **Say the keyword in the first two seconds.** Transcribed audio is indexed. - **Hook formulas:** numbered promise, contrarian open, result-first open, direct-address question. One per video. - **Payoff by second ten** or completion rate dies and the video never ranks. - **On-screen text carries keywords continuously.** Every section card is an indexed surface. - **Caption: keyword first, then context, then 3-5 hashtags.** One broad, two intent, one geo if local. Skip #fyp. - **End with a loop or comment bait.** Rewatches and comments are the strongest signals after completion. ## The transcript is the ranking surface TikTok auto-transcribes every video at upload, and search reads the transcript the way Google reads body copy. The caption is one text field. The transcript is the whole document. - **Say the exact target query verbatim in the first three seconds.** Not a paraphrase. "Here is how to clean suede shoes," not "Got scuffed suede? Watch this." Speech recognition indexes what you said, not what you meant. - **Repeat the head term two or three times, naturally.** Open, midpoint, close. More than that and you sound like a 2009 article and viewers bail. - **Put the query in speech AND on-screen text.** Speech recognition reads audio, text recognition reads overlays. Two separate indexed layers. Hit both, every time. - **Front-load proper nouns.** Brand, product, neighborhood names in the first sentence. Named entities anchor what the video is about, and earlier counts harder. - **Do not substitute slang for the keyword you want.** "These kicks" for thirty seconds ranks for nothing. Speak the term the way it gets typed. - **Audit your own transcript.** Turn on captions and read what the machine thinks you said. If your product name transcribes wrong, enunciate harder or carry it in the overlay. Tactical note: write the script before the shoot, with the target query placed at the open, midpoint, and close. A scripted 30 seconds out-ranks an improvised 60, because the transcript comes out already optimized. ## Comment seeding and reply farming Post a pinned comment restating the target query with added value (another indexed text layer). Reply to every real question with a video: the question sticker becomes your overlay keyword, the keyword research is already done, and the algorithm routes it to engaged viewers. Two reply-videos per week per account. ## Carousels: the underused search format Photo-mode carousels get indexed text on every slide (ten slides = ten keyword surfaces), cost near zero to produce, hold search placements longer than videos, and earn saves (a heavy ranking signal). Formats that work: ranked lists, price breakdowns, before/after sets, step-by-step tutorials, local roundups. ## Engineering the "others searched for" chips Each chip on a ranking video means the algorithm decided that video answers more than one search. That attachment is engineerable. - **One intent per video, two or three adjacent answers in the back half.** Main query owns seconds zero through fifteen; the back half picks off adjacent questions ("does it work on textured skin," "how much does it cost"). The transcript catches all of it and the video gets attached to each query it answers. - **Chips follow co-search behavior.** People who search A search B. A series on one theme makes your videos colonize each other's chips: five videos in a cluster and each one routes viewers to the other four. You become your own related-search results. - **Mine competitor chips as a content map.** Search the head term, open the top three videos, log every chip. Those are validated adjacent queries, free. One video per chip surrounds the head term instead of fighting for it head-on. ## Profile-level SEO The profile is indexed too. Every field is a keyword slot. - **Username and display name are two separate slots.** Username carries the brand, display name carries brand plus category keyword: "Brandname | NYC Med Spa." - **Write the bio in query language.** What you do, for whom, where. Keywords, not personality. The videos carry the personality. - **Pinned videos are category landing pages.** Three pin slots, one head term each. Pin the ranking video per term and swap in the new champion when one outperforms. - **Playlist names are indexed and searchable.** "Brooklyn dispensary tours," not "our faves." A playlist is the closest thing TikTok has to a category page. - **Route the link-in-bio to the page targeting the same query.