Both, and the two are tangled together. Some view-through conversions reflect genuine influence, especially from video and display that build awareness before a later purchase. But as ad platforms report them, view-through numbers are inflated: the platform credits itself for every user who saw an impression and later bought, including the many who would have converted anyway. A report cannot separate influence from coincidence. Only incrementality testing can.
What a view-through conversion actually is
A view-through conversion (VTC) is recorded when a user is served an ad impression, does not click, and then converts within a defined attribution window. Meta counts a 1-day view by default in sales campaigns; display and video networks often stretch the window to 14 or 30 days. The user never touched the ad. The platform simply knows the impression was delivered to someone who later purchased, and it claims that purchase.
Note the exact wording: served, not necessarily seen. An impression at the bottom of a feed scrolled past in half a second still qualifies. Viewability standards help, but a "viewable" impression can still be two seconds of partial, sound-off visibility that no human consciously registered. That gap between delivered and noticed is where most of the inflation lives.
Why platforms report view-through, and why the number inflates
There is an honest argument for VTCs. Most people never click ads, and upper-funnel formats genuinely shape demand. Click-only measurement punishes video and display for doing exactly the job you hired them to do: create awareness that pays off later through a branded search, a direct visit, or an email open. That part is correct, and dismissing view-through entirely is its own kind of error.
The inflation problem is selection bias. Platforms show ads to the people most likely to convert, and retargeting literally targets your near-buyers and recent visitors. When those people are served an impression and buy, the platform claims credit for a purchase that was already in motion. Summed across every channel, platform-reported conversions routinely exceed what your backend actually booked, and view-through credit is a large share of that double counting. Each platform grades its own homework, and view-through is the most generous grade on the sheet.
How to test whether views drive real value
The only way to know whether a view created a sale or merely coincided with one is a causal test. Four methods, from most rigorous to most practical:
1. Platform lift studies
Meta and Google offer conversion lift studies built on true holdout groups. A slice of your target audience is withheld from ads, and the difference in conversion rate between exposed and holdout users is the incremental effect, views included. This is the cleanest read on whether impressions moved anyone.
2. Geo holdouts
Turn a channel off, or vary its spend, in matched regions and compare backend sales. Geo tests work even where user-level tracking has collapsed, because they measure outcomes at the market level rather than the individual level.
3. Marketing mix modeling
MMM estimates each channel's contribution from spend and outcome data, with no dependence on impression-level attribution at all. It is the natural home for awareness media whose value never shows up cleanly in a click path.
4. Ratio sanity checks
If a retargeting campaign shows spectacular view-through ROAS against an audience of recent site visitors, assume most of it is claimed, not created, until a holdout proves otherwise. A number that looks too good on a warm audience usually is.
When to count views, and when to discount them
Views deserve weight when the format is genuinely attention-getting (sound-on video with high completion rates, high-impact placements), the audience is cold prospecting rather than retargeting, and lift tests have shown the channel drives incremental conversions. In those conditions a view is a plausible cause, not a coincidence.
Discount views heavily when the audience is retargeting or existing customers, the window is long (a 30-day view window on display credits almost anything), the inventory is cheap and low-viewability, or the campaign's click-based performance is weak and view-through is quietly doing all the reported work. The table below turns those instincts into signals you can read at a glance.
| Signal | Lean toward real value | Lean toward inflation |
|---|---|---|
| Audience | Cold prospecting | Retargeting, CRM lists |
| Format | Completed video, high-impact | Banner at the edge of viewability |
| Window | 1 day | 14 to 30 days |
| Validation | Lift test shows incrementality | No test, report only |
What this means for your budget
Do not zero out view-through, and do not take it at face value. A practical middle path has three moves. First, report click-based and view-inclusive numbers side by side, so no one confuses the two. Second, apply skepticism proportional to audience warmth, discounting retargeting harder than prospecting. Third, let periodic lift tests or MMM set the actual discount factor for each channel, rather than arguing about it in a meeting.
Remember that the tools you use carry their own bias. Click-path platforms like Triple Whale or Northbeam will undervalue awareness media by design, because a view with no click barely registers. Platform reports overvalue it by design, because self-attribution is their business model. The truth sits between them, and only a causal method can locate it.
Where a unified approach fits
This is exactly the gap a unified-measurement platform is meant to close. Admira combines multi-touch attribution with marketing mix modeling and lift testing on one first-party foundation, so view-heavy channels get judged by incrementality instead of self-reported credit, and your ROAS stops being a negotiation between dashboards that each want to win. If your prospecting and retargeting numbers lean hard on view-through credit, book a demo and we will pressure-test those views against real lift before you scale behind them.
FAQ
Are view-through conversions in GA4?
Not for non-Google media. GA4 only registers sessions, so an impression on Meta or TikTok with no click never appears in the data at all. This is one of the biggest reasons platform and GA4 numbers disagree so sharply on the same campaign, and why you should never expect the two to reconcile.
What is a reasonable view-through window?
Shorter is more defensible. A 1-day view window at least ties the impression closely to the purchase in time, so the causal story is plausible. Long windows on display inventory credit ads for purchases that have only the weakest connection to the impression, inflating ROAS without adding any real evidence of influence.
Do view-through conversions matter for the algorithm even if reporting is inflated?
Yes. View events give delivery systems more training signal, which can genuinely improve who sees your ads and how quickly the system finds buyers. You can benefit from views as optimization input while still discounting them as proof of value. The two roles are separate: signal for the machine, evidence for you.
How often should I run lift tests?
Test a channel whenever the spend on it is large enough that a wrong ROAS reading would change your budget. For most teams that means a few times per year per major channel, and always before scaling a channel whose entire case rests on view-through credit. A holdout costs a little short-term reach and buys you a decision you can defend.
.svg)
%20(2).png)
