Most analytics setups produce a dashboard nobody opens twice. It reports activity — sessions, pageviews, bounce rate, average duration — and none of it tells anyone what to do differently on Monday.
The fix is not more data. It is deciding, in advance, which small set of numbers would actually change a decision, and instrumenting those properly.
The five that earn their place
For most service businesses, everything worth knowing sits behind five numbers.
- Qualified enquiries per week. Not form submissions — enquiries that meet your basic fit criteria.
- Conversion rate by entry page. Which pages produce those enquiries, per visitor, not in absolute volume.
- Time to first response. The single operational metric most correlated with winning the work.
- Cost or effort per qualified enquiry, by channel. What each source actually costs to produce a usable lead.
- Close rate and value by source. Which channels produce clients, not just conversations.
Everything else is diagnostic: useful when one of those five moves and you need to find out why, useless as a weekly report.
Define an event before you build a chart
Analytics goes wrong at the definition stage, not the tooling stage. If 'lead' means a form submit to marketing and a qualified opportunity to sales, every conversation about performance is two conversations.
Write a short definition list: what counts as an enquiry, what makes it qualified, when the clock on first response starts and stops. Put it somewhere both teams can see. This is dull and it prevents most analytics arguments.
Track the funnel, not the page
A page view tells you almost nothing on its own. The useful unit is the step: landed, engaged with the key section, started the form, completed it, was qualified, booked, closed.
Instrumenting those steps individually is what lets you say something precise — 'sixty percent start the quote form and thirty percent finish it' — instead of something vague, like 'the page is underperforming'.
Respect the visitor while you measure
You can get everything above without invasive tracking. Prefer privacy-respecting, first-party measurement, collect only what feeds a decision, honour consent choices properly rather than cosmetically, and keep personal data out of your analytics tool entirely.
This is also a performance win: the heaviest scripts on most marketing sites are the ones collecting data nobody uses.
Review on a rhythm
Data without a meeting is decoration. Set a short monthly review with a fixed agenda: what moved, what we changed last month, what we are changing next, and what one experiment we are running.
Keep a written log of changes with dates. Six months in, that log — not the dashboard — is what lets you tell which decisions worked.
The test for any metric
Before adding anything to a report, ask what you would do differently if it doubled or halved. If there is no answer, it is not a metric; it is trivia with a chart.
Five numbers, reviewed monthly, with a change log beside them, will beat a fifty-widget dashboard in every business that has actually tried both.
Want to know what's holding your website back?
Run a free Insight Audit™ — the InsightWeave platform that scores your site and returns prioritised fixes.