Attribution
Assigning credit for a conversion across the touchpoints that led to it, and accepting that every method for doing so is wrong.
In depth
What it really means
Attribution assigns credit for a sale across the marketing touchpoints preceding it. Every model does this differently, and every model is wrong, because credit is not a physical property of a touchpoint and there is no correct answer waiting to be found.
That sounds defeatist and is the opposite. Once you accept that attribution is a decision-making tool rather than a measurement, you can pick a model, apply it consistently, and use the trend. Teams that treat attribution as truth spend years buying tools to find a number that does not exist.
The models, and what each one hides
| Model | Gives credit to | Systematically hides |
|---|---|---|
| First touch | The first interaction | Everything that closed the deal |
| Last touch | The final interaction | Everything that created demand. Usually flatters branded search and paid. |
| Linear | All touchpoints equally | That some touchpoints mattered far more |
| Time decay | Recent touchpoints more | Early awareness work |
| U-shaped | First and last, 40% each | The middle, where evaluation happens |
| Self-reported | Whatever the buyer says | Nothing systematically, but memory is unreliable |
The scale of the problem
A considered B2B purchase involves around 27 distinct interactions, buyers consume roughly eleven pieces of content before contacting a vendor, and 83% have already defined their requirements by the time they speak to sales.
Most of that happens in places you cannot track: a Slack group, a podcast, a colleague’s recommendation, an AI assistant’s answer. Buyers now typically make first contact around 61% of the way through their process, which means the majority of the journey is invisible to your analytics by design.
Pros & cons
Pros
- Applied consistently, it shows directional shifts that are genuinely useful.
- Forces the conversation about which channels get funded.
- Self-reported attribution is cheap, quick to add, and often more accurate than tracked data.
Cons
- No model is correct, and comparing across models produces contradictory answers.
- Systematically undercounts dark social, word of mouth, communities and AI answers.
- Expensive tools promise resolution they cannot deliver.
- Political. Whichever model is chosen advantages someone’s channel.
The mistake people make
Using last touch by default because it is what the tool shows. Last touch credits whatever happened immediately before conversion, which in B2B is almost always branded search or a direct visit. That makes brand look free and makes every piece of demand-creating content look worthless, which is precisely backwards. If you can only run one model, run first touch and add a self-reported field on your forms.
Best practices
- Pick one model, write down why, and keep it for at least a year.
- Add a free-text “how did you hear about us” field and read the answers.
- Treat attribution as directional, and never report it to two decimal places.
- Pair it with pipeline influenced so late content is not erased.
- Accept the dark funnel exists and stop trying to instrument it.
- Watch trends across quarters rather than absolutes in a month.
FAQs
What is marketing attribution?
Assigning credit for a conversion across the touchpoints that preceded it, using a chosen model.
Which attribution model is best?
None is correct. First touch plus a self-reported field is the most useful cheap combination for B2B SaaS.
Why does last touch mislead?
It credits the final click, which in B2B is usually branded search or a direct visit, so it makes the work that created the demand look worthless.
What is the dark funnel?
The untracked influence: communities, podcasts, private messages, word of mouth and AI answers. It is large and growing.
Is self-reported attribution reliable?
More reliable than most people assume, and often closer to reality than tracked data because it captures untracked channels.
Keep reading
Related on LymLyt
Beyond LymLyt
Further reading
Want this working on your site?
We build the content behind the term, ranked in search and cited by AI.
Book a 30-min call →