How last click works
Last click is the simplest attribution rule there is. Whichever click brought the person to the site immediately before they converted takes the entire conversion; every earlier click takes none. There is no sharing, no partial weighting and no judgement about which touch did the persuading.
Because it is a fixed rule rather than a model that has to learn, it works on any volume of data and gives the same answer from the same log every time. That is why it is the fallback when a data-driven model has too few conversions to train on, and why it is still what most people picture when they read the word conversions in a report. Google Ads has moved its own default towards data-driven attribution, but last click remains available as a setting and is still the model behind a great deal of everyday reporting.
One variation is worth knowing about: last non-direct click works the same way but skips over a direct visit and credits the marketing touch before it, on the reasoning that typing your address is not really a channel.
Why last click matters
It is the model a client can check. You can point at a single click, a single date and a single campaign, and nobody has to trust a black box. For a business with a short buying journey — an emergency plumber, a same-day service, a low-priced product — that simplicity is often accurate enough, because there genuinely was only one meaningful touch.
It also sets a floor. If a campaign is profitable on last click alone, it is profitable on any model, because every other model can only add credit to it. That makes it a useful test before scaling something.
Common mistakes with last click
The big one is judging awareness activity by it. Video, social and top-of-funnel content rarely close the sale, so under this rule they look close to worthless, and the first budget cut lands on the work that was creating the demand the closing channels harvest.
The second is misreading direct traffic. When tracking is lost — a stripped link, a blocked cookie, a click from an app that does not pass a referrer — the visit lands in direct, and last click hands the sale to nothing at all. That is a measurement failure being read as a channel.
How to act on it
Use last click where it fits and label it clearly wherever it is used, so nobody compares it against a shared model from another tool and thinks performance moved. Keep it as your auditable baseline rather than your only view.
Before cutting any channel on last-click numbers, look at the same period through an assisted or path report and see how often that channel appears earlier in the journey. If it shows up constantly on the way to conversions it is doing work the rule cannot see, and the honest next step is a proper analytics and tracking review or a holdout test rather than a cut.