Rules-based attribution models split conversion credit by a fixed and pre-defined formula. The quick tour, on one illustrative journey (someone sees a social ad, clicks a search result days later, then buys from an email):
- Last click: everything to the final touch: the email gets 100 percent.
- First click: everything to the opener: the social ad gets 100 percent.
- Linear: an even split: a third each.
- Time decay: more credit the closer to conversion: the email most, the search click less, the social ad least.
These are the models most companies actually start with, so they deserve a serious look rather than a sneer: exactly how each works, what each distorts, and when each is honestly the right pick.
Last click: the default that won’t die (ever)
Last-click attribution gives 100 percent of the credit to the final touchpoint before conversion. Someone searches your brand, clicks the result, buys: brand search gets everything.
It survives as the near-universal default for understandable reasons: it’s simple, it’s definitive, and it sits closest to the money.
The distortion is systematic, though: it flatters closers (brand search, retargeting, cart emails) and starves the channels that created the demand. The user who first met your product on social, searched it a week later and bought? Social gets minimum credit, every time. Scale that across a year of reports and the budget quietly migrates to the bottom of the funnel while the top starves.
Honest uses remain: bottom-funnel audits (what’s actually closing?) and platform hygiene checks. As a company-wide truth, it’s the distortion that’s cost the most marketing budgets the most money.
First click: last click’s mirror image
First click gives everything to the touchpoint that opened the journey, reversing last click’s logic and inheriting the mirror-image distortion: discovery gets over-credited, and everything that nurtured and closed the sale gets nothing.
It also carries a practical fragility the mirror doesn’t: the true first touch often predates your tracking entirely. The podcast mention, the friend’s recommendation, the article read months ago: none of it is in the data, so “first click” really means “first tracked click,” which is a different thing.
Its honest use: comparing demand-creation channels against each other. When the question is “which of our prospecting channels sparks the most journeys?”, first click’s bias points exactly where you’re looking, and that’s fine, as long as nobody mistakes the answer for the whole truth.
Linear: fairness by spreadsheet
Linear attribution divides credit equally across every touchpoint. Three touches, a third each. Its appeal is genuinely diplomatic: no channel’s advocate loses the meeting.
The flaw is what equality means here: a two-second glance at a display ad earns exactly the same credit as the demo call that sealed the deal. Long journeys make it worse, diluting every touch toward homogeneous noise: when everything matters equally, nothing informs a decision.
Its honest use: a neutral second opinion. When stakeholders distrust any weighted model’s assumptions, linear is the model with the fewest opinions to argue about, and comparing it against your primary model shows quickly how much the weighting matters.
Time decay: recency as a theory of influence
Time decay gives more credit the closer a touch sits to the conversion, usually on a half-life curve: a touch a day before conversion weighs heavily, one from three weeks ago barely registers.
The underlying theory is plain: recent influence is likelier influence. And as hand-set theories go, it’s the most defensible one in the rules-based family: for short sales cycles it approximates reality better than any other fixed rule.
Its distortion is the gentle version of last click’s: recency still isn’t causation, so the model tilts toward the funnel’s bottom, just gradually instead of absolutely. The discovery touch that genuinely started everything still gets shortchanged: politely, this time.
Verdict: the sanest hand-set default when data-driven attribution isn’t available, particularly for products bought within days or weeks.
View-through: the rule everyone forgets they set
There’s a fifth rule hiding in your reports, and most teams never chose it: view-through attribution. It credits an ad that was seen but never clicked, as long as the conversion lands within a set window after the impression.
The windows are attribution rules too, and platforms set them by default: a common pattern is a 7-day window for clicks and a 1-day window for views. Change the windows and reported performance changes materially, without a single customer behaving differently.
Three practical rules: know what windows every platform in your mix is using; align them where the platforms allow it, or at least document the differences before comparing numbers; and treat self-reported view-through with healthy skepticism, because the platform counting the views is the platform being paid for them: homework, graded by its author.
Both academic research and industry are aware of the importance of View-through. Most of platforms that generate non-click traffic like TikTok for instance, rely heavily of view-through attribution.
When rules-based is enough, and when it isn’t
Honest verdict time. Rules-based models are enough when the team is small, cycles are short, and the need is directional weekly tuning. Any conscious choice from this page beats the unexamined default, and consistency beats sophistication for learning.
They stop being enough at three recognizable moments: when serious budget moves start riding on the numbers; when journeys sprawl across walled gardens, devices and offline, so the trail itself is fragmentary; and when the funnel’s shape is exactly the thing you’re trying to learn, which a hand-set rule assumes rather than discovers.
Then you graduate, in two directions: data-driven attribution learns credit weights inside tracked journeys, and causal methods measure influence beyond them. The full family map is in marketing attribution models: choosing the right one.
Frequently Asked Questions
The model giving 100 percent of conversion credit to the single final touchpoint before purchase: the long-time default nearly everywhere. It systematically flatters bottom-funnel channels like brand search and retargeting, which is why they dominate last-click reports. Its simplicity is why it survives; its bias is why it misleads.
First touch credits the touchpoint that opened the journey, measuring demand creation; last touch credits the one that closed it, measuring conversion completion. Same journey, opposite verdicts: both are single-touch opinions with mirror-image blind spots. Neither sees the journey between its bookends.
The model splitting conversion credit equally across every touchpoint in the journey: three touches, a third each. Democratic and easy to explain, but a fleeting ad glance earns the same credit as the interaction that sealed the sale. Its honest use is as a neutral second opinion.
For short sales cycles where recent touches plausibly matter most, and anywhere data-driven attribution isn’t available: it’s the sanest hand-set default, tilting gently toward the conversion rather than handing everything to the last touch. For most teams it’s the best rule available before volume justifies algorithms.
Crediting an ad that was seen but never clicked, when the conversion lands within a set window after the impression. The windows are platform-set rules that materially change reported numbers, and self-reported view-through deserves skepticism. Check every platform’s window settings before comparing reports.
