Think Build Implement Repeat
Data & Scraping

Working Out What Marketing Actually Works, Cheaply

Last updated:

Sophistication is not the constraint

Multi-touch attribution models are marketed as the answer. For a business with modest volume and consent-limited tracking, they produce confident numbers built on partial data, which is worse than a simple approach used honestly.

The constraint is data capture at the moment of enquiry, not the model applied afterwards.

Three things that work

  1. Capture the source on the enquiry record — UTM parameters carried through the form into the CRM, automatically.
  2. Ask the customer. One free-text question on the form or in the first call: how did you hear about us? Self-reported and imperfect, and it catches the offline and word-of-mouth influences no tracking sees.
  3. Carry the source through to revenue, so you report on customers won by source rather than on leads by source.
The self-reported question is the one analysts dislike and business owners find most useful. It is the only method that captures “a colleague recommended you”, which is frequently the largest single source.

Report on what converted, not what arrived

Lead volume by channel flatters channels that generate enquiries and no business. Customers won and revenue generated by source is the number that should direct spend.

This requires the CRM to be updated with outcomes, which is the discipline the whole approach depends on.

Accept that some influence is invisible

  • The prospect who saw an ad, forgot, and searched your name three weeks later
  • The referral that came from someone who read an article you published
  • The event conversation that led to a search six months later
  • Anything on a device or browser that blocked tracking entirely

Attribution is directional. Treating it as precise leads to defunding channels whose contribution is real and unmeasurable.

Test by turning things off

The most reliable attribution method available to a small business is a holdout: stop a channel for a month and watch what happens to enquiries. Crude, disruptive and genuinely informative.

Do it in a quiet period, with one channel at a time, and give it long enough to account for your sales cycle.

What it costs to set up

Source capture through to the CRM is typically a few thousand pounds of integration work. The self-reported question is free. Reporting on won business by source is usually configuration.

That is the whole system, and it outperforms most modelled attribution in businesses of this size.

Frequently asked questions

What about tracking that consent blocks?

It will block some, which is why source capture at enquiry — a first-party record in your own system — and the self-reported question matter more than client-side tracking.

Should we use first-touch or last-touch?

Record both if you can, and report last-touch for immediate decisions and first-touch to understand discovery. Consistency matters more than which you pick.

How long should attribution look back?

Match it to your sales cycle. A business with a six-month cycle attributing on a 30-day window is systematically misreading its own marketing.

Is it worth paying for an attribution platform?

Rarely at small scale. The platforms are built for spend levels where a few percentage points matter in absolute terms.

Keep reading

Spending on marketing without knowing what works?

Source capture through to revenue is a small project with a large effect on spending decisions. Tell us what you run.

Book a free 30-minute call Get a project estimate WhatsApp us

Related services

What we build for problems like this one

Web ScrapingData Science