The task
A monthly paid media report for a Google Ads and Meta Ads client has an executive summary, performance by channel and by campaign, anomalies, and a plan for the next month. The consultant describes the tables as the most automatable part, which Looker Studio has solved for years. The slow part is the writing: what happened, why, and what to do next. According to the source, a full report takes between 4 and 8 hours when done properly.
What they did
The consultant keeps a fixed Looker Studio template per client or account type, with monthly KPIs, comparisons against the prior month and the same month a year earlier, per-campaign detail, and highlighted anomalies. The summary table is exported as a CSV file, because Looker Studio has no direct Claude connection. Claude then receives the current and prior month's data, the client's sector, products, objective, and spend, and any events such as a launch or budget change. The prompt asks for a short executive summary, a short comment per campaign, and a list of anomalies. It forbids invented reasons and tells Claude to write "needs review" where the data does not explain a variance. The source names Claude Sonnet 4.6 as the model. In the worked example, an anonymized fashion ecommerce client, the consultant checks that the dashboard refreshed and that conversions match Google Ads and Meta Ads, exports the table, notes the month's events, generates the draft, edits it, and builds the final PDF.
What the source reports
According to the source, the monthly flow for the example client takes about 75 minutes, compared with 4 to 5 hours a month before, a reduction of about 75 percent. The source says editing the draft takes 20 to 30 minutes instead of 2 to 3 hours of writing from scratch, and that quality stayed comparable according to client feedback. The one-time setup for that client took about 3 hours, according to the source.
Where the approval sits
The consultant stays in the loop at several points. Before the draft, the consultant verifies the dashboard refresh and the conversion figures and adds context for the month's events. After the draft, the consultant validates the figures, adjusts the tone, refines the next-month actions, and removes wrong inferences. The source says the plan for next month always needs the consultant's validation. The consultant also lays out the final PDF.
Worth knowing
This is one freelance consultant's account on a personal site, and the figures are self-reported from a single anonymized client. The consultant also sells services. The source lists cases where it advises against this flow: accounts with fewer than 3 months on the books, months with structural changes such as a tracking migration, reports for non-marketing executives, and clients who prefer fully human reports. It also warns that automated narratives built on unclean data produce errors that look professional.
Source: Personal blog · Jun 2026
Drafted with AI assistance and checked automatically. A person is responsible for the collection. How these entries are made
Prompt for your agent
Draft the monthly ad spend commentary
Turn the campaign figures for one month into a commentary that explains each change and flags any gaps, ready for your review.
More in this category
- Client reporting
A performance agency lets AI draft weekly client summaries for account manager review
A performance marketing agency has an AI model write plain-language summaries of the weekly changes in each client's campaigns. Each summary is saved as a draft, and the account manager reviews, adjusts, and sends it.
Read the entry - Client reporting
An agency has GA4 client reports drafted overnight for account managers to edit
An agency with 11 clients has AI agents draft weekly Google Analytics 4 reports on Wednesday night. Each account manager reviews and edits the drafts on Thursday before they go to the clients.
Read the entry
Ninety minutes live on your tools. One automation running before we hang up.