** Searcher intent on TikTok should land on the matching site page, not the homepage. Same intent, same entity, consistent across surfaces, which is the association GEO feeds on. ## Local TikTok SEO Target "[category] + [neighborhood]" literally in voiceover, overlay, and caption. Tag the location on every post (feeds nearby/map surfaces). Stitch local discovery creators. Employee-face content outranks brand-polish content because local search on TikTok is trust search. The same budtender-as-search-asset logic runs through our [budtender SEO playbook](/playbooks/budtender-seo/), and the guardrails for regulated categories live in the [cannabis brand social media rules](/playbooks/cannabis-brand-social-media-rules/). ## Ranking TikToks in Google Google indexes TikToks and renders the caption as the result title, so write captions like search headlines. Long-tail, local, and how-to queries are the entry points. Cross-post to YouTube Shorts with the same keyword packaging: Google's AI Overviews cite YouTube more than any other domain, so one production hits three search surfaces. ## Cadence, iteration, and the 10-video test 1. Ship ten videos per keyword cluster over two weeks with different hooks and the same keyword packaging. The algorithm picks the winner; do not guess. 2. Check rank weekly in a logged-out session and log positions. Watch the "Search" traffic source in analytics. 3. When a video ranks, make four more on adjacent long-tails. Rankings cluster. 4. Refresh decayed winners with a new hook. Never re-post the identical video. 5. Floor cadence: three search-targeted posts per week. Carousels count. ## The repost-and-iterate loop The cadence section covers videos that ranked and faded. This is the other case: a video that holds a search placement but leaks viewers. The rank proved the query; the retention graph says the execution is wasting it. 1. **Diagnose with the retention graph, not the view count.** Find where viewers drop. That segment is the leak. The keyword packaging works, so do not touch it. 2. **Same hook, tighter cut, re-shoot.** Keep the caption phrasing, overlay text, and opening line identical. Cut the dip segments, move the payoff earlier, ship shorter. 3. **Fresh takes on proven queries rank fast.** TikTok search promotes new uploads on validated queries faster than it re-ranks old videos, which drag their historical engagement data around forever. 4. **Run kill-and-replace monthly.** Audit ranked videos, flag rank-but-leak cases, re-shoot the worst two or three, let old and new coexist two weeks, then judge. Common outcome: the old keeps its placement and the new takes a second slot. Two positions on one query for the cost of a re-edit. ## Measuring TikTok SEO - Native analytics: Viewers tab, Performance tab. - Manual keyword rank checks in the TikTok search bar for priority queries. - Audience demographic insights. For deeper rank tracking, competitor benchmarking, and trend discovery, purpose-built tools exist. ## Social search is the next frontier - 64% of Gen Z use TikTok as a search engine. - 78% of TikTok users have purchased after seeing a product in a video. - 70% of beauty discovery happens on social platforms. ## How TikTok feeds GEO Two ways. First, TikTok drives branded search. 58% correlation with a seven-day lag in a Winston client engagement. Branded search drives direct traffic. Direct traffic drives conversion. Second, AI systems ingest video. YouTube is the #1 citation source for Google's AI Overviews, outpacing Reddit. TikTok videos cross-posted to YouTube Shorts increasingly appear in AI citation layers. Winning on TikTok compounds across search surfaces. ## Hashtags are a relevance signal, not a reach lever The hashtag advice floating around is mostly cargo cult. Twenty hashtags, trending hashtags, #fyp stacked five deep. None of that buys reach. Hashtags do one job in search: they tell the classifier what topic the video belongs to. Treat them as a relevance signal, not a distribution lever. - **Three to five specific tags plus one broad tag.** The specific tags classify ("#suedecleaning," "#suedeshoes"); the broad tag ("#shoecare") parks the video in the larger neighborhood. Fifteen more tags dilute the signal, they do not add chances to rank. - **Match the hashtag to the spoken and on-screen query.** The classifier cross-checks layers. Voiceover, overlay, and hashtag agreeing classify the video with confidence. A contradictory trendy tag is noise the classifier discounts. - **Avoid banned, flagged, and dead hashtags.** Tap a tag before using it; check view count and recency of top posts. No recent ranking content means dead or suppressed. Skip it. - **Branded and campaign hashtags are for aggregation, not search.** A #yourbrandchallenge tag groups campaign content but does almost nothing for query search, because nobody types it into the search bar. Keep the jobs separate. - **Stop spending the hook on hashtags.