Agencies and media buyers lose a predictable number of hours every week to the same task: assembling numbers from three platforms into one document so a client can look at it.
It is not difficult work. It is repetitive, it is deadline-bound, and it is the first thing that gets rushed when a campaign needs attention. Which means the report that represents your whole relationship is often built in the worst hour of your week.
Why it keeps happening
Nobody agreed what the numbers mean. The client thinks "leads" includes form fills. You think it excludes the spam ones. Both are reasonable, and the disagreement surfaces in the report, every month.
The data lives in three places with three definitions. Platform A counts a conversion on click, platform B on view, your CRM counts it when a human qualifies it. Adding them together produces a number that is precise and wrong.
The narrative is written from scratch. The only genuinely human part of the report is "why did this change", and it is the part done last, at speed, from memory.
The fix, in three moves
1. Agree one metric definition per client, in writing. Not a dashboard, a sentence. "A lead is a form submission with a valid phone number, deduplicated by phone, attributed to the campaign that produced the first touch." Signed off once, referenced forever. This single sentence removes most report arguments before they start.
2. Generate the numbers, write the narrative. The table is computed by a job that pulls the platforms on a schedule. No human retypes a number, ever. The only person-authored part is two or three sentences explaining the movement.
3. Send the same shape every time. Same sections, same order, same date. A client who knows where to look stops asking where to look, and the questions become about the work instead of the document.
The report that earns its slot
Four sections, in this order:
The number that matters. One metric, this period against last, with the target beside it. Not a wall of metrics. The client checks one thing first; put it first.
What changed and why. Two or three sentences. This is the only part that requires judgement, and it is the part clients actually read.
What we did. The work delivered. Not a task list, the changes made and the reasoning.
What is next. The decision the client needs to make, if any. A report with no asks is a report that gets skimmed.
Anything else belongs in an appendix nobody opens.
What to automate, and the one thing not to
Automate: the platform pulls, the joins, the period comparisons, the delivery on a schedule, and the archive of every past report in one place.
Do not automate: the explanation. A generated narrative reads as generated, and a client who suspects the explanation is a template will discount the numbers too. The whole value of the report is that a human looked at it and had an opinion.
How to tell it is working
- The report exists before the call is scheduled, not because of it.
- The numbers in it were never touched by a person.
- Two different clients could swap reports and still understand the structure.
- The only time you spend on reporting is the three minutes of writing the explanation.
If the report takes the morning, you are paying for it twice: once in the hours and again in the quality of the explanation you had no time to write.