Same campaign, three attribution models: $43.90 to $1.02

One campaign spent $1,756 and produced 40 first deposits, a cost of $43.90 per FTD on last-click. Counted with post-click credit, the same spend cost $10.98 per conversion; with post-view credit, $1.02. The figures come from the ad network Kadam. Nothing about the campaign changed between the three numbers. Only the attribution models did.

Results at a glance

The table holds the campaign's numbers as Kadam's CEO presented them in a 2026 conference talk. All of them are reported by Kadam and not independently verified.

Context

The case is a measurement argument made by a seller of top-of-funnel traffic. Kadam's CEO, Sergiy Sergienko, framed it against a market where, in his view, acquisition costs keep rising and payback keeps getting longer, so teams cut whatever does not show up in their reports.

Attribution models are the rules that decide which ad touches get credit for a conversion. Of the common marketing attribution models, last-click is the bluntest. In last-click reporting, whichever ad is clicked last takes all the credit, so a push notification can absorb the value of video and banner views that came days earlier. Multi-touch attribution models spread credit across several touches, which is why they tend to rate display and video higher. Our guide to attribution models covers the types and how Google and Meta apply them.

Kadam has a stake in the conclusion. Its own site lists popunder, push, in-page push, banner, native and video formats, so a model that gives display more credit makes more of its inventory look profitable.

What they did: comparing three attribution models

Kadam took one block of spend and counted its results under different rules, without changing the traffic. Each rule asks a different question about the same users:

• Last-click: which conversions followed a click that was the final tracked touch? Here, 40 FTDs.

• Post-click: which users clicked an ad and converted at some later point? Here, 160 conversions.

• Post-view: which users saw an ad, never clicked it, and converted later? Here, 1,730 conversions.

Kadam itself framed this as last-click against counting delayed conversions; we split the second view into post-click and post-view because it reported them separately.

A blended CPA, as the speaker used the term, divides spend by conversions credited to both views and clicks rather than to the last touch alone. The talk did not give a blended figure for this campaign, and the post-click and post-view counts should not be added together, because the overlap between them and with the 40 last-click FTDs is unknown.

The arithmetic is simple, and we checked it: $1,756 divided by 40 is $43.90, by 160 is $10.975, and by 1,730 is $1.015.

The talk did not say which formats this campaign used. As a general scheme, the speaker recommended warming cold players up with video, banner and native ads, then using pre-landers, push and popunder to convert, and retargeting to bring lapsed players back. The reasoning is that last-click credits whichever ad was clicked last, so the early touches in such a sequence usually show nothing in a last-click report.

Results

The three counting rules put the same campaign's cost per acquisition about 43 times apart, from $43.90 to $1.02. That spread is the result. None of the three numbers is a better or worse campaign; each answers a different question about it.

One detail shows how loose the post-view count is. The campaign produced 1,490 registrations, which by our arithmetic cost about $1.18 each. The post-view figure of $1.02 per conversion is cheaper than a sign-up. That is hard to square with first deposits alone, so the post-view count probably includes other events or players.

The talk paired the numbers with a test on Kadam's own brand. To try the idea on itself, the network stopped buying display for its own brand and put that money into PPC, SEO and social ads, where attribution is easier to read. By the company's account, searches for "Kadam" then fell 35%. When display came back, the metrics returned to normal within three to four days. Kadam presented this as evidence that display was feeding brand search and organic traffic that last-click credited to other channels.

A second company the speaker did not name saw visits fall 17% and leads fall 54% after pausing display. He said leads then rose 224% once buying resumed, without saying from what base. With no name, dates or baseline, the example cannot be checked.

What did not work

The case makes a fair point about last-click, but the data behind it has gaps that limit what anyone should do with it.

• The conversion is undefined. The talk called a conversion a target action, for example a deposit. But 1,730 post-view events is more than the 1,490 registrations, so at least some of them cannot be first deposits by this campaign's registrants. The talk does not say what they were.

• View-through credit flatters display. A person who saw an ad and later deposited might have deposited anyway. Post-view counting is a model, not proof that the ad caused the deposit. Google Ads, for example, reports view-through conversions in their own column and keeps them out of its main "Conversions" column.

• Nothing is disclosed about the setup. No period, GEO, format mix, view window or click window was given. A one-day and a 30-day view window would produce very different post-view counts.

• The brand test was not controlled. It was one company's observation, with no holdout group and no stated period, and it measured searches, not revenue.

• The source sells the traffic. An ad network benefits when buyers credit impressions.

Nothing in the talk shows that anyone paid $1.02 for a depositing player. The only deposit count tied to this spend is the 40 last-click FTDs.

Takeaways for affiliates

The useful lesson is to separate the number you budget on from the numbers you use to judge a funnel.

• Budget on paid FTDs. Operators usually pay CPA on qualified players tied to an affiliate's tracked click and reported back by postback, not on view-through credit. Some networks do credit views (Awin pays on a post-view cookie only when no click cookie exists), so read the program terms.

• Keep click and view conversions in separate columns. Treat post-view as a signal about reach, the way Google Ads keeps view-through out of its main conversion count.

• Ask what a "conversion" means. Before comparing digital attribution models in a network report, get the event, the windows and whether click and view counts are deduplicated. Counted and paid conversions can differ a lot, as our KPI dispute case illustrates.

• Test a pause before cutting top-of-funnel. Kadam's recovery took days, so a short holdout by GEO or time block can show whether push and search conversions drop without display.

• Reconcile network counts every week. Ask the network for click-only and view-only conversions in separate columns, and check each against operator-reported FTDs weekly. A custom attribution model built this way can guide creative and channel mix without touching how payouts are counted.

FAQ

What are attribution models?

Attribution models are the rules that assign credit for a conversion to the ad touches that came before it. Last-click gives all credit to the final click; other models spread it across several touches. See our attribution models guide for the different types.

What is the difference between post-view and post-click conversions?

A post-view conversion is credited to an ad the person saw but did not click. A post-click conversion follows a click on the ad, but is credited even if the person did not convert straight away. Platforms set a time window for each, so the counts depend on those settings.

Why can one campaign show both a $43.90 and a $1.02 cost per acquisition?

The spend stays the same while the number of credited conversions changes. Last-click credited 40 FTDs; post-view credited 1,730 conversions, a looser event, so the cost per conversion fell to $1.02.

What are the four types of attribution?

Common single-touch and multi-touch rules include first click, last click, linear, time decay and position-based, plus data-driven. As of October 2026, Google Ads offers only last click and data-driven, and Meta's standard attribution lets advertisers choose click windows of 1 or 7 days and a 1-day view window.

Sources

• Google Ads Help, About view-through conversions

• Google Ads Help, About attribution models

• Meta Business Help Center, About attribution models and attribution settings

• Awin Help, Awin cookie types and attribution hierarchy

• Kadam, Advertising network website

• Case figures: Kadam CEO's 2026 conference talk, as relayed in trade press; original slides or recording not located.