The off-brand problem is 3x bigger than leaders think. Here's how to fix it with AI.

Brands still matter in an AI world. In fact, they matter more.

That's the headline from Adobe's new research, Brand Intelligence for an AI-Native World. Adobe and research firm Incisiv surveyed over 600 brand, marketing and creative professionals at large enterprises, plus more than 1,400 consumers and business buyers. 87% of enterprise marketing and creative leaders expect trusted brands to become more important as AI use grows.

The report is worth reading in full. But one finding stood out for us, because it's the problem Brandkit was built to solve: the brand knowledge exists, it just isn't available when and where people (and now machines) need it.

Consistency builds trust. Inconsistency costs it

The research draws a straight line from consistency to trust, and from trust to loyalty:

  • 83% of consumers say a brand that consistently looks, sounds and shows up the same way drives trust.
  • 88% say an inconsistent brand expression decreases trust.
  • 90% will recommend a brand they trust, and 82% will stay with it even after a bad experience.
  • 75% of enterprise respondents say their organisation has suffered real business consequences from off-brand content.

One retail VP in the study put it simply: consistency is "the only thing a customer can check quickly."

This applies to every brand, whatever its size. Your brand is expressed by many people outside your brand team: staff in other departments, agencies, resellers, franchisees, partners and media. Most of them will never speak to your brand team. Consistency depends on how easy you make it for them. Tourism boards and place brands know this well, but the same is true for any brand.

static-file-2800

(Fig: Is this caravan featuring McDonalds and Airbnb branding on-brand?)

Leaders reckon 9%. The people doing the work think 27% is off-brand!

Adobe found that 74% of content approvers admit they have signed off on off-brand content.

There's also a gap in how big people think the problem is. Senior leaders estimate 9% of their public-facing content is off-brand. The people doing the work put it at 27%, three times higher.

Why? The report points to three root causes.

1. Guidelines are necessary, but not enough.

Master brand guidelines average 140 pages (yes really).

And:

  • 79% of practitioners say guidelines are already too long to be useful
  • 71% say there are too many separate brand documents
  • 67% find it hard to know if they have the current version?
  • **58% **say different guideline documents contradict each other

2. The real rules aren't written down.

70% of practitioners rely on unwritten rules and institutional knowledge to decide what's on brand. 89% say that when guidelines don't cover a situation, the answer lives with a few people or one team. 63% say it usually takes more than a week to get a definitive answer.

3. Review is manual, and the bottleneck is people.

The most common brand control is manual review by senior leaders (83%). Only 10% use automated or AI-based brand-checking tools, and 75% of those who have tried them report a mixed or negative experience. The tools only check what's written down as a rule, so they inherit the same gaps as the guidelines.

In short: a long PDF, a few busy experts, and a lot of guesswork in between.

What the research suggests and why it matters for everyone

The answer is brand intelligence: your brand knowledge, organised so that people and AI can find it, understand it and apply it. Brand managers have always held this knowledge. The challenge now is making it available beyond the brand team.

Adobe's report makes four recommendations that are useful for any brand, at any size:

  • Make guidelines useful to AI. Write in clear, definitive language. State the don'ts alongside the dos, because AI can't infer boundaries the way a designer can.
  • Capture unwritten rules and new decisions. Record the judgement calls, exceptions and answers that currently live in people's heads.
  • Move beyond static rules. Give systems enough context to reason about situations the rulebook doesn't cover.
  • Provide access wherever the work happens. Surface brand knowledge where designers, reviewers and external partners already work, and make it consumable by AI agents through open protocols like MCP.

That last point is the big one. Brand governance used to be about people. Now it's about people and machines.

Who is reading your brand guidelines now?

Your guidelines have three audiences today:

  1. Your team and collaborators. Staff, agencies, partners, resellers and journalists who need answers quickly.
  2. Public AI models. ChatGPT, Claude, Gemini and Perplexity form their view of your brand from what they can find and read on the open web.
  3. AI agents and tools. The assistants your team and partners use to write copy, build presentations and create content.

A 140-page PDF serves none of these well. People don't read it. AI models often can't parse it. And agents can't reach it.

How Brandkit puts brand intelligence to work

Here's how we approach each of Adobe's recommendations in Brandkit.

1. Digital brand guidelines, written for people and AI

In Brandkit, Digital Brand Guidelines are web pages, not PDFs. They're built as Knowledge Assets that sit alongside your logos, photos and videos in your Brand Portal.

That solves several of the problems in the research at once:

  • One current version. Update a page and everyone sees the change. No hunting for the latest PDF.
  • Short, linked sections. Break a 140-page document into focused pages for logo use, tone of voice, colour, photography and so on.
  • Structured for machines. Clean, plain-text pages are easy for AI to read and cite.

You can also add to or edit your digital brand guidelines in real time, just open, edit and save.

Adobe's advice on clear, definitive language applies here too. "Use as appropriate" helps nobody. "Use the reversed logo on dark backgrounds only" helps everyone, human or machine.

