You get asked "what is happening on TikTok in this category?" and you have three days to answer it properly. Your listening suite counts posts and mentions. Nothing in your stack counts searches. TikTok market research changes shape once you can pull TikTok's own search-demand numbers into a chat, code a comment sample in one pass, and save every figure as a dated file you can footnote.

What you'd ask on day one

Three questions, run on 19 September 2026 against the energy-drinks category. Each result below is the actual output and links to its snapshot.

Category demand, by country

Try this prompt in Claude, ChatGPT or Cursor with TokConnect connected

How much search demand is there for energy drinks on TikTok right now? Give me the top topics with their 7-day search volume, how many videos exist for each, which countries TikTok attributes the searches to, and the last 30 days for the biggest one.

Topic7-day searchesVideosCountries TikTok lists
soft drink and energy drink images4,800,239587Brazil
energy drinks for staying awake4,750,823832US, Sri Lanka, UK
soft drink and energy drink stories4,728,580116None listed
energy drinks in hospital settings4,603,3180None listed

Four topics in one category, each worth 4.6 to 4.8 million searches a week, one of them with no videos at all. The 30-day window for "energy drinks for staying awake" came back with five reported days: 1,039,899 on 10 September, then 9,794, 108,644, 1,635,680 and 4,750,822 on the 18th. The gaps are days TikTok did not report, not days with no searches.

Data: TikTok Creator Search Insights via TokConnect, pulled 19 September 2026. Open the snapshot. Six tool calls.

Who is doing the searching

Try this prompt in Claude, ChatGPT or Cursor with TokConnect connected

For the topic "energy drinks for staying awake", give me the age, gender and country split over the last 30 days, and tell me what share of the audience TikTok could not classify.

  • Gender: 66.7% female, 0% male, 33.3% unclassified
  • Age: 33.4% aged 18 to 24, 33.4% aged 35 and over, 16.6% aged 25 to 34, 16.6% other
  • Country: 100% United States

The third TikTok could not classify is unknown, not male. Say so in the footnote and the number survives review. One call, one credit.

Data: TikTok Creator Search Insights via TokConnect, pulled 19 September 2026. Open the snapshot.

A coded comment sample

Try this prompt in Claude, ChatGPT or Cursor with TokConnect connected

Find the two most-discussed videos for "best energy drink", pull about 100 comments, and code them: which brands get named, how often, and what are people actually arguing about?

Three calls returned 140 comments across two videos. Seventeen brands came up by name:

CodeComments
Names Monster28
Names Red Bull24
Names Alani, Dzik or Ghost17
Raises a sugar, caffeine or sweetener concern12
Asks for a zero-sugar or "ultra" variant9
Names water, coffee, tea or milk instead8
Challenges the creator's credibility or sponsorship6

The sweetener argument runs both ways inside the same thread, and six people went after the ranking rather than the drinks. A mention count will not give you that.

Data: TikTok search and comments via TokConnect, pulled 19 September 2026. Open the snapshot. Commenter identifiers removed.

What it replaces

What you do todayToolWhat it showsWhat TokConnect adds
Track brand and category mentionsBrandwatchMentions, sentiment, hashtags. TikTok accounts get the top 1,000 mentions, refreshed dailySearches instead of mentions: a 7-day count per topic and a daily series going back 180 days
Build the quarterly category deckTalkwalkerSentiment, virality maps, AI summaries. Aggregates export, raw TikTok rows do notEvery call comes back as JSON you keep, at topic and comment level
Quote platform behaviour in a reportStatista, YouGov ProfilesSurvey and panel data on claimed usage and attitudesTikTok's own count of what people typed, for your exact category
Check whether a term is risingGoogle TrendsGoogle and YouTube search interestTikTok search interest, which Google Trends does not carry
Look a topic up in TikTok's own toolCreator Search InsightsSearch popularity, content gaps and audience split, in the app and on desktop with a loginThe same numbers outside the app, plus the history view TikTok does not offer
Read comments under category videosScrollingWhatever you had time to read, unrecordedUp to 50 comments a call, coded in one pass, saved

None of the listening suites above carries TikTok search-volume data. They count what was posted. This counts what was typed into the search bar, which is the closest thing TikTok has to stated intent.

