Analytics tools are generous with numbers and unhelpful about which ones matter. A standard dashboard will show you dozens of metrics, most of which will not change a single decision you make.
A metric earns its place by one test: if this number moved, would I do something differently? If not, it is decoration.
Vanity metrics and what to use instead
| Commonly reported | Problem | Better question |
|---|---|---|
| Impressions | Measures delivery, not interest | Did anyone act? |
| Followers | Accumulates; does not indicate current reach | How many saw the last post? |
| Total sessions | Hides that sources behave differently | Which sources produce conversions? |
| Bounce rate alone | A single-page visit can be a success | Did they do what the page was for? |
| Cost per click | Cheap clicks can be worthless | Cost per qualified outcome? |
None are useless. They are diagnostic, not evaluative. Impressions help explain why conversions fell. They do not tell you whether the month was good.
Three layers worth tracking
Structure your reporting so each number has a job.
1. Business outcomes
Revenue, enrolments, bookings, qualified enquiries. These answer "did it work". They are lagging. They tell you about the past — but they are what the work is for.
2. Conversion metrics
Conversion rate, cost per acquisition, lead-to-customer rate. These explain how the outcome happened and are where most optimisation decisions get made.
3. Diagnostic metrics
Traffic by source, click-through rate, scroll depth, page speed, form abandonment. These tell you where a problem is. You look at them when layer two moves unexpectedly, not every week.
Before adding a metric to a report, write the sentence: "If this goes down, I will ______." If you cannot finish it, leave the metric out. Reports that contain only actionable numbers actually get read.
Attribution is genuinely imperfect, plan for it
Attribution assigns credit for a conversion to the touchpoints before it. It is useful and it is never fully accurate, for reasons that are structural rather than fixable:
- People switch devices, research on a phone, buy on a laptop.
- Privacy controls, tracking prevention and ad blockers remove data by design.
- Offline steps, a phone call, a conversation, a visit — are invisible to web analytics.
- Different models credit different touchpoints, so two reports of the same reality disagree.
Practical consequences: expect platform-reported conversions to exceed analytics-reported ones, because each platform credits itself. Use one source as your reference and stay consistent, watch the trend rather than the absolute number, and ask new customers how they found you, self-reported attribution catches what tracking misses.
Learn this by actually doing it.
ForGrowth is built around one idea: you learn digital marketing by doing digital marketing, on real tools and real projects.
Explore the Program →Set up tracking that reflects the business
Default analytics tracks pageviews. Pageviews are not outcomes. Define events for the actions that matter:
- Enquiry form submitted
- Phone number tapped
- WhatsApp opened
- Key page reached
- Meaningful scroll depth on long pages
Then verify each one fires. Test in real time, submit the form yourself, and confirm the event appears. An untested conversion setup is a very expensive way to be confidently wrong for a month.
Segment, or averages will mislead you
An overall conversion rate of 3% might be 6% on desktop and 1.5% on mobile, a serious mobile problem hidden by an average. Or one campaign carrying four that lose money.
Segment by device, source, campaign, and location. The interesting information is almost never in the total.
Watch the trend, not the day
Daily numbers on small volumes are mostly noise. Three conversions instead of five is not a 40% decline; it is a Tuesday.
- Compare like periods, week to week, not weekday to weekend.
- Wait for enough data before concluding anything.
- Note external events. A festival, a holiday, a price change all show up in the data and none are campaign performance.
A report that gets used
The best reporting habit is small and regular. Weekly, on one page:
- The business outcome number, versus last week.
- Cost per qualified outcome.
- The single biggest change, and the likely reason.
- One thing being changed this week as a result.
Point four is what separates reporting from analytics. A report that never leads to a decision is a record, not an analysis.
Questions about upcoming batches?
Weekday and weekend options are available. Talk to a ForGrowth advisor about which one fits your schedule.
Enquire Now →