** Lead the caption's first line with the literal query, not a wall of tags. Push hashtags to the end where they classify without crowding out the phrase you want to rank for. ## The Series and playlist play for binge-ranking Profile-level SEO covered naming a playlist with a head term. The deeper mechanic: a playlist (or paid Series) turns loose videos into a structure where one ranking video pulls the rest up. Grouping them changes the unit of competition from one video to one collection. - **The playlist name is an indexed, searchable surface.** Name it the way the head term gets typed ("Brooklyn coffee shop tours," "first-time edibles guide"). The playlist itself can surface for the query, separate from any single video. - **One ranking video pulls the collection.** When a searcher lands on a video in a playlist, TikTok surfaces the playlist and offers the next video. The strongest video becomes the front door; weaker videos inherit the traffic. - **Sequence by search intent, not post date.** Definition first ("what is X"), comparison second ("X vs Y"), how-to and cost third, local or branded last. Next-video autoplay walks the viewer down the funnel. - **Build the playlist before you have ten videos in it.** Create it at video two or three and keep adding. Waiting until the set is full means the early videos never got the binge structure. - **Series (the paid, gated format) is for the deepest evergreen demand.** Reserve the paid Series for the few topics with durable demand and a genuine teaching sequence. For most accounts the free playlist does the binge-ranking job. ## Duets and Stitches as a query-association tactic Local SEO mentioned stitching discovery creators in passing. The general tactic is bigger: a Stitch or Duet attaches your video to another, and TikTok reads that attachment as a topical relationship. Stitch a video that ranks for a query and you associate your content with that query, using someone else's ranking as a foothold. - **Stitch what already ranks for your target query.** Find the video at the top of your money keyword and stitch it. The original's query relevance bleeds onto yours. You are borrowing a ranking, not starting from zero. - **Pick the video by query, not by follower count.** A small creator's video that ranks for "how to deglaze a pan" beats a huge creator's video about nothing in particular. Rank relevance is what transfers. - **Add the answer, do not just react.** Extend or correct the original with your own keyword-rich payoff. A reaction with no added value gets no completion and no rank. - **Duet for side-by-side comparison content.** Duets run two videos in parallel, built for "X vs Y" and "does this actually work" formats. The comparison framing puts both the head term and the comparison query into your content. - **Stitch your own catalog to build internal association.** Stitching your older ranking video ties the new one to the same query cluster and signals the account owns the topic. It is the on-platform version of internal linking. ## Reading TikTok's search-intent signals in the analytics The measurement section is the manual baseline. This is the layer underneath: TikTok's analytics tell you which searches your videos already surface for and how search traffic behaves once it lands. - **Read the search query data, not just the traffic-source percentage.** The analytics show which searches your video surfaced for, frequently not the query you targeted. The gap between intended and actual query is your next batch of briefs. - **Segment retention by traffic source.** Search viewers arrived with intent; if they bail early, the video answered a different question than the one they typed. Filter retention to search traffic to learn whether the video satisfies the searcher. - **Watch for videos surfacing on queries you did not target.** An accidental ranking on an adjacent query is the classifier finding demand you missed. Make a deliberate video for that term while the account already has credibility on it. - **Double down on search-sourced winners, not view-count winners.** A search-sourced video is a durable ranking asset; a feed hit is a spike that decays. Allocate production behind search traffic even when the raw view count is smaller. - **Track the search share of traffic over time per topic.** Climbing search share means the account is winning the topic and deserves more videos. Flattening or falling means the rankings are decaying and need a refresh. ## TikTok Shop and the search-to-purchase loop For brands that sell physical product, TikTok Shop closes the loop the rest of this playbook opens. A product-tagged video can surface in search and carry the buy button in the same frame. The optimization is the same discipline: name the problem the way people search it, then attach the product to the answer. - **Product-tagged videos surface in search, so caption them like search videos.