The short definition
Attribution is assigning credit for a sale to the marketing that caused it.
The question sounds simple and is not, because customers rarely take a single path.
Why it is hard
A realistic Houston customer journey:
- Sees your van in the neighbourhood in March
- Has a problem in June and searches your trade plus the area
- Sees you in the local pack, reads reviews
- Asks a neighbour, who mentions your name
- Searches your business name directly
- Calls
Analytics records step five. The van, the reviews and the neighbour are invisible, and they may have been the decisive parts.
This is not a flaw in a particular tool. It is the nature of the problem.
The common models
Last touch. All credit to the final interaction. Simple, and it systematically over-credits branded search and under-credits everything that created awareness.
First touch. All credit to the first known interaction. Over-credits discovery, ignores what closed the deal.
Linear. Credit split evenly across known touchpoints. Fairer, still limited to what was tracked.
Position-based. More weight to first and last. A reasonable compromise.
Every model is a simplification. The useful question is not which is correct but which misleads you least for the decision at hand.
What actually works for a small business
Two things together:
Track what you can. Conversion tracking, call tracking numbers, campaign parameters on links.
Ask. "How did you hear about us?" recorded on every enquiry. Imperfect, and it captures word of mouth and offline sources that no tool can see.
Neither alone gives a real picture. Together they are enough to make good decisions.
What attribution cannot do
It cannot prove causation. That a customer clicked an ad does not prove the ad caused the purchase — they may have been going to call you regardless, having already decided.
The honest question is incrementality: what would have happened without the spend. That is harder to answer and usually needs a deliberate test rather than a report.
The practical standard
You do not need perfect attribution. You need enough to tell which channels are clearly working, which are clearly not, and which are unclear.
That is achievable with basic tracking, an asked question, and a record of which enquiries became jobs.
Starting
Add a source field to your job records and ask every caller. Sixty days of that beats any attribution model applied to data you do not have.
Frequently asked questions
What are the main attribution models?
Last touch, first touch, linear and position-based. Each is a simplification — the useful question is which misleads you least for the decision you are making.
Why does attribution undercount word of mouth?
Because referred customers usually search your business name before calling, so the referral is recorded as branded search. The neighbour who recommended you never appears in any report.
Can attribution prove a channel caused a sale?
No. A customer clicking an ad does not prove the ad caused the purchase — they may have already decided. That question is incrementality, and it needs a deliberate test rather than a report.
What is good enough for a small business?
Enough to tell which channels are clearly working, which clearly are not, and which are unclear. Basic tracking, asking how they heard about you, and recording which enquiries became jobs.