2. A BrandBot that answers questions in seconds, not weeks

A BrandBot is an AI assistant that learns your brand from the guidelines, FAQs and assets you publish in Brandkit. You can add it to your Brand Portal, your CMS pages or embed in your website(s).

Instead of waiting a week for an answer, a partner can ask:

  • "Can I use the logo on a co-branded event poster?"
  • "What's our tone of voice for social media?"
  • "Which product name do we use in customer-facing copy?"

The BrandBot answers from your published brand knowledge. It works for staff, agencies and partners, at any hour, in any time zone.

3. Capture the unwritten rules as FAQs

The research shows that most brand decisions rely on knowledge that lives with a few people. The fix is simple, if not always easy: write it down.

Every time your brand team answers a question that the guidelines don't cover, add it as an FAQ in Brandkit. Admins can add and edit FAQs at any time, and the BrandBot uses them straight away. You can also review BrandBot conversations to spot the questions people keep asking, and fill those gaps.

Over time, your FAQs become a record of your brand's judgement calls. That's the "decision trace" Adobe describes, without a large enterprise system.

4. Make your brand readable by public AI, through the open web

Large language models learn from the open web, including datasets such as Common Crawl. AI search tools also fetch live pages to answer questions. If your brand story, facts and guidelines are locked in PDFs or behind logins, these tools will fill the gaps from other sources.

Brandkit helps here in two ways:

  • Public Brand Portal pages can be indexed by search engines and AI crawlers, so the information you choose to share is the information AI finds.
  • An automatically generated llms.txt file gives AI tools a clean, plain-text index of your published knowledge. You can see ours at brandkit.com/llms.txt and even more specific index for Brand Guidelines guideline.txt and FAQs faqs.txt

You decide what's public. Keep sensitive material private, and publish what you want the world, and its AI models, to know.

5. Connect AI agents directly through MCP

The Model Context Protocol (MCP) is an open standard that lets AI tools like Claude connect to other systems. Adobe names it as a key way to make brand intelligence available to the AI tools your teams use.

Brandkit's MCP server lets authorised AI assistants search your Brandkit account and pull in your approved guidelines, knowledge and assets. So when someone on your team asks their AI assistant to draft a campaign brief, it can work from your actual brand, not a guess.

A simple checklist to get started

You don't need to be an enterprise to do this. Start here:

  1. Move your guidelines out of PDF and into short atomised text based assets.
  2. Rewrite vague rules as clear dos and don'ts.
  3. List the ten questions your brand team answers most often, and publish them as FAQs.
  4. Add a BrandBot to your Brand Portal so partners can self-serve.
  5. Decide what's public, and make sure AI crawlers can read it.
  6. Connect your AI tools to your brand knowledge through MCP.

The bottom line

Adobe's research confirms what brand managers have long known: consistency builds trust, and trust builds loyalty. What's changed is who needs to understand your brand. It's no longer just your team and partners. It's every AI model and agent that writes, summarises or recommends on your behalf.

The brands that win will be the ones that make their knowledge easy to find, easy to read and easy to use, for people and machines alike.

That's what Brandkit is for. A home for your brand, for people and AI.

Book a demo


Frequently asked questions

What is brand intelligence?
Brand intelligence is your brand knowledge (who you are, what you stand for, and how you look and sound) organised so that people and AI can find it, understand it and apply it. It includes your brand guidelines, the answers to common questions, and the judgement calls your brand team makes every day.

Why do brand guidelines fail?
According to Adobe's research, master brand guidelines are too long with an average 140 pages, and 79% of practitioners say they're too long to be useful. Teams also struggle with too many documents, out-of-date versions and contradictions.

What is a BrandBot?
A BrandBot is a custom AI assistant, powered by Brandkit, that answers questions about your brand using the guidelines, FAQs and assets you publish. It can be added to a Brand Portal, CMS pages or a embedded in a corporate website.

How do AI models learn about my brand?
Large language models learn from public web content, including datasets like Common Crawl, and AI search tools read live web pages. Publishing clear, public brand information and an llms.txt file helps AI describe your brand accurately.

What is MCP and why does it matter for brands?
MCP (Model Context Protocol) is an open standard that connects AI assistants to other systems. It lets AI tools draw on your approved brand knowledge directly, rather than guessing.


Source: Adobe, Brand Intelligence for an AI-Native World (2026). Research conducted with Incisiv; surveys fielded in North America, July–August 2026. Statistics quoted are from this report.

Happy branding :)

The off-brand problem is 3x bigger than leaders think. Here's how to fix it with AI.

Adobe research finds employees estimate off-brand content is three times more common than leaders think. Discover why guidelines fall short—and how AI-ready brand knowledge, BrandBots and MCP can help teams build consistency and trust.

Asset type post
ID #933756
Word count 1918 words

Licence

Licence: Worldwide Paid and Unpaid Available to anyone for royalty free use in paid and unpaid media worldwide, provided Brandkit benefits from such use, and David Vaassen is credited (optional).
Expiry: No expiry date
Release date:
Added at:
Updated at:

Tags

Loading