** Tagging a product does not exempt the video from the keyword rules. Voiceover, overlay, and caption still need the literal query. - **Name the product problem, not the product.** People search the problem ("how to fix frizzy hair," "gift for someone who has everything"), not your SKU. Build around the problem query and present the product as the resolution. - **Tag the product the video actually demonstrates.** The association between the on-screen demonstration and the tagged product earns the click. A generic tag on an unrelated demo does not convert. - **Build a shoppable answer for each money query.** Take the priority queries with purchase intent and make one product-tagged video per query. Ranking and revenue are the same motion here. - **Pin the shoppable winner to the profile.** When a product-tagged video ranks and converts, pin it. The pinned slots become a shoppable storefront for your three best search-to-purchase answers. ## Search ranking is not the For You algorithm Most TikTok advice optimizes for the For You page, then assumes search works the same way. It does not. The For You feed is a recommendation engine that guesses what a viewer wants from their watch history. Search is a retrieval engine: a human typed a string and TikTok returns the videos that best answer it. Same app, opposite jobs. - **Watch-time is scored against intent, not the average.** A video can underperform on the feed and still rank in search if the search-referred viewers (the ones who typed the query) finish it. Segment retention by traffic source, because search-referred completion is the number that moves search rank. - **Literal keyword match carries weight the feed does not give it.** The feed matches a video to a person and ignores the exact phrase. Search matches a video to a phrase, so the exact string in the caption, the on-screen text, the auto-caption of your spoken audio, and the comments all become retrieval signals. Feed your text to search and your behavior to the feed. - **The auto-caption of your spoken audio is a search index, not a subtitle.** When you speak the query aloud, TikTok auto-captions it into the same index that catches your written caption. Search leans on it hard; the feed barely uses it. That is why saying the query out loud in the first three seconds is a search tactic specifically. - **Comment keywords feed retrieval.** When commenters repeat the query (or you seed it in a pinned comment), that text counts toward what the video is about for search. The feed does not promote a video because its comments contain a keyword; search can surface it for exactly that reason. - **Freshness and decay differ.** The feed front-loads new uploads then drops them. Search keeps returning a video for as long as it satisfies the query, which is why search-optimized videos have a months-long tail and feed-optimized videos have a 48-hour spike. Tactical note: on every video, compare search-referred retention to overall retention. Higher search retention means a ranking asset; lower means the video got classified for a query it does not answer, and you are about to lose the placement. ## The searchable-hook framework A hook has to stop the scroll (a feed job) and declare the query the video answers (a search job). The searchable hook does both: it is the literal phrase a human types, said out loud and printed on screen in the first three seconds. Title the video as the query, not as the vibe. - **"How to X."** The instructional query. Hook: "Here is how to X," frame-one text "HOW TO X." Not "I finally figured out the secret to X," which buries the search string. - **"Best Y for Z."** The recommendation query, highest commercial intent. Hook: "These are the best Y for Z," overlay "BEST Y FOR Z." The "for Z" qualifier is where the long-tail and the conversion live, so never drop it to sound cleaner. - **"X vs Y."** The comparison query, pure bottom-of-funnel. Hook: "X versus Y, here is the honest answer." The easiest to rank, because most creators make brand-loyalty content instead of neutral comparisons and the supply is thin against real demand. One query per video. Name it in the hook, let the back half pick up adjacent queries through the transcript. Name the format too ("ranked," "tier list," "honest review," "step by step"); format words are searched and they set the completion-rate expectation. If you would not type your hook into the search bar, it is a feed hook. Rewrite it until it reads like a query. ## The Google TikTok carousel and short-video pack One TikTok can rank in two search engines, and the second is Google. Google surfaces TikTok in the short-video pack, the video and image tabs, and inside AI Overviews as a cited source. - **Google reads the caption as the title and the on-screen text as content.