Why Last-Click Makes Display Look Worthless
Last-click attribution credits only the final interaction before a purchase. Display and video ads are mostly built to be seen early in a journey, not clicked at the moment of purchase, which means they lose almost every credit contest to whichever cheap, high-frequency format happens to sit closest to the sale — even when that format's clicks were never the reason the shopper decided to buy.
What this looks like in a real account
The structural mismatch, plainly stated
Display and video formats are designed for exposure — a banner or a video ad reaches a shopper who wasn't actively searching, plants awareness, and moves on. The shopper's actual purchase, when it happens, is very often triggered by a completely separate later moment: a search, a branded query, a retargeting click days later. Last-click attribution was never built to credit the first thing; it credits the last thing, structurally and by design. Applying it to display isn't measuring display badly — it's asking a tool that only ever answers "what closed the sale" to also answer "what started it," which it can't do.
The worked example that shows the size of the gap
Take the sharpest illustration from our own DSP data: third-party mobile in-app video generated 48.3% of every click in our book in a recent 31-day period across 27 advertisers, and 3.5% of the sales. That format ran at a 1.56% click-through rate against a 0.28% portfolio average — on some exchanges above 20%, a rate that's a thumb reacting to a full-screen interstitial, not a genuine intent signal. Those clicks cost $0.33 each, which is why the segment looks attractive on a pure cost-per-click report, and they carried a $26.94 cost per acquisition against $6.83 portfolio-wide. A last-click-only view of that inventory would show it winning almost half the account's click volume — and, judged purely on that click share, would make it look like the most "engaging" format in the account, when the actual sales evidence says the opposite.
Why cheap, high-frequency clicks beat genuinely persuasive display under last-click
Last-click attribution doesn't ask whether a click reflected real purchase intent — it just checks whether a click happened closest to the sale. A format that generates enormous click volume through accidental taps and interstitial mishits will win the last-click credit contest against a genuinely persuasive display or streaming impression almost every time, purely because clicks, of any quality, are the currency last-click counts. The same 48.3%-of-clicks, 3.5%-of-sales pattern is the clearest evidence available that click volume and actual influence are not the same thing, and a measurement system built entirely around clicks will systematically reward the format that generates the most of them, not the one that does the most persuading.
The corrected comparison: display versus video in the same inventory
Isolating format from placement makes the mechanism even clearer. In the same mobile-app inventory class — same apps, same devices, same audiences — display units returned 6.29x while video units in that identical inventory returned 1.46x. Mobile-app display cost $5.52 per attributed purchase; mobile-app video cost $26.94. The difference isn't the audience or the placement — it's the format and the click behaviour it produces. That comparison is also a reason not to conclude that programmatic video doesn't work on Amazon generally: Amazon's own video surfaces behave very differently from third-party in-app video, and lumping the two together under one "video" line item hides exactly the distinction that matters.
The common mistake, including ours
The mistake is optimising toward cost-per-click as a proxy for quality, without checking what's actually driving a format's low CPC. We've seen — and had to correct — a bid strategy that shifted budget toward in-app video specifically because its CPC looked so favourable on a last-click report, before the $26.94 cost-per-acquisition number surfaced and made clear the cheap clicks weren't converting at anywhere near the rate the CPC alone implied. Cheap clicks are, in our own words now, the single most expensive thing to optimise toward on DSP — a lesson that cost real budget before it became a standing rule in how we review supply sources.
Why this compounds with the assisted-conversion problem
The same mechanism that makes last-click undercredit display also inflates how good the closing-touch format looks, because the sale it's crediting itself might have been substantially decided earlier, by exposure last-click never sees. That's the other side of the coin covered on the assisted-conversions page on this site — a channel showing a strong last-click ROAS isn't necessarily doing more real work than a weaker-looking upper-funnel one; it may just be sitting closer to the purchase moment by design, collecting credit that a fuller, cross-channel view would distribute differently.
What to measure instead of last-click ROAS for display
Break performance out by supply source and format before trusting any blended last-click number — the account-level average hides exactly the kind of inversion the in-app video example shows. Weight click-through rate against cost-per-acquisition, not against itself, since a high CTR paired with a high CPA is the specific signature of accidental or low-intent clicks. And where a format's real contribution is genuinely upper-funnel, measure it the way the upper-funnel measurement approach on this site describes — completion rate, viewability, Brand Lift, geo-tested search lift — rather than expecting a last-click report to ever tell that format's real story.
| Format (same inventory class) | Share of clicks / CTR | ROAS / CPA |
|---|---|---|
| Third-party mobile in-app video | 48.3% of all clicks, 1.56% CTR | 3.5% of sales, $26.94 CPA |
| Mobile-app display, same inventory | Lower click volume, lower CTR | 6.29x ROAS, $5.52 CPA |
| Portfolio blended average | 0.28% CTR | $6.83 CPA portfolio-wide |
Which one you should actually pick
Any advertiser with DSP reporting access can break out performance by supply source and format to check for this pattern themselves — it doesn't require special tooling, just looking one level below the blended account number. reMKTR reviews supply-source-level CPA against CTR on every managed account specifically because this exact mistake has cost us real budget before, as part of the discipline behind Full Circle's $500M+ in managed Amazon spend across 100+ brands.
Shortlist on the job, not the feature grid. Pull your search-term report for the last 90 days and total the spend against terms that produced no orders — 33.6% on the account above. Then ask each vendor on your list what they would do about it in week one, and see who answers with a process rather than a screenshot.
Common questions
Does this mean display and video advertising doesn't work on Amazon?
No — the same data shows mobile-app display returning 6.29x in the identical inventory class where video returned 1.46x. The issue is specific to formats with abnormally high accidental-click rates, not display or video as categories.
Why does in-app video get so many clicks if they're not real intent?
Full-screen interstitial video units are prone to accidental taps and mishit clicks, producing click-through rates well above the portfolio average — in our data, over 20% on some exchanges — without a matching increase in genuine purchase intent.
Should I stop bidding on cost-per-click for DSP display and video?
Weigh CPC against cost-per-acquisition for the same placement before trusting it as a quality signal — a low CPC paired with a high CPA is the specific pattern that indicates low-intent, high-volume clicking rather than genuine engagement.
How do I know if a format in my account has this same problem?
Break out click-through rate and cost-per-acquisition by supply source. A format with a CTR well above your portfolio average paired with a CPA well above it too is worth investigating for the same accidental-click pattern.
We show the method before the number.
Claim the free auditRead next
- Criteo Pricing: The Fee Stack Inside Your BudgetPricing · criteo pricing
- Criteo Alternatives: Sorted by Why You Are LeavingAlternative · criteo alternative
- Pacvue Pricing: No Public Number — What to AskPricing · pacvue pricing
- Skai vs Pacvue: Contracts, Not Feature GridsHead to head · skai vs pacvue