View-Through Attribution

View-through attribution (VTA) credits a conversion when a person was exposed to an ad or video and converted later without a captured click. The platform (TikTok, Meta, a mobile measurement partner) applies a lookback window, sometimes 1 day, sometimes longer, and a set of identity guesses. The founder sees “view-through installs” next to “click-through installs” and has to decide whether the first column is a gift or a trap.

It exists because skippable video is how people actually behave. They watch, they do not tap, they search the app that night. Last-click would miss them. Over-eager VTA would also claim them when they were going to install anyway.

Why it matters to an indie iOS or SaaS founder

Creator posts and Spark Ads are video. Video is VTA’s native habitat. If you take TikTok’s view-through column as ROAS truth, every campaign looks like a hero. If you ignore it entirely, you may kill a creator whose audience searches rather than taps in-app browsers.

Your constraint is cash. App Store proceeds do not care which column a DSP smiled at. Build a sandwich:

• Floor: unique codes, click IDs, ASA, anything that needed an action.
• Register: App Store Connect / Play statistics for the same window (downloads, proceeds).
• Ceiling: platform VTA. If the ceiling is 8x the floor and the register did not move, the ceiling is a story.

Organic creator posts often have no VTA object at all unless you Sparked. Do not invent one. Time-boxed lift on the register is your organic VTA analogue, and it is still an estimate.

Never let a salesperson (or a dashboard tooltip) turn VTA into “we know this creator drove 12,410 installs.” You do not know that. You know a model.

How to use VTA without getting captured

Set lookback windows as short as you can still believe (a 1-day view-through is less mythical than a 7-day on a high-traffic brand campaign). Do not overlap three paid channels and then credit each with VTA on the same install. Pause other tests when you need a clean read. Compare VTA-heavy campaigns against a week of register data, not against each other in the ads UI.

For creator negotiations, keep VTA out of the bonus structure unless both parties can see the same report and agree it is directional.

How JumpTrend approaches it

JumpTrend does not provide an MMP or a VTA model. It tries to make the exposure more relevant, the viewer was already the kind of person who watches this creator, because the creator came from the avatar feed. Relevance can make view-through more plausible and still not make it countable.

Measurement remains yours. JumpTrend will not guarantee downloads to fill a VTA spreadsheet. See measuring creator-driven installs (glossary) and the longer pillar guide.

A sentence you can put in a board update (or a note to yourself)

“Click-based and code-based installs were X. App Store downloads versus baseline were Y. The ads platform reported Z view-through. We are making spend decisions on Y, paying talent on deliverables, and treating Z as an upper bound.”

If you cannot say that sentence, you are either not measuring or you are presenting a DSP’s model as cash. The longer pillar on measuring creator-driven installs is the worksheet that produces X, Y, and Z on purpose.

Multi-channel double counting

If ASA, Meta, and TikTok all claim view-through or last-click on the same week of branded search, the sum will exceed App Store downloads. That is not a mystery. Deduplicate by believing the store first, then assigning incrementality only when you isolated a channel. Creator tests love isolation. Dashboards love overlap. Be on the side of isolation.