
Strategy & Craft
Splitting an ad budget: the 70-20-10 rule
Seventy per cent on what works, twenty on scaling, ten on experiments. The rule is old, simple, and violated in almost every account.
Strategy & Craft
A forty-page monthly report is a way of not being read. One page, five numbers, and three sentences that say what changes.

Marketing reports fail in a specific way: they document activity thoroughly and say nothing about what should happen next. They are read once, filed, and never referenced.
Line 1: the headline. One sentence stating whether the month was good, bad or flat, and why.
Block 2: the five numbers. Enquiries, cost per enquiry, close rate, cost per customer, revenue from advertising. Each with the same month last year beside it.
Block 3: what changed. Two or three bullets naming specific causes: a campaign launched, a competitor's price change, a channel that broke.
Block 4: what we recommend. Two or three actions with owners and dates. This is the only part that justifies the report existing.
Appendix: everything else, for whoever wants it.
Month-on-month comparison in most businesses measures seasonality, not performance. A roofing firm's October is always better than its February, and reporting that as improvement is meaningless.
Year-on-year for the same month removes the seasonal effect and reveals the trajectory. Where the previous year is unavailable, use a rolling three-month average.
The most useful discipline in reporting is naming problems before anyone asks. It costs nothing, it builds credibility, and it prevents the conversation in which a director finds a number the report avoided.
The format: what went wrong, what caused it, what we are doing, and by when we will know whether that worked.
| Common inclusion | Why to remove |
|---|---|
| Screenshots of every advert | nobody reads them |
| Platform engagement scores | not comparable, not actionable |
| A list of tasks completed | describes effort, not result |
| Follower counts | correlate with nothing |
| Fourteen charts of the same data | pads the document |
If the report is presented rather than sent, the rule tightens further: three slides. Where we are, what changed, what we propose. Everything else answers questions if they arise.
A presented report that runs longer than fifteen minutes has stopped being a decision meeting and become a briefing.
Once a quarter a longer document is worth producing, because the questions are different: what have we learned, which assumptions turned out wrong, and what should we stop doing.
That document is read, because it asks questions people care about rather than reciting numbers they have already seen.
One page plus an appendix nobody is required to read.
Whoever will be asked to act on it. Reports written by the agency for the agency's benefit describe activity, not results.
More from Strategy & Craft

Strategy & Craft
Seventy per cent on what works, twenty on scaling, ten on experiments. The rule is old, simple, and violated in almost every account.

Strategy & Craft
Percentage-of-revenue rules are widely quoted and mostly wrong. The answer comes from your own numbers in four steps.

Strategy & Craft
A bad brief produces bad work at full price, and the client pays twice. Eight questions, two pages, three hours of your time.