Which Marketing Metrics Should You Stop Reporting?
Retire the Number, Keep the Question It Was Meant to Answer
Short answer
Stop reporting any metric that has never changed a decision, and replace each one with the narrower figure that would. The usual candidates are total impressions as a headline, follower totals, cross-channel sums, average sentiment scores, estimated competitor spend, and post volume as an output measure.
1. Which marketing metrics should we stop reporting?
Eight metrics come up in almost every small-team report and earn their place in none of them. Each has a narrower replacement that answers the question the original was standing in for.
- Vanity metric
- A measurement that reliably rises with activity and cannot fall for any reason a reader would want to act on, so it reports effort rather than result.
| Stop reporting | Report instead | Why the swap works |
|---|---|---|
| Total impressions as a headline | Reach on the specific pieces you chose to promote | Impressions rise with spend and with repeat delivery, so the headline moves without anyone deciding anything |
| Follower count totals | Follower growth rate per channel, and share of your tracked set | A cumulative total only ever rises, so it cannot signal a problem. A rate and a share can both fall |
| Interactions summed across channels | One index per channel, measured against a baseline from that same channel | A sum of unlike units is dominated by whichever channel produces the biggest raw counts |
| An average sentiment score | The three most common complaint themes, with counts and quotes | Compressing review text into one number destroys the only useful part, which is what people are complaining about |
| Estimated competitor ad spend | Active ad count and creative refresh dates, both observed | Spend estimates are modelled and unverifiable. Ad counts and start dates are published facts |
| Website sessions as a success measure | Conversions and qualified conversations from those sessions | Sessions reward any traffic at all, including the traffic you would rather not have paid for |
| Email open rate | Click rate, reply rate, and list growth net of unsubscribes | Open tracking depends on a loaded image, and clients that prefetch images register opens nobody made |
| Post count as an output measure | Posts published against planned slots, with the theme each one served | Volume answers whether the team was busy. Slot completion answers whether the plan actually ran |
2. What makes a metric worth stopping?
Three questions decide it: has it changed a decision recently, can it fall for a reason you would want to know about, and is it observed rather than modelled. Two failures is enough to retire it.
Run every metric in the report through these three, in writing:
- Has this number changed a decision in the last two quarters? If nobody can name the decision, the metric is being reported out of habit rather than need.
- Can it go down for a reason you would want to hear about? A metric that only rises, or that falls only when something obvious breaks, carries no information in either direction.
- Was it observed or was it modelled? Modelled figures lose their caveats inside a recurring report and eventually get quoted as measurements, which is worse than not having them.
The first question does most of the work, and it is uncomfortable to ask out loud, which is why it rarely gets asked. A report where nine of eleven metrics fail it is completely normal, and it is not a sign anyone did anything wrong. Reports grow by addition because adding is easy and removing requires a conversation.
3. Why do vanity metrics survive in reports for so long?
Because they only go up. A number that cannot deliver bad news is comfortable to present, easy to collect automatically, and nobody has an incentive to be the person who removes it.
Consider a month where engagement halved and the follower total rose by forty. Both statements are true, and only one of them opens with growth. Given a choice, a report will lead with the second, and after a year of that the follower line is the metric everyone recognises and the engagement line is a chart nobody reads.
Four reasons a metric outlives its usefulness:
- It is monotone. Cumulative totals cannot fall, so presenting one is never uncomfortable.
- Somebody senior asked for it once, and nobody was ever told the question had been answered.
- It is the cheapest thing to collect, so it arrives in the report automatically whether or not anyone wanted it.
- It is comparable to what other companies quote publicly, which makes it feel like an industry standard rather than a habit.
None of those reasons is bad faith, and all four are fixed the same way: a scheduled review of the report itself, not just of the numbers in it. Once a quarter, ask which line item changed a decision, and delete on the spot.
4. Are impressions and reach useless, then?
No. Both are correct measurements with a specific place: reach for analysing your own content, impressions for paid reporting where impressions are the billing unit. Neither belongs as the headline of an organic report.
