Measuring Marketing Now That Attribution Is Genuinely Broken

There was a period when digital marketing could credibly claim to know exactly which advertisement produced which sale. That period is over, and pretending otherwise leads to worse decisions than accepting it.

Cookie restrictions, privacy features, multi-device journeys, and platform data limits mean no system sees the whole picture. Meanwhile every platform reports generously on its own contribution, so if you add up what Google, Meta, and your analytics each claim, you will have attributed the same customer three times.

Here is how to measure usefully anyway.

Accept that the numbers overlap

Someone sees your Instagram post on Monday, searches your name on Wednesday, clicks a Google ad on Friday, and enquires. Meta claims it. Google claims it. Your analytics records a paid search conversion. All three are partly right.

The mistake is picking whichever number is highest and reporting it, or picking whichever is lowest and concluding a channel does not work. Both happen constantly.

Measure the total, not just the parts

The most reliable signal available to most businesses is simple: total enquiries, tracked monthly, against total spend.

It sounds crude. It is also the only number nobody can inflate. If you spent more and got proportionally more enquiries, something is working. If spend doubled and enquiries did not move, something is not.

Track it as a running chart rather than month by month, because monthly variation is noisy and the trend is what matters.

Ask people how they found you

Genuinely the most under-used measurement tool available. One question at the point of enquiry, asked consistently, and within two months you have a picture that no analytics platform can give you.

It is imperfect. People misremember, and they often name the last thing they touched rather than the thing that convinced them. But it captures word of mouth, offline advertising, and referrals, all of which are invisible to digital tracking and often substantial.

Make it a dropdown on the form and a standard question on calls.

Use unique routes for offline channels

For radio, print, hoardings, and events, build attribution into the creative rather than hoping to infer it afterwards.

• A dedicated phone number used in that channel only
• A short URL or QR code to a campaign-specific page
• An offer code mentioned only there

None of this is difficult and it converts an unmeasurable channel into a measurable one.

Watch branded search as a proxy

When awareness advertising works, people search for your name. Branded search volume in Search Console is one of the cleaner indicators of whether upper-funnel activity is doing anything.

If you run a radio campaign and branded searches rise during and after the flight, that is evidence, even without a click path.

Run holdout tests when a channel is large enough

The most rigorous method available: turn a channel off for a defined period, or run it in some areas and not others, and observe what happens to total enquiries.

It feels uncomfortable and it is the only way to know a channel’s real incremental contribution. Businesses that do this occasionally discover a channel that reported well was mostly claiming credit for demand that existed anyway.

Report on decisions, not on data

A report that shows twenty metrics and recommends nothing is not a report, it is a data dump. The question every month is what to do differently, and the numbers exist to answer it.

A useful monthly report contains: total enquiries and cost, the trend against previous months, what changed and why, what we learned, and the specific recommendation. Three pages at most.

Metrics worth ignoring

• Impressions and reach. Context at best, never a result.
• Followers. Almost entirely unrelated to revenue.
• Click-through rate in isolation. A high rate on traffic that never converts is a problem, not an achievement.
• Bounce rate, which is widely misinterpreted. Someone reading your whole page and leaving satisfied is not a failure.
• Keyword rankings alone, without traffic or enquiries attached.

The three numbers that matter

For most businesses, everything reduces to:

• Total qualified enquiries this month against previous months
• Total marketing cost including fees and spend
• What proportion of enquiries closed, and at what value

From those three you can calculate cost per customer, which is the only figure that tells you whether to spend more or less. Everything else is diagnostic detail explaining why those numbers moved.

Being honest about uncertainty

An agency presenting a perfectly clean attribution picture is showing you a simplification, and possibly one designed to flatter their own channel.

The better position is knowing roughly what works, being clear about what is uncertain, and running occasional tests to reduce that uncertainty where it matters most. That is less satisfying than a dashboard with precise percentages, and considerably more likely to be true.

Read our FAQs on reporting and results, or ask how we would measure your marketing.