Marketing attribution models, judged against your CRM

A marketing attribution model is the rule that decides how credit for a lead or sale is shared across the ads and channels that touched it. Common models give all the credit to the first touch, all of it to the last touch, or spread it across every touch in between. None of them is accurate if your CRM doesn't record where each deal came from.

The common attribution models

Each model below answers the same question with a different rule. "Touch" means a tracked interaction: an ad click, a visit from an email, a form fill.

First-touch

All credit to the first tracked interaction. Use it when you want to know which channels find new people.

Last-touch

All credit to the final interaction before the conversion. Use it when you want to know what closes, knowing it ignores everything that came before.

Linear

Credit split equally across every touch. Simple and fair-looking, but it treats a passing impression the same as a demo request.

Time-decay

More credit to touches closer to the conversion. Suits short cycles where recent activity matters most.

Position-based (U-shaped)

Most credit to the first touch and the touch that created the lead, commonly 40% each, with the rest shared across the middle.

W-shaped

Equal large shares to the first touch, the lead-creation touch and the opportunity-creation touch, commonly 30% each, with the remainder across the others. It needs your CRM to record when a lead became an opportunity.

Data-driven

Credit assigned by an algorithm comparing paths that converted with paths that didn't. It is the default in GA4 and Google Ads, but it only sees the touches that platform can track.

Single-touch vs multi-touch

First-touch and last-touch are single-touch models: easy to run, and each blind to most of the journey. Every other model is multi-touch. Multi-touch looks more complete, but it depends entirely on how many touches are actually recorded. A multi-touch model built on half the touches is not more accurate than a single-touch one, just more confident.

Data-driven attribution in GA4 and the ad platforms

In 2023 Google retired the first-click, linear, time-decay and position-based models from GA4 and Google Ads, leaving data-driven and last click. Meta, LinkedIn and Microsoft each attribute conversions inside their own platform, using their own windows. So every platform runs its own model on its own data, and each one is generous to itself. That is how the totals end up larger than your sales, as described in Meta and Google both claim the same conversion.

A worked example

Take one B2B deal worth $120,000 with four tracked touches:

• A LinkedIn ad brings the first visit.
• A Google search ad brings them back, and they download a guide. This creates the lead.
• A Meta retargeting ad brings a return visit.
• A Google brand search leads to a demo request. This creates the opportunity, and the deal later closes.

Here is how each model credits the same $120,000. Time-decay uses 10/20/30/40% across the four touches; U-shaped and W-shaped use the common splits above.

Same deal, same touches, and LinkedIn's share runs from nothing to all of it. The model is a choice about what you value, not a measurement. Pick it before looking at which answer you prefer.

Why models disagree with your CRM

Every model only sees the touches that were tracked. Consent banners, ad blockers, people switching devices, and sales calls never appear. If the CRM record has no source, the deal can't be credited to anything at all. This is why the first job is checking how many CRM deals carry a source, and why GA4 and the CRM never match exactly.

Choosing a model for long B2B sales cycles

Start with the data, not the model. Check what share of closed deals in the CRM have a recorded source, and whether lead and opportunity dates are captured. If both are solid, a W-shaped model fits B2B well because it rewards the touches that move a deal between stages. If they are not, compare first-touch and last-touch side by side: the gap between them shows how much the choice matters, and fixing the tracking is worth more than any model.

Whatever you choose, agree it in advance and run every channel through the same model, so no platform gets to mark its own homework. That is what cross-channel attribution means in practice, and it is the first thing a free marketing data audit checks.

Questions people ask