Marketing attribution models explained

What it is

Attribution is the practice of giving credit for a sale, signup or qualified lead to the marketing touchpoints that came before it. A touchpoint can be an ad click, an email visit, a webinar signup or a sales call that started from a campaign. The model is the rule that splits credit. First touch credits the first known interaction. Last touch credits the final one. Linear splits credit evenly. Position based weights the first and last touches more. Time decay favors recent touches. Data driven models learn weights from your conversion paths.

Attribution is not incrementality. Incrementality asks whether the sale would have happened without the campaign. Attribution asks which recorded touchpoints were present and how a model scored them. Cookies, consent and cross device behavior leave gaps. Treat attribution as a map with missing roads, not a ledger.

Most companies need a shared definition before a complex model. Decide what counts as a conversion, which channels you track, how long a touchpoint stays eligible and how you handle direct traffic. Those choices change the answer more than the model name.

How it works

Tracking starts at the link. Put UTM parameters on paid and owned links: utm_source, utm_medium, utm_campaign, and where useful utm_content and utm_term. Keep a naming sheet so values do not drift. Capture click IDs such as gclid and fbclid. Add hidden form fields for first source, last source, landing page and referrer. Push those fields into the CRM on the lead or contact record.

Then join touchpoints to outcomes. Google Analytics 4 has attribution reports for conversions you define. Ad platforms show their own view, usually from clicks and views they recorded. Your CRM shows the source on the opportunity or deal. A clean setup connects these views through common campaign names, dates and contact IDs. Without that join, each team argues from a different report.

Model choice controls the story. First touch shows what starts interest. Last touch shows what closes. Linear shows breadth. Position based shows the opening and the final push. Time decay shows recent influence. Data driven models can surface paths you did not expect, but they need enough volume and clean inputs. Pick the model that answers your question, write down the lookback window, and keep it stable for a quarter.

Why it matters for a business

Attribution helps you decide where to spend next. If paid search looks weak on last touch but starts paths that close later, cutting it can starve the pipeline. If retargeting gets credit for sales that would have happened anyway, it looks better than it is. A model gives finance and marketing a shared language for those debates.

It also improves sales conversations. When a rep sees the webinar, pricing page visit and demo request on one record, the first call starts warmer. When a marketer sees which campaign source sits on closed revenue, the weekly report moves from clicks to pipeline.

Use attribution to find leaks: forms that lose source data, campaigns that send traffic to untracked pages, channels that only appear late in the path. Fix the data first, then use the model to guide tests.

Common mistakes

The common mistake is reporting last click as the whole truth. Last click is easy and useful for closing campaigns, but it undervalues early education. The next is dirty UTM tags. One team uses email, another uses Email and a third uses newsletter. Reports split into fragments and trust drops.

Teams also compare numbers across tools without adjusting definitions. Analytics, a paid social platform and your CRM count conversions at different moments, with different identity rules and windows. Expect differences. Document why they exist instead of hunting for one perfect number.

Direct traffic and offline touchpoints get forgotten. A buyer who types your brand name after hearing a podcast shows up as direct. A sales call that started at an event may have no digital source. Add a self reported field on forms and let sales update the source when they learn it. Do not change models every month. A new model resets the baseline. Pick one primary model for budget talks and one secondary for channel learning, then freeze them for a quarter.

How to get started

Write a one page definition sheet. Name the conversions, channels, lookback windows and rules for direct traffic and offline leads. Assign one owner for UTM naming. Put the sheet where marketing, sales and finance can find it.

Fix collection before dashboards. Audit your top landing pages for UTM coverage and hidden form fields. Check that the CRM stores source, campaign and landing page on the lead and carries them to the opportunity. Confirm consent capture in analytics and ad tools. Run a test campaign with known tags and follow it from click to closed deal.

Then pick two models to compare, such as first touch and last touch, or position based and data driven. Build one report that shows spend, pipeline and closed revenue by channel. Review it monthly with sales. When data is stable, add holdout tests or paused campaigns to check what attribution cannot tell you.

You do not need a large stack. Clean CRM data, tagged links, a source field on forms and one dashboard beat a complex model on messy data. If you want the campaign plumbing set up in one pass, one option is the Cloudverg Marketing automation build at a fixed price of $7,500. It covers sending domains, forms, lead scoring, CRM routing and campaign reporting tied to pipeline. Compare it with doing the work in house or hiring a specialist.

Frequently asked questions