Blog / Digital growth /

Choosing a meaningful success measure for a new resource

Your new resource is ready to publish, and somebody asks how you’ll measure success. “Traffic” is the easy answer. It’s also the one that tells you least about whether anyone was helped.

Start with the job the resource is supposed to do. A guide that answers a support question, a template that helps someone get started and a page built to generate enquiries each need different evidence. Pick the measure after you’ve named the job, not after the numbers come in.

Start with the intended change

Write one sentence describing what a reader should be able to do after using the resource. Then choose an observable action close to that outcome. Downloads, engaged visits, completed forms or tool use may fit, but only if they represent the intended job.

A pageview isn’t proof of usefulness. A download count says little if the file never gets opened. Define what counts and where the data comes from.

Pair outcome with context

Use one primary measure and a small number of diagnostic ones. For a guide, look at search entrances, navigation behaviour and a relevant next action, within your analytics setup and consent rules. For a lead resource, separate the form completion from qualified follow-up.

Document the limits: missing tracking, repeat visits, changes in promotion. And don’t choose a flattering measure after the results arrive. I’d rather publish a modest number I trust than a big one I can’t explain.

Review and adapt

Give the resource a review period that suits its traffic and purpose. Check whether readers can find it, whether it answers the question and whether the next step is clear.

If the number is low, look at distribution and usability before you call the topic a failure. A good success measure helps you make the next decision. It isn’t there to fill a report.

Keep the work reviewable

Choose one question the current setup can answer before you build a dashboard of everything.

Record the date, source and conditions behind each number. A launch week looks nothing like a normal one.

Check one ordinary case against one exception, such as a page where tracking is missing or a download behind a different form.

Know which system is the source of truth and which only shows a copy. A pasted figure in a slide should never outrank the analytics property.

Keep confirmed results, likely explanations and next steps in separate lines of the report.

Revisit the measure when something real changes, such as a new template, a consent update or a new distribution channel, rather than only on a calendar reminder.

If two reports disagree, check the definitions, the timing and the scope with their owners before you combine them. Keep the difference visible until you know why it exists.

Look at the path a reader actually takes to the resource, not only the page where you expect them to start. A link in a newsletter and a search result can lead to quite different behaviour.

Before you apply one measure to every resource on the site, try it on a single representative page and check what it reports after publication.

A useful handover names the person who can answer a question about the content, the analytics or the audience. I’d rather see a name than “marketing”.

If your data is incomplete, say what is missing and pick one next check that reduces the uncertainty. Don’t invent a plausible figure to make the report look complete.

Once you’ve changed the resource, go back to the same example and confirm that the reader’s task still works.

Keep the maintenance note brief but specific: the owner, the source and the event that should trigger another review. A real change, such as a redesign or a new tracking setup, beats a repeating reminder.

When you report a result, put its limitation in the same sentence or paragraph as the recommendation. That way a colleague sees what is known and what isn’t.

Record each change to the measure with a short reason and an owner, and keep the earlier definition available for comparison.

Know which system is the source of truth for each number and which merely displays a copy. A figure pasted into a deck should never override the analytics property it came from.

Test the measure on one representative resource before applying it to the whole library, and check the live result after publishing. Extend it only when the report matches what you meant it to show.

Separate a confirmed result from a likely explanation and from a proposed action. They can sit in the same report, but they carry different weight, and a reader should be able to tell which is which.

What useful behaviour should this resource make easier?