PPC reporting: five numbers instead of a dashboard with thirty charts
A good report is not judged by its number of charts but by how many decisions came out of it. Which numbers to watch monthly, which weekly, and how to write a report someone actually reads.
- A report should answer three questions: what happened, why, and what we will do about it.
- Five numbers monthly are enough: non-brand result against margin, cost per customer, brand share, marginal CPA and a check against orders.
- A metric nobody has ever decided anything on does not belong in a report.
A typical monthly report runs to twenty pages, forty charts and ends with “we continue to optimise”. Nobody reads it all and nobody decides anything from it. Producing it costs several hours a month that could have been spent in the account.
A good report has one criterion: how many decisions came out of it.
Three questions a report should answer
- What happened? A few numbers against target and against the same period last year. Not against last month, if the field is seasonal.
- Why? A cause, not a description. “Revenue fell 12 %” is a description. “Revenue fell because two main categories were out of stock for three weeks” is a cause.
- What will we do? What we will do, what we expect from it, and when we will evaluate. Without that part a report is just accounting for the past.
Five numbers that suffice monthly
| Metric | Why this one | Compare it with |
|---|---|---|
| Non-brand cost of sale or ROAS | Shows how you are doing at winning new customers | Break-even from margin, not an industry average |
| Share of brand in the result | Tells you how far the headline figure can be trusted | The previous period; when it rises, the result is being distorted |
| Cost per customer | For services the only meaningful number | Margin on a deal |
| Marginal CPA | The only number for a budget decision | Gross margin on one conversion |
| Ratio of conversions to orders | A check that the numbers can be trusted | Its own value last month |
For online shops we would add the share of the range that received at least one impression in the month. It is the one metric from the feed that belongs in a management report.
What to check how often
- Daily, automated: anomalies. A jump in spend, a drop in conversions, disapproved products. This is not for a person to read — it is for something to alert on. We built ClientDeck for it, because a manual morning check across several accounts does not work.
- Weekly: the search terms report for the most expensive campaigns, budget status, new warnings in Merchant Center.
- Monthly: the five numbers above, the check against orders, the budget decision.
- Quarterly: review targets against current margin, review structure, decide what to shut down.
- Annually: incrementality for brand and remarketing, a full audit.
Metrics that do not belong in a management report
Not because they are useless — a specialist needs them. But no decision comes out of them:
- Impressions and reach. Not a result, a volume.
- Click-through rate. Easily raised with a discount in the headline that destroys the margin.
- Quality score. A diagnostic tool, not a goal.
- Impression share as a target. One hundred per cent loses money for most accounts.
- Average position and similar historical relics.
- Lead count without qualification. The most dangerous number in B2B reports.
The test we use: if I delete this chart, will anyone’s decision change? If not, the chart goes.
Comparing over time without fooling yourself
- Compare with the same period last year, not with last month, anywhere there is seasonality.
- Watch the number of working days. April and March differ more than you would think.
- Do not calculate short periods at low volumes. At fifty conversions a month a weekly comparison is noise.
- Mind the conversion window. Numbers for the last week will still grow. A report produced on the third of the month understates the end of the previous one.
How to write the report
A structure that works and fits on one page:
- One sentence of summary. “Month on plan, non-brand cost of sale 22 % against a 25 % target, held back by a stock outage in category X.”
- Five numbers with comparison and target.
- What we did and with what result. Including what did not work — otherwise that attempt gets repeated.
- What we will do next, with an expected result and a review date.
- What we need from you. The most commonly omitted and most important part: a budget decision, data from the CRM, a fix on the site.
Log the interventions that did not work and why you reverted them. A report containing only the good news is not taken seriously — and more importantly, nobody learns anything from it.
What to take away
A report is not evidence of work done, it is a decision-making tool. Five numbers, one sentence of summary, a list of what was done and a list of what will be. When no decision follows from a report, you do not need a better chart — you need different numbers.
Common questions
Which PPC metrics should I report to management?
Five numbers are enough: non-brand return compared with break-even from your margin, the share of brand in the overall result, cost per acquired customer against margin on a deal, marginal cost per conversion as the basis for budget decisions, and the ratio of conversions to real orders as a check on data credibility. For online shops it pays to add the share of the range that received at least one impression in the month. Impressions, reach, click-through rate and quality score do not belong in a management report.
How often should I check PPC campaigns?
Daily only through automated alerts on anomalies such as a jump in spend or a drop in conversions — that is not for a person to read. Weekly, go through the search terms report for the most expensive campaigns and check budget status. Monthly, evaluate the key numbers and decide about budget. Quarterly, review targets against current margin and the account structure. Once a year, verify incrementality for brand and remarketing and run a full audit.
Why are click-through rate and quality score not worth tracking as headline metrics?
Because no business decision comes out of them. Click-through rate is easily raised by putting a discount in the headline that destroys the margin, so a higher number does not have to mean a better result. Quality score is a diagnostic that hints where to look for a relevance problem, but it is not a goal in itself. Both belong in the hands of a specialist working in the account, not in a management report that decides about money.