Marketing Attribution Models, Explained

Two people can look at the exact same customer journey — a paid ad click on Monday, an organic search visit on Wednesday, a phone call on Friday — and reach opposite conclusions about which channel "worked," simply because they're applying different attribution rules. That's the part most explanations of attribution skip: it's not a measurement problem, it's a modeling choice. There's no single objectively correct answer for how to split credit across touchpoints — only models that make different tradeoffs, and some that fit a small service business's data volume far better than others.

Why attribution models exist in the first place

If every customer converted after a single interaction with a single channel, attribution wouldn't be a question — you'd just count conversions per channel and be done. Real customer journeys rarely work that way. Someone might see your Google Business Profile, visit your website a few days later from an organic search, then call after seeing a retargeting ad. Three touchpoints, one conversion. Attribution models exist to answer a genuinely ambiguous question: how much credit does each of those three touchpoints deserve?

The main models, explained

• First-touch attribution. 100% of the credit goes to whatever channel first introduced the customer to your business. Simple to understand and useful for understanding what drives initial awareness, but it ignores everything that happened afterward — a channel that gets someone in the door but contributes nothing to actually closing them still gets full credit under this model.
• Last-touch (last-click) attribution. 100% of the credit goes to the final touchpoint before conversion. This is the default most small businesses use whether they realize it or not, since it's the simplest to track and requires no modeling — whatever a lead says brought them in, or whatever channel a form submission's UTM parameters point to, gets the credit. Its weakness mirrors first-touch's: it ignores everything that happened earlier in the journey, which can undervalue channels like SEO that often plant the seed without capturing the final click.
• Linear attribution. Credit is split evenly across every touchpoint in the journey. Fairer in the sense that no touchpoint is ignored, but it treats a passing glance at an ad and a 10-minute website visit as equally valuable, which usually isn't true.
• Time-decay attribution. Credit is weighted toward touchpoints closer to the conversion, with earlier touchpoints getting progressively less. This is a reasonable middle ground for longer sales cycles, where the final nudge probably deserves more credit than an interaction from six weeks earlier, but earlier touchpoints shouldn't be zeroed out entirely.
• Position-based (U-shaped) attribution. A fixed, heavier share of credit (often 40% each) goes to the first and last touchpoints, with the remaining share split across whatever happened in between. This reflects the intuition that how someone first found you and what finally convinced them both matter more than the middle steps.
• Data-driven attribution. Instead of a fixed rule, credit is assigned algorithmically based on patterns in your own actual conversion data — which touchpoint combinations tend to precede a conversion versus not. This is what Google Analytics 4 uses by default for most reports, but it requires a meaningful volume of conversion data to produce a stable, trustworthy result, which is exactly the constraint that makes it a poor fit for very low-volume small businesses.

Which model actually fits a small service business

For most local service businesses converting a modest number of leads per month, the honest answer is: last-touch, tracked consistently, is usually good enough — not because it's the most accurate model in theory, but because the more sophisticated models need conversion volume they don't have to produce a reliable answer. A multi-touch or data-driven model built on 20 conversions a month will bounce around from one small sample to the next in ways that look like insight but are really just noise.

The practical middle ground worth adopting even at small scale: track last-touch as your primary number, but also note first-touch when you can (often from the intake question "how did you first hear about us," compared against "what made you call/submit today"). That gives you a rough sense of which channels are doing the introducing versus which are doing the closing, without needing a full modeling system.

What attribution can't fix

No attribution model compensates for missing data. If you don't have call tracking, UTM parameters, or a consistent intake question capturing lead source, there's no model sophisticated enough to correctly split credit across touchpoints you never recorded in the first place. Attribution modeling is the last step in a chain that starts with actually capturing source data — see call tracking, explained and how to actually track marketing ROI for the tracking infrastructure that has to exist before any attribution model has real data to work with.

The takeaway

Attribution isn't a solved problem you're failing to implement correctly — it's a set of tradeoffs, and the right choice depends on your conversion volume and how long your typical sales cycle runs. For most small service businesses, that means picking last-touch (or a light first-touch/last-touch hybrid), applying it consistently, and being honest that any single-touch model will systematically undercount channels — often SEO — that do more of their work earlier in the journey than at the final click.