Automated client reporting for a digital agency
A reporting platform that aggregates web analytics, search, paid and rank-tracking data into branded monthly client reports — now with AI-written commentary grounded in the real numbers.
The AI commentary is fed the exact figures rendered on each report, section by section — so it explains what actually happened, rather than making things up. Pragmatic AI, in the one place it saves real hours.
This platform is the reporting engine for a digital marketing agency. It pulls data from web analytics, search console, paid search and rank tracking, and turns it into branded monthly client reports — consistent, scheduled, and maintained by a very small team.
We built it around a builder-driven pipeline: each report block fetches from a data-source service, computes month-on-month and year-on-year comparisons, renders tables and charts, and caches its output. Scheduled jobs generate reports and deliver them, and a provider-agnostic AI layer now drafts the written commentary from the exact numbers on the page.
It turns a week of copy-paste into a scheduled job, and the AI drafts the commentary from our real numbers. Our reports have never been more consistent.
A multi-stage approach.
Discovery & builder design
We designed a builder-driven engine where each report block is a self-contained unit — fetch, compare, render, cache — so new report types could be added without touching the rest.
Multi-source aggregation
We integrated web analytics, search console, paid search and rank tracking behind one pipeline, with per-client branding, colours and configurable rules.
Surviving GA4
When Google retired Universal Analytics, we rebuilt the entire data layer on GA4 — swapping the engine while the reports kept running.
AI commentary & AI-search
We added a provider-agnostic AI layer that drafts each section's commentary from the rendered numbers, and reporting on visibility in Google's AI Overviews.
What changed.
Every source, one report
Web analytics, search, paid and rank-tracking data flow into a single branded report, with month-on-month and year-on-year comparisons computed automatically.
AI that writes the analysis
A provider-agnostic AI layer drafts each section's commentary from the exact figures on the page — grounded, not generic — with the model swappable behind the scenes.
It survived GA4
When Google forced everyone off Universal Analytics, we rebuilt the data layer on GA4 without the reports ever stopping.
Reporting on the AI-search era
The platform now measures how clients appear in Google's AI Overviews — share of voice, AI search volume and traffic value.
Reports eating your team's week?
If your team is assembling the same reports by hand every month, we'll help you automate the whole pipeline — and add AI where it genuinely helps.