Which attribution model should you use?

MarqIQ has two touches to work with, not four

Each form submission stores a first-touch source and a last-touch source. Nothing in between is kept.

So a model that promises to weight a middle touch has no middle touch to weight. A journey of six visits arrives at the report as two values: where that person came from the first time, and where they came from the time they filled the form in.

When both values are the same — somebody found you on LinkedIn and came back through LinkedIn — the submission counts once, at full credit, under every model.

The same submission under every model

Take one submission. First touch is Google organic. Last touch is LinkedIn Ads. Here is what each button does with it.

When first touch is the honest answer

First touch tells you what is finding people who did not know you. That is the question a demand-generation budget is set against.

• Use First Touch to argue for money at the top of the funnel — content, events, PR, paid social.
• Use Last Touch to see what closes the loop — branded search, retargeting, an email that brought somebody back.
• Use Linear when the honest answer is that both mattered and you do not want to pick.
• Use Time Decay when your buying cycle is short and the recent touch genuinely did most of the work.

First touch is not the honest answer when a channel is only ever the last step. Branded search almost never finds anybody. It is where people go once something else has told them your name. Under First Touch it looks worthless. Under Last Touch it looks like your best channel. Read both before you judge it.

Sourced and influenced are two different numbers

The Attribution tab answers which channel gets the credit for a submission. The Influence report, in the left-hand menu under Intelligence Studio, answers a different question: which analyst firms and press outlets touched the pipeline at all.

Do not add the two together. An outlet that touched a deal and a channel that created it are different claims, and the combined figure means neither.

Fix one model for reporting, use the rest for diagnosis

Pick one model, write it on the slide, and keep it for four quarters. Keep the model fixed, or a growth figure may only mean somebody changed the model between the two quarters.

The other models are a diagnostic tool. Flipping between First Touch and Last Touch over the same quarter tells you how much of a channel's case rests on the last click. A channel that scores well under Last Touch and poorly under First Touch is closing, not finding.

Reports you build yourself do not use these models

The report builder has no model selector. It gives you the underlying fields instead: First-Touch Source, First-Touch Medium, First-Touch Campaign, Last-Touch Source, Last-Touch Medium and Last-Touch Campaign.

Grouping a report by First-Touch Source gives you a first-touch report with every submission counted once. It will not match the First Touch figure on the Attribution tab exactly. The builder opens on a 90-day window and reads at most 50,000 submissions, while the Attribution tab's Monthly period covers the last 12 months.