App marketing agency research: audit, pitch and brief from TikTok
An app marketing agency runs the same TikTok research for every client. TokConnect does it in one chat: recent posts, saves per 1,000 views, category searches, comments and rival apps, pulled live.
- 10 minclient audit
- 3pitch numbers
- 1UGC brief
What changes when the app isn't yours
An in-house app marketer knows one product and one set of rivals. Our guide for app marketers covers that job. An agency starts cold, often on a new app every week, and has to show up to the first call knowing the account better than the client does.
So the agency version of TikTok research is a repeatable sequence: the same seven pulls for every client, the same three numbers for every pitch, and one content brief that a UGC team can shoot. Below, each step runs on a real app, Calm, with real numbers from 10 October 2026.
| Job | In-house team | Agency |
|---|---|---|
| Starting point | Knows the app | Cold, per client |
| First deliverable | Content plan | Audit, then pitch deck |
| Research volume | One app, deep | Many apps, same steps |
| What it needs from a tool | Depth on one category | Speed, repeatability, one bill across clients |
The 10-minute client audit, run on Calm
Seven tool calls, in this order. Each answers one question a client asks in the kickoff meeting. Calm is a useful sample: a big app with an active account, so every step has something to show.
-
Call 1
Profile
user_profilecalm1,139,821 followers, 1,553 videos, 7,312,710 likes.
-
Call 2
Last 30 posts
creator_posts, 12 August to 7 OctoberMedian 25,900 views. 26 of 30 passed 10,000 and 4 passed 100,000. Median length 28 seconds.
-
No call
Saves per 1,000 views
Computed from the same posts
Median 8.5 saves per 1,000 views. The best, 13.8, came on "Which sound do you find most calming?" (133,100 views).
-
Calls 3 and 4
Category searches
search_topics"calm app" and "sleep sounds"6 of 9 sleep-sound searches we cite have 0 videos. None of the 20 topics for "calm app" was the app by name.
-
Calls 5 and 6
Curve and searchers
search_popularityandaudience_demographics54.7% of searchers for "sleeping sounds for sleep" are 35 or older.
-
Call 7
Comments
video_commentson the best post50 of 111 comments read. Viewers argue about thunder and ask for loops.
What the audit found: sound posts carry the account
We split Calm's 30 posts by caption. The 16 that mention sleep, sound, rain, noise or music had a median of 36,500 views and 9.7 saves per 1,000 views. The other 14, mostly short mood clips, had 15,400 and 6.7. All four posts over 100,000 views were sound posts, and eight of the 30 ran about a minute.
That is the first slide of the audit: more long sound videos, fewer short mood clips, with saves as the proof, because a save means someone plans to play it again at bedtime.
Public posts from Calm's verified TikTok account, pulled 10 October 2026. The split uses caption words, so treat it as a first pass to check by eye. Every post and number.
Audit the TikTok account of [client app]: followers, median views and saves per 1,000 views on its last 30 posts, which content types beat the median, the top searches in its category with video counts, and the themes in the comments on its best post.
Audit a client appSearches the client could own
The account already makes rain and white-noise videos. TikTok's search data says people look for exactly that, in phrasings with nothing filed under them. "fan and rain sounds to sleep" drew 294,070 searches in 7 days with 0 videos.
Two of the gaps carry the client's own word. "Calm noises" (260,478) and "calm audios" (249,125) have no matching videos, and the app's account is named Calm. Saying those exact phrases on screen and in the caption is a cheap test.
