An Amazon PPC report template for agencies: the sections and metrics to include, weekly or monthly, and how to build it once in Claude for every client.
Every Amazon agency sends a PPC report, and most are produced the same way: someone downloads campaign data from the Ads console on the first working day of the month, pastes it into a spreadsheet, repairs the formulas that broke, and writes three paragraphs the client rarely reads past the first. Multiply that by twenty clients and reporting quietly turns into a part-time job.
This post gives you a template for the report itself, section by section, with the metrics that earn a place in it and the ones that do not. It then shows how to produce that report for every client from a single prompt in Claude over DataDoe data, and which delivery route suits which client: a written report, a spreadsheet link, Looker Studio or a Claude Data dashboard.
A client report answers three questions. Did advertising do its job this period? What did the agency change? What happens next? Anything that serves none of them belongs in an appendix or nowhere. Clients forgive a short report. They do not forgive one in which they cannot find the number they care about.
The report also needs a fixed point of comparison: the previous period, and the target agreed when the engagement started. If the account came to you through an audit, that audit is your baseline. Our Amazon PPC audit checklist ends with exactly that table.
Three to five numbers and one sentence. For most sellers that means ad spend, ad sales, ACoS, TACoS and total sales, each shown against the previous period and against target. The sentence names what drove the change. A client who reads nothing else still knows how the month went.
Weekly figures for the last 13 weeks, so seasonality and the effect of your changes are both visible. Put ad sales beside total sales: the client sees at a glance whether advertising moves the whole business or only its own numbers.
Sponsored Products, Sponsored Brands, Sponsored Display and, where the client runs it, Sponsored Television, each on its own row with spend, sales, ACoS, ROAS and share of spend. Put the attribution window in the table header, because it is not the same for every ad type (more on that under the traps below).
Not the whole campaign list. Pick the five campaigns with the largest change in spend or ACoS, in either direction, with one line on why. Eighty campaign rows tell a client you have data. Five with reasons tell them you understand their account.
Three short lists: converting queries you promoted to keywords, queries you added as negatives together with what they had been costing, and the top non-brand queries by sales. This is where the work becomes visible, and it answers the question many clients never say out loud: what does the agency actually do each week?
A dated list of what changed: bids, budgets, new and paused campaigns, placement adjustments, rules. DataDoe keeps the change history of Sponsored Products and Sponsored Brands campaigns, ad groups, ads and targets, and every Action run through DataDoe is logged with Amazon's response, so this section can be written from records instead of memory.
Two or three lines from outside advertising that explain the numbers: a product running low on stock, a lost featured offer, a price change, organic sessions falling. Ad results often move because of the listing, not the campaigns. Sessions and conversion come from Amazon's Business Reports, covered in our guide to Amazon Business Reports.
Three planned actions, each with the effect you expect. Next month's report opens by checking them.
ACoS and TACoS do different jobs, and clients confuse them. A two-line definition at the foot of the first page saves a call; ACoS vs TACoS has wording you can borrow.
Both, at different lengths. The weekly report is five lines in an email or a Slack message: spend, sales, ACoS against target, the biggest change of the week and anything that needs the client's decision. The monthly report is the full template above. A quarterly review adds the strategic questions: next quarter's budget, product launches, new marketplaces. Clients who get the weekly note read the monthly report more closely, because nothing in it comes as a surprise.
The template above is really a prompt. Write it once, and Claude produces the report from DataDoe data for any client in your organization.
Claude reads campaign, search term and placement data, the change history and, where Seller Central is connected, total sales and stock levels. Check the source numbers before anything goes out. For the weekly round across the portfolio, the same prompt works with "for every client Group" and a limit of one paragraph per client.
Best for the monthly narrative. Claude drafts it, an account manager checks it, and it goes out as a PDF or a document under your agency's name.
For clients who live in spreadsheets. A recurring Export runs on a schedule and offers a static link that Excel or Power BI can connect to directly, so the client's own workbook refreshes without anyone emailing files around. Recurring Exports work with non-PII sources, which covers the advertising tables this report uses.
For agencies with a reporting stack of their own. DataDoe shares the data as a read-only linked dataset in your Google Cloud project, and the Integrated option gives you Seller, Vendor and Ads data already cleaned and joined. On the Base plan there is no monthly integration fee: queries draw AI Tokens, and Google bills the compute. Looker Studio reads BigQuery natively, so one report design can serve every client and filter by account. Our post on Amazon data in BigQuery walks through the dataset.
For clients who prefer to click around rather than read. Claude Data builds a dashboard from a plain description and writes one query per chart, and anyone with access can open the query behind a number. The steps are in our guide to an Amazon dashboard in Claude Data, and the ways to connect the data are on the Claude Data page. Before sharing one with a client, read Anthropic's notes on what viewers see, and keep buyer personal data off the page.
One Base plan at $97 a month covers the agency: unlimited client connections, 10M data rows, 2,000 AI Tokens a month and one seat. Every additional seat costs $29 a month and brings 500 more AI Tokens. The part that grows with your portfolio is stored data, which you control table by table and with retention periods. There is a 14-day free trial, and an account manager for every customer. The full agency setup is on DataDoe for agencies, and what Claude can do inside each ad account is on Amazon Ads in Claude.
A headline with three to five numbers, spend and sales over time, results by ad type, the campaigns that moved, search term work, a list of changes, business context and the plan for next period. Keep everything else in an appendix.
A short weekly note and a full monthly report suit most clients, with a quarterly review for budget and strategy.
Yes. Connect every client account to one DataDoe organization, group them by client, and run the same template in Claude, in Looker Studio over BigQuery, or through recurring Exports.
Show both. ACoS judges the campaigns, TACoS judges what advertising does for the business, and a report that shows only one of them hides half the story.
Yes, in several ways: a recurring Export link in their own spreadsheet, a Looker Studio report over BigQuery, or a DataDoe seat whose access covers only their Group.