Which attribution model should I use?

Data-driven where volume supports it, it distributes credit by observed contribution instead of position rules. The habit worth keeping: run one period through two models, and the channels that swing most mark your real uncertainty.

Why does attribution never match my sales numbers?

Models only see tracked touches: word of mouth, LINE chats, offline visits and cross-device gaps are invisible, and consent-declined traffic drops out. Attribution allocates the observable; the order database stays the total. Use each for what it is.

Analytics

Attribution Model

An attribution model is the rule that decides how credit for a conversion is assigned across the touchpoints a user interacted with before converting.

Models range from last-click, which credits the final touch, to data-driven, which distributes credit across the journey. For example, last-click may undervalue early awareness channels that started the path. The model you choose shapes which channels appear most effective.

Every model is an opinion about credit, and the choice quietly sets budget strategy: last-click systematically starves awareness channels, first-click starves closing ones, data-driven redistributes by observed contribution but stays a model of the touches it can see. Untracked influence, word of mouth, LINE conversations, offline, is invisible to all of them. The practical posture is comparative: read the same period through two models, note which channels swing most, and treat those swings as the measure of your uncertainty. Incrementality tests, geo holdouts, remain the only ground truth attribution can be checked against.

Example

One sale, three touches: a Facebook ad introduced the brand, an organic search compared options, a brand-name ad closed. Last-click hands all credit to the brand ad; the report says prospecting "doesn't work" and its budget gets cut, after which sales fall. Data-driven attribution splits credit across the path, and the same journey now shows prospecting as the step everything else depended on.

Frequently asked questions

Which attribution model should I use?: Data-driven where volume supports it, it distributes credit by observed contribution instead of position rules. The habit worth keeping: run one period through two models, and the channels that swing most mark your real uncertainty.
Why does attribution never match my sales numbers?: Models only see tracked touches: word of mouth, LINE chats, offline visits and cross-device gaps are invisible, and consent-declined traffic drops out. Attribution allocates the observable; the order database stays the total. Use each for what it is.

Working on Attribution Model? See how our SEO audit can help.

Explore SEO audit

Back to glossary

Attribution Model

Frequently asked questions

Related terms