How it fits your week

  1. Connect once. One config block in Claude, ChatGPT, Cursor or Gemini. The connect page has each one; most analysts run it in Claude Desktop, where the deck gets drafted anyway.
  2. Ask in chat. Type the category question the way you would put it to a junior analyst. The agent picks the tools and gives you the table.
  3. Save the answer. Ask for the raw JSON and keep it beside the deck. When someone queries a figure in November, the file still says what it said on the day.

The numbers come from two places, live. Search volume, the 7-day and 180-day series, the popularity score, the content-gap flag and the audience split are TikTok's Creator Search Insights data. Videos, creators, hashtags, sounds and comments are ordinary TikTok search.

What it does not do: no posting, no scheduling, no ads, no transcripts, no private or personal data, no affiliation with TikTok. It reads the same surfaces a logged-in person sees.

Straight answers

"Is this a valid sample? I can't put a non-probability number in a client report."

It is not a sample. It is TikTok's own count of searches on TikTok over seven days, global unless a country is picked, which is the same figure the app shows a creator. Treat it as a search-log source, the way you already treat Google Trends, and describe it that way in the footnote. ESOMAR's guideline on social media research is the frame for observed public behaviour, and it holds that what people do in a public place is not private to a researcher.

"What is the methodology? I have to answer that before it goes in the deck."

TikTok does not publish how it computes its popularity score, so do not claim one. State what the number is: searches for a topic cluster over the last seven days, global unless you set a country, pulled on a named date. Do not present it as category incidence or market share. Our search volume guide gives you wording you can paste into a methodology note.

"What will legal ask about TikTok's terms?"

It reads what TikTok shows a logged-in user, and nothing else. No private data, no personal data collection, no posting, no ad account. Commenter identifiers are stripped from anything we publish. If compliance wants the official research route instead, the TikTok Research API is limited to academic and non-profit researchers, which rules out commercial insights teams.

"We already pay for a listening suite."

Keep it. It answers "who is talking about us and how do they feel". This answers "what are people looking for". A category with rising searches and falling mentions is a different brief from one where both are flat.

The plan that fits

One tool call spends one credit, whatever it returns. The three questions above cost eleven credits: six for the category sweep, one for the audience split, one for the video search and three for the comment pages. A full category study with five topics opened, two audience splits and 200 coded comments lands around 15 to 20 credits. Two studies a week plus a standing Monday pull across six categories is roughly 200 to 250 credits a month, so Starter at $49 for 1,000 credits covers it with room for the requests that land on a Friday afternoon. Move to Growth when a second analyst starts running their own pulls. Begin on the free 100 credits and count your own first month. Full pricing and the credit table.

Guides for insights teams

Questions market research and insights teams ask

Does TikTok publish search volume anywhere?

Yes, inside Creator Search Insights. It shows "search popularity" for a topic: searches over the last seven days, global unless you pick a country. It lives in the TikTok app and on desktop at tiktok.com/csi with a login. There is no export, no API and no history view, which is the gap this fills.

Is a TikTok search-volume figure comparable to a Google keyword volume?

No, and do not put them in the same chart. Google's keyword tools give a monthly average. TikTok's figure is a seven-day count for a topic cluster rather than one exact phrase. The search volume guide has wording you can reuse in a methodology note.

Can I use this data in a client report?

Yes, cited the way you would cite Google Trends: name the source, give the date you pulled it, and describe the metric. Every call returns JSON we save with a date, so the figure is still checkable months later. Do not present it as category incidence or market share, because it is a count of searches, not of people.

We already pay for Brandwatch or Talkwalker. Why add this?

They count posts and mentions on TikTok and neither carries search-demand data. This answers a different question: what people typed into TikTok's search bar, how that moved over 180 days, and who was doing it. Categories where searches rise while mentions flatten are exactly the ones worth a brief.