Tracking which creator drove which install

you track creator performance with unique codes and links where the platform allows them, spike-matching against posting times where it does not, and one per-creator sheet that adds it all up. view counts do not decide retainers, tracked installs do. the sheet is boring, and it is the entire difference between a roster and a list of people you pay.

Unique codes and links

give every creator their own trackable path into the app, and accept that each path leaks. the workable options:

• a creator code redeemed inside the app, for a bonus, a trial extension, anything cheap. codes survive the screenshot-and-search behavior that breaks link attribution, because the user types the code after installing.
• a unique link per creator in their bio or comment, routed through your site so you count the click before the store swallows the trail.
• the blunt instrument that outperforms both: a “how did you hear about us” screen in onboarding with a free text field. self-reported attribution is unfashionable and it routinely catches creators the links miss, because most viewers watch, remember, and search the app name later.

none of these is precise. app store handoffs strip data, viewers install days after watching, and codes only get typed when the incentive is worth typing. run all three anyway; directionally consistent beats precisely blind.

Spike-matching by posting time

when link data is thin, match install spikes to posting timestamps. keep a timeline of every creator post, then lay daily, ideally hourly, installs over it. a post that moves the needle shows up as a bump in the hours and days after it goes live, and when creators post in staggered windows, which you should schedule deliberately, the bumps are separable. it is not courtroom evidence, one creator posting during another’s tail muddies the read, but across a batch it reliably sorts creators into moved-something and moved-nothing, which is most of the decision you need.

this is also the argument for staggering your batch calendar in the first place: attribution gets designed into the schedule, not reconstructed after. and it gives you a free sanity check in reverse: a week with no posts scheduled shows you your organic baseline, and any creator whose claimed impact vanishes against that quiet week was riding the baseline all along.

Why view counts lie

views measure the platform’s interest in the video, not the audience’s interest in your app. a creator can pull half a million views with a hook that has nothing to do with the product, and produce zero installs because the product was wallpaper. another does forty thousand views with a demo so specific that the comments fill with “what app is this” and installs follow for a week. pay and retain on views and you optimize your roster for entertainers. the mix that took ZuAI from 10K to 2M users at a $0.02 blended CAC was steered by exactly this discipline: every channel measured well enough to cut what was not converting, whatever its surface numbers looked like. how that arithmetic actually works is in the ZuAI CAC arithmetic.

The per-creator sheet

one row per creator post, and the columns decide retainers for you:

creator, platform, post date, deal type, fee paid, views at day 7, multiple over the creator’s own baseline, code redemptions, link clicks, self-reported mentions, estimated installs, cost per estimated install, and a decision column: retest, retain, or cut.

fill it the same day each week, no exceptions, because a sheet three weeks stale is a sheet you will never trust again. the decision column is the point: when the row says a creator’s three posts cost more per install than your paid ads do, the retainer conversation ends before sentiment starts it. and the weekly pull of views, clicks and redemptions into the sheet is exactly the kind of dull repeated work that belongs in the n8n growth automations playbook; automate the collection, never the judgment.

batch-level verdicts about the channel come first, and that logic lives in how many creators before you judge the channel. the sheet is what turns those batch verdicts into individual ones.

this is one piece of the creator seeding playbook. the breakdown call is free. Let’s talk Growth.