** It pulls the caption, the visible overlay text, and increasingly the spoken transcript. A video packaged with the literal query in all three layers is also the one Google can slot into a short-video pack. Optimize once for both indexes. - **The query types that trigger the pack are predictable.** How-to, product and "best" queries, recipe and tutorial, and "watch" intent. A well-packaged TikTok has a real shot at the carousel even with modest follower count, because Google ranks the answer, not the account. - **Recency is a Google ranking factor for video, harder than for text.** Google's video packs skew fresh, so a recent TikTok outranks a better old one for the same query. Re-shooting a proven query refreshes both your TikTok rank and your Google video rank. - **Cross-post to YouTube Shorts to widen the surface.** Google owns YouTube and cites it most in AI Overviews. The same vertical video as a Short with a keyword title and description becomes a second Google-eligible asset. One shoot, three surfaces. See [why video is the most undervalued GEO lever](/playbooks/video-is-the-most-undervalued-geo-lever/). - **You do not control whether Google picks TikTok or YouTube, so ship both.** Google chooses the source per query. Have your answer present on more than one platform and let it choose. ## Turn one TikTok into a Google-indexable page A TikTok ranks inside TikTok and maybe inside a Google video pack. It cannot rank as a blue link or get cited cleanly by an AI engine that prefers quotable text. The fix is to turn the script you already wrote into a plain web page on a domain you own. This playbook exists as an HTML page and a clean Markdown twin at the same path for exactly that reason. - **The script is already the article.** You scripted around a literal query with the answer front-loaded, which is the structure of a page that ranks. Paste the transcript, tighten it into prose, publish. - **Embed the TikTok in the page and the page in the caption.** The page carries the text Google indexes and the engagement of the video; route the link-in-bio to that page. The two surfaces point at each other. - **Serve a clean Markdown twin for the AI crawlers.** AI engines prefer structured, low-noise text. A Markdown version (same content, no nav chrome, no video player) is the format they quote most reliably. - **Add FAQ schema with the exact queries.** The questions people typed to find the video become visible FAQ entries and matching FAQPage structured data, which earns AI Overview and answer-box placement for the same query. - **Build the page once per money query, not per video.** Build the indexable page for queries with commercial intent and durable demand, point three or four supporting videos at it, and let the page be the canonical text answer while the videos own in-app search. Same discipline as our [how to get cited by ChatGPT in 2026](/playbooks/how-to-get-cited-by-chatgpt-in-2026/) playbook. ## The 30-day TikTok SEO sprint A concrete four-week sprint to take an account from no search presence to ranking on a defined query cluster. Run it once to stand up a category, then drop into the weekly mining loop to maintain it. **Week 1: research and infrastructure.** 1. Run the 90-minute keyword research per category and output one keyword sheet. 2. Pick one head term and the eight to twelve long-tails clustered around it. One cluster, not five. 3. Fix the profile: display name carries the category keyword, bio in query language, link-in-bio routed to the matching page, three pins assigned to the cluster's top terms. 4. Create the playlist named with the head term before any video lives in it. **Week 2: ship the 10-video test.** 1. Produce ten videos against the one cluster: same keyword packaging, different hooks. Script each before the shoot with the query at open, midpoint, and close. 2. Mix formats to keep cost low: talking-head, photo-mode carousels, one or two reply-with-video answers. 3. Package every one: query in the spoken hook, query as frame-one text, query first in the caption, three to five specific hashtags plus one broad. Add each to the playlist. 4. Cross-post each to YouTube Shorts with a keyword title and description. **Week 3: measure, double down, stitch.** 1. Check rank logged-out for every cluster term and log positions. Read the Search traffic source and search query data per video. 2. Identify the two or three videos pulling search traffic. These are your rankers. 3. Make four adjacent-long-tail videos off each ranker. Rankings cluster. 4. Stitch the videos already ranking for your head term and harvest new comment questions into reply-with-video briefs. **Week 4: build the page, prune, and set the floor.