Attribution models

The attribution model decides how a sale's credit is divided between the ads that preceded it. Clakta offers six. Change yours from the dashboard toolbar or any Ads Manager page — it re-reports your existing data instantly.

Last Click

Attributes 100% of the sale to the last paid click before the order.

The default, and the most common starting point. Only paid clicks are eligible; organic visits and unclicked impressions are ignored. If the journey has no paid click inside the window, the order is unattributed.

Use it for: bottom-of-funnel decisions, direct-response campaigns, and any time you want whole-number conversions that are easy to explain.

Blind spot: it gives all the credit to whatever closed the sale, so prospecting that created the demand looks worthless.

First Click

Attributes 100% of the sale to the first paid click in the journey.

The mirror image. Same eligibility rules, opposite bias.

Use it for: judging which campaigns actually introduce new customers, and valuing top-of-funnel prospecting.

Blind spot: retargeting and closing campaigns appear to contribute nothing.

Linear

Divides credit equally among every touchpoint — including organic.

Paid clicks, paid views and organic visits inside the window all count, each receiving 1 / n of the sale.

Use it for: seeing the whole journey, including how much of your revenue arrives through channels you are not paying for.

Blind spot: an organic visit is treated as equal to a paid click, which systematically dilutes paid performance. Do not judge ROAS with this one.

Linear Paid

Divides credit equally, but only among paid channels.

Same equal split, with organic touchpoints excluded. Paid clicks and paid views share the sale.

Use it for: comparing paid channels against each other fairly when a typical customer sees several ads before buying. This is usually the best default for multi-channel advertisers who find Last Click too harsh.

Blind spot: every paid touch counts equally, so a cheap incidental impression weighs the same as the click that did the work.

Clicks & Deterministic Views

Weights clicks above views, and recent touches above older ones.

The most sophisticated model. Each touchpoint gets a weight:

• A click starts at weight 1.0 and decays with a 7-day half-life
• A view starts at weight 0.25 and decays with a 1-day half-life

Weights are normalized so the order still distributes exactly 1.0 total credit. A click yesterday therefore outweighs a click six days ago, and any click outweighs a view.

Use it for: the most realistic picture when journeys are long and involve several channels.

Triple Attribution

Gives every paid channel that touched the order 100% credit.

Deliberately does not deduplicate. If a sale was touched by Meta and Google, both get a full conversion. Totals across channels intentionally exceed your real order count. A channel touched twice still counts once.

Use it for: comparing Clakta against what each ad platform reports in its own dashboard, since this mirrors how they each claim the same sale independently.

Choosing

What the model does not change

• Total revenue and total order count. Those come from your orders, not from attribution.
• Ad spend. That comes from the platforms.
• Platform-reported conversion columns. Those are never modified.

Only Clakta-attributed conversions, attributed revenue and the metrics derived from them — ROAS, CPA, profit — respond to the model.

Related

• How attribution works
• Attribution windows
• Unattributed orders