Measuring content that isn't an advert

Content gets reported in views because views are available, unambiguous and large. They are also close to meaningless on their own, and everybody involved privately knows it.

The difficulty is that the honest alternative — that the effect is real but hard to attribute cleanly — is not something most people feel able to present. So the view count gets reported again, and the question of whether any of it worked stays unanswered for another year.

There is a more useful position between those two.

Why views persist

A view count is a genuine measurement of something. It is simply a measurement of distribution rather than effect.

It tells you the piece was served and briefly retained. That distinction matters most for short-form, which is served to people who were not looking. It does not distinguish between someone who watched four seconds while scrolling and someone who watched it twice and forwarded it to a colleague. Those two are worth wildly different amounts, and averaging them produces a number that describes neither.

It survives because it is easy to obtain and hard to argue with. Both are the wrong criteria.

What content is usually actually for

Most commissioned content is trying to do one of four things, and each has a different signal.

Shorten the sales conversation. Something for the salesperson to send after the first call. The signal is whether they send it, and whether the questions it answers stop being asked.

Make an audience believe something specific. The signal is what people say when they arrive — unprompted mentions in enquiries, in calls, in meetings.

Reduce a burden. Recruitment content answering the same twenty questions, or onboarding material replacing a session run monthly. The signal is time recovered.

Establish presence with a small, known group. Twelve people in the world can award the contract. The signal is whether those twelve saw it, which is a question about placement and not a number a platform can supply.

None of these is measured by views. Two of them can be answered by asking three people.

What to record instead

Four things, none of which need a new tool.

Send rate. For anything intended to be used by a team, whether it is used at all is the first and most honest indicator. If the sales team is not sending it, nothing else matters.

Question displacement. Write down the objections currently coming up. If the piece works, some of them appear less often. This is unglamorous and unusually reliable.

Unprompted mention. Whether anyone references it without being asked. Rare, and worth more than six figures of impressions when it happens.

Completion, not views. If a platform gives you one number, take the proportion who finished rather than the number who started. Watch time is the closest thing to attention that platforms actually report.

Say the attribution problem out loud

Content marketing rarely produces clean attribution, and pretending otherwise is what makes budgets fragile.

A prospect may see a film in March, remember nothing consciously, and enquire in September after a colleague’s recommendation. No tracking model captures that honestly. Models that claim to are assigning credit by rule, not by evidence.

The more defensible position is to say plainly: this is a contributing channel, here is what moved while it was running, and here is what we would expect to see if it stopped. That is a harder conversation for a quarter and a much sturdier one over two years — because the moment a channel is defended with a number everyone quietly distrusts, it is one budget review from being cut.

Set the measure before commissioning

The practical change is to decide what would count as working before anything is made.

At the point the brief is written, answer three questions:

1. What should be different afterwards — for whom, and observable how?
2. What are we recording now, so there is a baseline to compare against?
3. When do we look — and are we prepared to conclude it did not work?

The third is the one that gets avoided, and it is the one that makes the other two worth anything. A measure nobody is willing to fail is not a measure. It is a reassurance.