** 1. Turn the best script into a Google-indexable page with FAQ schema, embed the winning video, point supporting videos and the link-in-bio at it. 2. Run kill-and-replace on rank-but-leak videos: diagnose the drop, re-shoot with the same packaging and a tighter cut. 3. Lock the floor: three search-targeted posts per week (carousels count) and the 30-minute Monday mining loop. 4. Write the cluster up: which terms ranked, which videos pull search traffic, what search share is doing. That report decides the next cluster. Tactical note: the sprint is one cluster on purpose. Ten focused videos on one query cluster out-rank fifty scattered across ten topics, because the search classifier rewards topical concentration. ## Common TikTok SEO mistakes - **Optimizing the caption and winging the script.** The transcript is the whole document and a search index. Script the query into the audio first, caption second. - **Writing hooks for the feed and expecting search rankings.** If the hook is not a phrase a human would type, it is built for the For You page, not the search bar. - **Stuffing twenty hashtags, including #fyp.** Hashtags are a relevance signal, not a reach lever. Three to five specific tags plus one broad, matched to the spoken and on-screen query. - **Substituting slang and brand-speak for the searched term.** "These kicks," "our hero product." Nobody types those. Speak the term the way it gets typed. - **Posting one video per topic and moving on.** Search rewards topical concentration. Ten videos on one cluster builds an account the classifier trusts. - **Judging videos by view count instead of search traffic.** A decaying feed spike is worth less than a smaller video pulling search traffic for months. - **Re-posting the identical video to refresh it.** TikTok suppresses duplicates. Re-shoot with a new hook and the same packaging. - **Ignoring the profile.** Display name, bio, pins, and playlist names are indexed keyword slots. The videos carry the personality; the profile carries the keywords. - **Treating TikTok as a closed loop.** A TikTok that never becomes a YouTube Short or an indexable page is one search surface when it could be three. ## What Winston does - TikTok keyword research mapped to buyer-intent prompts. - Caption, hashtag, on-screen text optimization per priority video. - Voiceover scripting for keyword density without sounding forced. - Cross-platform distribution: YouTube long-form, TikTok short-form, Reels. - Monthly branded-search correlation tracking. ## FAQ **How do you rank in TikTok search?** Say your exact target query verbatim in the first two seconds of the voiceover, put it in the on-screen text overlay, and lead the caption with it. TikTok indexes the transcribed audio, overlay text, and caption as three separate layers and rewards literal phrase matches. Mine keywords from search-bar autosuggest and Creator Search Insights, ship ten videos against one cluster, then double down on whatever ranks. **What is TikTok SEO?** Optimizing video content to appear in TikTok's in-app search results through keyword-rich captions, hashtags, on-screen text, voiceovers, and metadata. **Why does TikTok SEO matter for brands?** 74% of Gen Z uses TikTok as their go-to search tool. 78% have purchased a product after seeing it in a video. 70% of beauty discovery happens on social platforms. **How does TikTok search relate to GEO?** Two ways. Top-of-funnel brand discovery (58% correlation between TikTok views and Google branded search). And AI systems ingest video content (YouTube is the #1 citation source for AI Overviews). **How is TikTok search ranking different from the For You algorithm?** The For You feed is a recommendation engine that matches a video to a person using watch behavior. Search is a retrieval engine that matches a video to a typed query using text. Search leans on literal keyword matches in the caption, on-screen text, the auto-caption of your spoken audio, and the comments, and on search-referred retention, while the feed mostly reads behavior and front-loads new uploads. **Do TikToks rank in Google search results?** Yes. Google indexes TikTok and surfaces it in the short-video pack, the video and image tabs, and as a cited source in AI Overviews. The same packaging that ranks a video in TikTok search earns the Google carousel slot, and cross-posting to YouTube Shorts widens the surface because Google cites YouTube most in AI Overviews. **How do I say a keyword so TikTok indexes it?** Say the exact target query out loud in the first three seconds so the auto-caption transcribes it into the search index, using the literal phrase a human types rather than slang. Put the same query as frame-one on-screen text and lead the caption with it, so the spoken auto-caption, the overlay, and the written caption all classify the video.