Time it, too. Daily readings for "sleeping sounds for sleep" peaked at 2,432,851 on 16 August and read 246,423 on 10 October, so the season is past its summer high. The people searching skew older: 54.7% are 35 or older and 19.9% are 25 to 34, with women at 51.6% and men at 29.0%. Every searcher in the panel was in the United States.
| Search, as typed | 7-day searches | Videos |
|---|---|---|
| fan and rain sounds to sleep | 294,070 | 0 |
| sleeping rain sounds | 267,626 | 0 |
| Calm noises | 260,478 | 0 |
| calm audios | 249,125 | 0 |
| sleeping sounds for sleep | 246,424 | 723 |
| sleep raining sounds | 188,118 | 0 |
| white sounds to sleep | 176,311 | 0 |
| Sleep sounds | 172,117 | 870 |
| water sounds for sleep | 113,634 | 836 |
"sleeping sounds for sleep", median daily reading
What the comments on the best post ask for
We read 50 of the 111 comments on Calm's top post, the rain video with 291,400 views and 2,912 saves. They read like a product feedback form.
The most-liked comment, "I need a big thunderstorm" (20 likes), sits beside one who dislikes thunder when sleeping (13). Make both versions.
A spookier version, visuals with the sounds, and "how do you do a loop". Each is a follow-up video or a pinned reply.
Three people say rain makes them need the bathroom, and two say it sounds like frying. Humour you can reply to on camera.
One viewer says they have used the app almost every day since 2022. "I love the calm app" is the line to pin.
Top 50 comments by TikTok's relevance order, pulled 10 October 2026. Commenters are not named.
Pitch-deck numbers for a prospect
The same pulls turn into a pitch for the next client. Say the prospect is Headspace, a direct rival with more followers than Calm. Three numbers fit on one slide, and each comes from a public pull you can rerun in the meeting.
| App account | Followers | Posts pulled | Median views | Saves per 1,000 views | Photo posts |
|---|---|---|---|---|---|
| Calm | 1,139,821 | 30 | 25,900 | 8.5 | 0 |
| Headspace | 1,761,594 | 30 | 2,236 | 3.1 | 9 |
| Insight Timer | 3,626 | 20 | 545 | 2.5 | 9 |
Public posts from each verified account, pulled 10 October 2026: Calm 12 August to 7 October, Headspace 23 August to 10 October, Insight Timer 16 July to 8 October.
Check before you put it on a slide
Four Headspace posts have views far ahead of their likes: 4,100,000 views with 2,621 likes, 615,100 with 137, 499,600 with 155 and 366,100 with 106. That is under 1 like per 1,000 views, while its other 25 posts got a median of 26.7. Ask the client whether those four were promoted before you benchmark against them. Without them, Headspace's median is 2,194 views.
| Search, as typed | 7-day searches | Videos |
|---|---|---|
| meditation benefits | 319,604 | 851 |
| how to meditate with adhd | 299,710 | 0 |
| calming meditation | 290,970 | 1,136 |
| how to start meditating | 191,619 | 757 |
| morning meditation | 169,850 | 1,168 |
The pitch writes itself from there: a rival's sleep-sound format pulls more than ten times the median views, and a 299,710-search question about ADHD has no video. For the outreach side, our guide to winning TikTok clients with search data covers the cold email and objections.
Build pitch numbers for [prospect app] against [its main rival]: followers, median views and saves per 1,000 views on each account's last 30 posts, posts with views far above their likes, and the top searches in [its category] with no videos.
Build a pitch slideSlideshow and UGC research for app clients
Most app agencies sell volume: UGC creators, faceless accounts, slideshows. The research question is what format already wins in the category. We pulled the top 20 TikTok videos for "sleep app".
Small accounts dominate. 14 of the 20 came from accounts under 1,000 followers, and the best of those reached 915,600 views from an account with 147 followers. Seven of the 20 captions name the same app, SOMNR, across two accounts, each ending on the same line: track your sleep with SOMNR. Repetition, not polish, is the pattern.
The format is short. The median video ran 14 seconds and 11 of 20 were 15 seconds or less. Only one carried an #ad tag. Carousels belong in the plan too: Headspace's most-viewed post of the 30 was a photo post, and 9 of its 30 posts were slideshows. Our TikTok slideshow guide covers how to build them.
