Every agency I’ve worked with has the spreadsheet. One row per client: hex codes, font names, a link to the latest guidelines, who signs off, and a notes column that says things like “NEVER use the old tagline, see email from March”. It starts tidy. By the twelfth client it’s the most important file in the building and nobody trusts it.
I was on the client side of that relationship for years, sending agencies the PDF and hoping. Here’s how I’d run a multi-brand studio now.
One workspace, every brand
In Brandbook, a workspace holds as many brands as you need. Each one has its own logo, palette, type, voice, imagery and rules, its own library and its own public brand book. Switching between them is one click in the brand switcher, and everything on screen changes with it: the rules Create follows, the checks, the library.
Adding a client is the same as starting your own brand: paste their website, wait a minute, review what the scan found. If they have an old PDF, add it as a source and settle the disagreements. You’ll have a working system before the kickoff call ends.
See every brand at once
The agency view puts every brand on one screen, with its health score and how it moved since the last scan. A brand drifting downwards is a conversation to have with the client before they notice it themselves. A brand holding steady is one you can leave alone this week.
That health score comes from checking the client’s own live site against their own rules, so it means something specific. It isn’t a vibe. It’s off-palette colours, voice slips and weak contrast, counted.
Your tools know which brand they’re in
The quiet cost of twelve brands is context switching. Your designers remember which green is which. Your AI tools don’t. They happily write a Northbound caption in Pulp & Co.’s voice if you paste the wrong guide.
One Brandbook API key covers every brand in the workspace. Connect it to Claude, Cursor or your own scripts once, and the agent can list the brands, pull the right one by name and check its own drafts against that client’s rules. The brand is part of the request, not part of your memory.
Approvals without the email chain
Every piece of work gets checked before it goes to the client, so the first round of feedback isn’t about the logo being too small. Anything that needs a person goes into the review queue on Check, with the score and findings attached. You can set a score above which work is approved automatically, so your senior people spend their time on the work that actually needs them.
Watch what goes out
Clients post things you never see. With Monitor, you can connect a client’s Instagram, Facebook, LinkedIn, TikTok or website and have new posts checked against their rules as they appear. When something drifts, you know first, and you can bring it to the client as a helpful note instead of a surprise.
A brand book you can hand over
Every brand gets a public brand book at its own address, with five themes to choose from and downloads for the logo, brand.md and tokens.json. It’s the thing you send the client’s other suppliers, instead of a PDF that’s out of date by the time it lands. When you change a rule, the brand book changes too, and its What’s changed page says what moved.
The spreadsheet was never the problem. Having to keep it true by hand was.
Moving over
You don’t have to move everyone at once. Start with the client whose brand drifts most, because that’s where you’ll feel the difference. Scan their site, fix what the scan got wrong, connect your tools, and run a month of work through checks. Then do the next one. By the twelfth, you’ll have stopped opening the spreadsheet, and someone can finally archive it.
