Automated reporting
Most reports take more time to make than to read. Exporting numbers from three tools, pasting them into a spreadsheet, updating charts, writing a commentary and sending the whole thing out. Every week or month, again. I build reports that build themselves: at a fixed time the data is retrieved, processed and turned into a report with numbers and a readable commentary. You open it and can get straight to work on what it says.
Example: weekly report on paid advertising performance
An automation I built that runs every week: a report on the performance of paid ads in the previous week. Every Monday morning the workflow pulls the figures from the ad platforms, compares them with the week before and with the same week last year, and calculates per campaign what changed in spend, clicks, conversions and cost per conversion.
Then it goes beyond a table. An AI step looks at the notable differences and writes a short commentary: which campaigns became more expensive and why that probably is, where budget could be put to better use, what needs attention this week. The result is a report that reads like the summary from a colleague who has already looked at the numbers. It lands in the inbox before the first meeting of the week starts.
How such a report is built
- Retrieve. At a fixed time the workflow pulls data from the sources that matter: ad platforms, Google Analytics, Search Console, your CRM, your accounting software, a web shop or a spreadsheet. Via APIs, so no exporting.
- Process. The data is cleaned, merged and compared with the previous period. Deviations larger than a set threshold are flagged.
- Interpret. An AI step writes a short commentary in plain language based on the figures and the deviations. You decide upfront on the tone, the length and where the emphasis should be.
- Deliver. The report goes out as an email, PDF, Google Doc, Slack message or dashboard update to whoever needs to read it. Including your clients, if you run an agency.
The setup differs per report, but the building blocks are the same. That is why a second report is usually built faster than the first.
Which reports you can automate
Marketing and SEO
- Ad performance per week or month
- Organic traffic, rankings and conversions
- Client reports for agencies
Sales and clients
- Pipeline and new leads from the CRM
- Quotes sent, accepted, expired
- Client satisfaction and reviews
Finance and operations
- Revenue, outstanding invoices and cash flow
- Hours and project progress
- Stock and orders from your web shop
What a report does and does not do for you
An automated report collects and interprets, but does not decide. The AI commentary is a first reading of the numbers, not a replacement for your judgement. That is why the workflow also shows what the commentary is based on and flags uncertainties. And a report is never better than the data that goes into it: if conversion tracking is wrong, the report is wrong too. During setup I therefore first check whether the sources are reliable.
More about how I handle data on the page about data retrieval and filtering.
What it costs and what is included
A report from one or two sources with an AI commentary usually comes in between 1,000 and 2,500 euros one-off. Reports from more sources, with multiple recipients or client-specific variants range from 2,500 to 4,000 euros. After that you pay a fixed monthly fee for maintenance: platforms like Google and Meta change their APIs regularly, tracking settings change, and the questions you ask of a report shift the longer you work with it. Maintenance also includes refining the commentary based on your feedback. See pricing and maintenance.
Frequently asked questions
Is this the same as a dashboard in Looker Studio or Power BI?
No, they complement each other. A dashboard shows numbers at the moment you look. An automated report comes to you at a fixed time, with a commentary on what has changed. Many clients use both: the dashboard to dig deeper, the report to know where to look.
Can the report be in my own layout or house style?
Yes. The report can be formatted as an email, PDF or Google Doc with your logo, colours and layout. For agencies sending client reports that is usually a requirement.
How reliable is the AI commentary?
The commentary is based on the figures the workflow itself retrieved and calculated; the AI does not invent numbers. What the AI does do is interpret, and there it can draw a wrong conclusion. That is why I show what a remark is based on, set the tone and level of certainty, and refine the instructions based on your feedback in the first weeks.
Which sources can you pull data from?
Almost anything with an API: Google Ads, Meta Ads, LinkedIn Ads, Google Analytics, Search Console, Ahrefs, HubSpot, Pipedrive, Exact, Moneybird, Shopify, WooCommerce, Google Sheets and many more. For sources without an API I look at exports by email or file exchange.
Which report do you still make by hand?
Send me an example of your current report and the sources the numbers come from. I will show you what the automated version could look like. Or see the overview of all automations first.