The problem is never that the number is wrong, it is that it is answering a question nobody asked in that meeting. Reach tells you how many people had the chance to respond, which is exactly what you want when judging whether a piece of content worked. Impressions tell you how much delivery you bought, which is what you want when judging a media plan.
Put either at the top of a monthly summary and it becomes a headline that grows with budget and cadence, which makes the report a description of activity. The fix is placement rather than deletion: keep both in the content and paid sections where they answer something, and let the summary carry results.
5. How do I retire a metric leadership asked for?
Bring the replacement to the same meeting, explain which decision the old metric was standing in for, and keep the old number available on request. Removal without a substitute reads as hiding something.
Five steps, and the order is what makes it work.
- Find out what the metric was originally for. Ask the person who requested it which question they wanted answered. It is usually a reasonable question, and the metric is a poor proxy that got established because it was the number available at the time.
- Bring the replacement to the same meeting. Show the old metric and the new one side by side for one period, so the audience can see what the swap buys them. A removal presented without an alternative will be read as an attempt to bury a bad number.
- Retire one metric at a time. A report that loses six lines in one month feels like a different report and invites a debate about the whole thing. Losing one per cycle is barely noticed and reaches the same place within two quarters.
- Keep the retired number available on request. Keep collecting it and put it in an appendix or a linked sheet. Availability removes the only real objection, which is that somebody might need it and no longer be able to get it.This also protects the history, so the series is intact if the metric is ever reinstated.
- Record the change inside the report. One line saying what was removed, what replaced it and when. Without it, the next reader sees a series that simply stops and assumes the data broke.
6. Which metrics should stay in every report?
Five: what competitors changed, your cadence against the field, conversions or qualified conversations, share of your tracked set, and one quality measure that carries text rather than a score.
| Channel | Posts per week | Accounts measured |
|---|---|---|
| 1.98 | 71 | |
| 1.68 | 77 | |
| 1.04 | 56 |
That is the pattern for every metric worth keeping: it either describes a change somebody has to respond to, or it comes with the base needed to judge it. Volume without a base, and rates without a comparison, both fail.
The five that survive, and what each is for:
- What competitors changed this period. The only section that reliably produces decisions, and the one most reports omit.
- Cadence against the field, per channel. Answers whether the publishing plan is competitive, not whether the team was busy.
- Conversions or qualified conversations. The result, however imperfectly attributed, stated with its known gaps.
- Share of your tracked set on one metric. A figure that can fall while your own numbers rise, which is exactly why it is worth reporting.
- One quality measure carrying text: the top complaint themes, or the actual comments. Numbers cannot say what to fix.
Oppira assembles the competitor-change section and the cadence comparison from daily collection, which matters here because those are the two lines most likely to be dropped when a report is assembled by hand under time pressure.
Key Takeaways
Swap, never simply delete
Every retired metric needs a named replacement that answers the question it was standing in for, or it returns in a quarter with more authority.
Three questions decide it
Has it changed a decision in two quarters, can it fall for a reason you want to know about, and was it observed rather than modelled.
Monotone metrics carry no information
A cumulative total cannot fall, so it can never signal a problem. Replace totals with rates and with share of a tracked set.
Impressions and reach are misplaced, not useless
Reach belongs in content analysis and impressions belong in paid reporting. Neither should headline an organic summary, where both grow with budget.
Cadence beats volume because it has a base
Tracked accounts averaged 1.04 posts per week on LinkedIn in June 2026, which turns your own cadence into an answerable question.
Retire one metric per cycle
Removing six at once starts a debate about the whole report. One per cycle is barely noticed and reaches the same place in two quarters.
Frequently Asked Questions
Explore More
Related analyses, benchmarks, and industry insights
Related Guides
Glossary Terms
Turn reading into a reaction
Oppira watches your competitors, keeps your playbook current, and drafts the response. Start free and see your market clearly by tomorrow.