Reading Your Analytics: A Quick Primer
Cost per verified action is the north star: it ties spend to a real, checked outcome and is comparable across campaigns. Read seven-day rolling averages, never daily figures.
Most campaign dashboards show you the least useful number first, largest and at the top. This is a short guide to reading past that.
Vanity vs. Value
Impressions and raw clicks feel good and tell you almost nothing. A campaign with half the clicks but twice the verified actions is the better campaign, every time.
The problem with click metrics is not that they are wrong — they are accurately counting something. The problem is that optimising them actively degrades results. Cheaper clicks are reliably found among the traffic least likely to do anything, so a campaign optimised for cost per click gets better at the metric and worse at the outcome.
Worked example, same spend:
| Campaign A | Campaign B | |
|---|---|---|
| Spend | $500 | $500 |
| Clicks | 2,000 | 800 |
| Cost per click | $0.25 | $0.63 |
| Verified actions | 40 | 95 |
| Cost per verified action | $12.50 | $5.26 |
Campaign A wins on the number most dashboards lead with. Campaign B is more than twice as efficient at the thing you are buying. If you were optimising cost per click you would cut B.
Cost per Verified Action
This is your north star. It tells you what you actually paid for a real, checked outcome — and it is directly comparable across campaigns.
The comparability is the underrated part. Cost per click is not comparable across campaigns with different objectives, audiences or packages, because a click means something different in each. A verified action means the same thing everywhere: somebody real did the thing, and it passed verification.
That makes it the right basis for every decision you make:
- Which campaign gets more budget? Lowest cost per verified action.
- Which segment do I cut? Highest cost per verified action.
- Did that change work? Compare cost per verified action before and after.
- Is this channel worth it at all? Cost per verified action against what that action is worth to you.
Note that spend is only drawn against verified activity, so there is no separate waste figure to subtract. Failed reviews are not billed. Your reported spend and your useful spend are much closer together than on an impression-based channel — which is also why comparing your cost per click against a display network is meaningless in both directions.
Reading Trends, Not Days
Look at seven-day rolling averages. Daily numbers swing for reasons that have nothing to do with your campaign.
There is a second, more specific reason to avoid daily figures here: verification lag. A thorough review takes longer to write and longer to verify, so today’s numbers systematically over-represent fast, shallow activity. Read a campaign at hour twelve and the careful publishers have not reported yet.
The failure mode is expensive and common. You judge early, cut the segment that was merely slower to report, and scale the one that produced quick shallow results.
The rule: give any change at least 48 hours before judging it, and prefer a seven-day rolling average to a daily figure. If you need a faster read than that, the answer is a bigger sample, not an earlier look.
How Much Data Is Enough?
Roughly fifty verified actions per week. Below that, week-to-week variation is mostly noise and you will end up optimising against randomness — reading a difference that would reverse itself next week.
If your budget cannot produce fifty a week, run for two weeks instead of concluding after one. Longer beats louder.
A Weekly Reading Routine
Twenty minutes, once a week, beats checking daily:
- Pull cost per verified action for the campaign as a whole, seven-day average.
- Compare against last week. Direction matters more than the absolute number.
- Segment it — by region, interest, device, publisher tier — and rank.
- Cut the bottom 40% and move that budget to the top performers. Keep total spend flat.
- Read the actual reviews, not just the counts. Note anything raised independently by three or more reviewers.
- Make one change. Write down what you expect to happen.
Step six matters more than it looks. Writing the prediction down before the change is the cheapest way to find out whether you are actually learning or just reacting.
Metrics That Mislead
- Impressions. Measures delivery, not interest.
- Raw clicks. Optimising these finds the worst traffic.
- Click-through rate in isolation. A high CTR to a page nobody converts on is a well-written promise you are not keeping.
- Total actions without cost. A campaign producing more actions at triple the cost is not winning.
- Any single day. Verification lag guarantees it is incomplete.
Where Campaign Analytics Stops
Analytics tells you where something goes wrong. It never tells you why. A drop-off at step three is a fact; the reason is a question only feedback answers.
The efficient loop: analytics to find where, reviews to find why, analytics again to confirm the fix worked. Verified user feedback: how to tell real signal from noise covers the qualitative half properly.
Frequently Asked Questions
What is a good cost per verified action?
There is no universal number — it depends entirely on what an action is worth to you. The useful comparison is against your own baseline over time, and against the revenue a verified action eventually produces. A $12 cost per verified action is excellent for enterprise software and ruinous for a $9 product.
Why is my cost per click worse than my display campaigns?
Because you are buying something different. A verified action includes a real person’s attention and, on engagement campaigns, their writing. Comparing it to an impression-based CPC is not a like-for-like comparison in either direction.
How long before I can trust a number?
48 hours minimum for any single change, and roughly fifty verified actions before a weekly figure is more signal than noise. If your budget cannot produce fifty a week, extend the period rather than concluding early.
My numbers dropped sharply for one day. What happened?
Almost certainly nothing about your campaign. Daily figures swing for unrelated reasons, and verification lag means any single day is incomplete by construction. Read the seven-day rolling average.
Should I look at total verified actions or cost per verified action?
Cost per verified action, for decisions. Total actions without cost attached will tell you a campaign spending triple is winning. Use the total only to check you have enough volume to trust the average.
Related Reading
Setting a budget that actually converts covers sizing budget around the fifty-action rule. Understanding campaign targeting options covers making cuts by segment, and advertiser campaign promotion: 5 mistakes that waste your budget covers the reporting errors that cost the most.
On getting conversion data in without manual tagging, see the new SaaS platform integrations.
Further Reading