Find the top TikTok videos for "[client app category] app": views, length, follower count of each account, which apps they name, and the hook in the first line. Then write a brief for 10 UGC videos for [client app].
Write a UGC brief
Top 20 TikTok results for "sleep app" and public brand posts, pulled 10 October 2026. Individual creators are described by follower count, not named.
Multi-client economics: credits, not seats
TokConnect plans are sold by credits, and each lookup uses one credit, whether it returns one result or twenty. A lookup that fails isn't charged. No plan lists a seat limit, so adding a strategist doesn't add a line to the bill. The Calm audit above took 7 lookups; the whole page took 13, plus 6 for the cover images.
71 audits like Calm's a month. Fits 10 clients with a weekly audit each (280 credits).
285 audits a month. Fits 50 clients on a weekly audit (1,400 credits).
1,142 audits a month. Fits 100 clients on a weekly audit (2,800 credits), with room for pitches.
Plans and credit rules from TokConnect pricing, checked 10 October 2026. Audits a month are credits divided by 7, rounded down; a weekly audit is 4 a month. The 50 free credits cover 7 audits and do not refill.
Where it fits an app agency stack
| Job | Where it lives |
|---|---|
| Installs, attribution, ROAS | The client's attribution tool and ad accounts |
| Store keywords and rankings | An ASO tool; see ASO keyword research with TikTok data |
| Posting and scheduling | Your scheduler |
| Client audits, category searches, rival app accounts, comments, top videos | TokConnect, in its chat or inside ChatGPT, Claude and Cursor |
| Ad spend, CPMs, installs, private account analytics | Not in TokConnect: it reads public TikTok only |
Running agency work beyond apps? TikTok tools for agencies shows the same workflow on a consumer brand.
Questions app marketing agencies ask
What does an app marketing agency do on TikTok?
It researches, plans and ships TikTok content for app clients: an account audit, the category searches worth answering, UGC and slideshow briefs, creator sourcing and a monthly report. Our Calm audit shows the research half: 7 lookups returned its median of 25,900 views on the last 30 posts and six sleep-sound searches with no videos. TokConnect runs those lookups for any client app from one chat.
How do you audit an app's TikTok account?
Pull the last 30 posts and take the median views and the median saves per 1,000 views, then split the posts by content type to see which kind beats the median. Add the top searches in the app's category, who searches them, and the comments on the best post. Calm's sound posts had a median of 36,500 views against 15,400 for its other posts.
Why measure saves per 1,000 views for app content?
Views depend on how far TikTok pushed a post; saves show that someone plans to come back to it, which is closer to an install. Dividing by views lets you compare accounts of any size. Calm's median was 8.5 saves per 1,000 views, Headspace's 3.1 and Insight Timer's 2.5 on 10 October 2026.
How much should an agency budget for TikTok research data?
Budget by lookups, not by people. On TokConnect each lookup uses one credit, a 7-lookup audit like our Calm example costs 7 credits, and Starter at $49 a month for 500 credits covers 71 of them. Growth ($99, 2,000 credits) and Pro ($199, 8,000 credits) cover 285 and 1,142, and no plan lists a seat limit.
How we got these numbers
We pulled public TikTok data through TokConnect on 10 October 2026, 19 calls in all:
user_profilefor Calm, Headspace and Insight Timer, andcreator_postsfor their last 30, 30 and 20 posts;search_topicsfor "calm app", "sleep sounds" and "meditation app" (20 records each);search_popularity(180 days) andaudience_demographics(last 30 days) for "sleeping sounds for sleep";video_comments(50 top comments) on Calm's most-viewed post;search_videosfor "sleep app" (20 videos);video_detailfor the six covers shown.
Search counts are TikTok's global 7-day search popularity from Creator Search Insights. Every phrase quoted is under 1 million searches and treated as a typed search; TikTok's grouped labels and off-topic results are left out. Saves per 1,000 views is saves divided by views, times 1,000, with the median taken per account. App brands are named; individual creators and commenters are not. Open the data.