Monthly reporting for seven restaurant brands, from two weeks to a few days, with the time put back into growth

Problem

The monthly leadership report had been built in Excel by the two people before me. Pulling the data, sorting it and tidying it took at least two weeks every month, around the day job of product uploads and weekly meetings. Even then, leadership found it hard to read and harder to act on. Underneath the report, the advertising couldn't be measured properly. The accounts ran on generic templates. Google Tag Manager wasn't set up correctly, so conversions weren't tracked, and there was no reliable history to set ad spend against sales. The websites had their own quiet problem. Events stayed listed after they happened, promos outlived the promo, and Google showed different opening hours from the site. Some of these issues had sat on the sites for a long time.

Constraints

I run the reporting about 90% of the time. Leadership and the other departments only read it, so it has to make sense to someone who never opens a dashboard. The tools sit on the group's marketing account, not mine, so they have to keep working when I'm not there. AI can lay out a report but it can't be the source of a number: every figure has to trace back to the platform it came from. Budget mattered too. The report runs on four connectors only, GA4, Google Ads, Meta and Google Business Profile. Email, TikTok and the online shops can plug in later, once they earn the cost. And with tracking only fixed this year, there was no clean history to benchmark against.

Process

I worked on the reporting and the ad setup at the same time, because each freed up time for the other. For the report, Windsor.ai pulls those four sources for every brand into one place. Claude turns that into the same layout for each brand, written to be read rather than decoded: short, consistent, and each section ending in what to do next. I check the figures against the platforms before anything goes out. The only real wait now is for the connectors to pull and the platforms to update. For the ads, I replaced the templates and rebuilt Google Tag Manager and conversion tracking, so spend can finally be measured against sales. Claude audited the setup, researched the options and recommended a fix. I reviewed it and approved it, and Claude made the change while I got on with the rest of the week. In September 2026 I added the Site Truth Monitor. On Monday and Thursday mornings it checks seven brand sites and nine Google listings against what they should say, and gives the team one ranked list of what to fix, with the evidence for each item.

Result

A report that used to eat half the month is now done in under a week, and that's my estimate rather than a tracked figure. It was the first thing leadership noticed. The time it freed up goes into the work the role was meant for: SEO, AEO and GEO, website fixes with the developers on issues that had sat for too long, and new projects to grow each brand's traffic and, from there, bookings and sales. The monitor's first full run, on 25 September 2026, flagged 35 issues across seven brands, three of them urgent: an event the next day that couldn't be booked, nine of ten event pages on one site returning a server error, and a shop Google showed as closed on a day it was open. It's early on the advertising side. I've been in the role about six months, the tracking is now in place, and the next step is to scale spend now that it can be measured.

What I'd do differently next time

I'd have started using AI for building much earlier. For years I used it for planning, copywriting and research. Tools like Claude and Codex turned out to be far more useful for the slow, repetitive parts of the job, and that's where the real time came back.