Meghan Gendelman, Canva's B2B CMO, Is Right That Brand Is Now Infrastructure. The Harder Build Comes Next.

A new genre of executive essay has taken over the feed, and most of it blurs together: a nod to AI changing everything, a tidy reframe, a closing Question Every Leader Should Be Asking. Meghan Gendelman's entry stands out from that pile. It is specific, it is honest about the problem, and it names something most marketing teams have been feeling for a year without quite putting words to it.

Writing under the headline "The brand infrastructure question every CMO should be asking", Canva's B2B CMO argues that now that AI lets every team ship content at speed, staying on brand can no longer depend on someone policing the output after the fact. It has to live in the infrastructure. Locked colors, locked type, structured templates, defined edit zones, a system clear enough that neither a rushed junior employee nor a confident language model can wander off brand. She is right, and the results are on her side.

She is right, and the wins are real

The diagnosis holds up completely. AI dropped the cost of producing competent content close to zero. Anyone in the building can now generate a deck, a one-pager or a campaign brief in the time it takes to write the prompt. A brand "system" that lived as a three-year-old PDF in a shared drive was already losing, and against this kind of volume it barely registers. Governance belongs inside the tools now rather than bolted on at review, and that shift is overdue.

The proof points are strong. FedEx rolled Canva Enterprise out to 1,400 teams across 45 countries and watched brand-review requests fall 77 percent in three months. Docusign ran a full global rebrand across thousands of assets in four months. Stripe scaled content output twentyfold across dozens of markets while keeping the brand intact. Her 500-plus approved templates that make "off-brand creation the harder option" are exactly the kind of leverage most brand teams are missing.

All of that is foundation, and it is worth building on rather than just applauding, because the foundation solves one risk cleanly and leaves a second one open.

The risk the infrastructure leaves open

Her frame is tuned to divergence: chaos, drift, the off-brand deck with the almost-right blue. That problem is real, it produces visible panic, and her system genuinely solves it. Sitting right next to it is a second risk the frame does not reach, running in the opposite direction, toward convergence. Sameness. On-brand and completely forgettable.

The evidence here is some of the most replicated work in the early study of generative AI. In a 2024 study in Science Advances, researchers at UCL and the University of Exeter handed some writers AI-generated story ideas and left others to work alone. The AI-assisted stories scored higher on creativity and quality, with the biggest lift going to the weaker writers. Across the whole group, though, those stories landed measurably closer to one another than the human-only ones did. The authors called it a social dilemma: each writer comes out ahead while the collective pool of novel ideas quietly shrinks. A 2026 meta-analysis covering 19 studies found the same homogenizing effect, small but statistically sturdy, and present in real-world creative work rather than just the lab.

The mechanism is simply what a model does. It pulls its output toward the statistical center of its training data, which is the category average, the thing everyone has already seen a hundred times. Point a thousand marketing teams at similar models running off similar template libraries and they do not fan out. They crowd into the same competent middle. Researchers call it algorithmic monoculture, and the term fits: a whole category drawing on the same source and converging on the same look.

Gendelman's own company makes this case better than any outside critic could. In Canva's 2026 "State of Marketing and AI" report, 70 percent of consumers said AI-generated ads are "missing their soul." Sixty-nine percent said they expect the future of advertising to be "the same AI-generated slop." Mentions of "AI slop" jumped ninefold, and Canva's writers named the moment "volume without vision." What consumers keep flagging, by a wide margin, has little to do with brands looking inconsistent. They are reacting to everything looking the same.

Consistency and distinctiveness are not the same job

Gendelman reads the "missing soul" finding as a clarity problem, a case where, in her words, "the ingredients were right, but the clarity wasn't there." Clarity genuinely helps, and she is right to push for it. It is the start of the answer rather than the whole of it. Clarity makes a brand correct. Distinctiveness makes it noticed. Those run on different axes, and a brand can score full marks on one while flatlining on the other.

The science of brand growth draws the line cleanly. Byron Sharp and the Ehrenberg-Bass Institute spent decades showing that brands grow through mental availability, the ease of coming to mind at the moment of purchase, and that mental availability rests on distinctive brand assets: colors, shapes, characters and phrases that are unmistakably yours. Sharp's rule is to be consistent and distinctive together, repeating the things that make a brand impossible to confuse with anyone else. A locked template enforces consistency of format, which matters enormously. It does not, by itself, protect distinctiveness, and regression to the mean goes after distinctiveness first.

This is the part worth adding to her argument rather than arguing against it. Her infrastructure is a distinctiveness-preserving machine only when a human first decides, deliberately, what the distinctive thing is. Left on autopilot, it preserves format beautifully and lets the more valuable asset quietly erode.

What to build on top

If brand-as-infrastructure is the operating system, distinctiveness is the application it exists to run. Four practical ways to build that layer, all of which sit comfortably on top of the system Gendelman describes:

Split the work in two and treat the halves differently. The commodity layer, meaning internal decks, localized variants, signage, the hundredth social tile, is where consistency at speed is pure upside, and where her templates and AI should run at full output. The distinctive layer, meaning your voice, your hero work, the category-breaking swings, is where the goal is to be unmistakable rather than efficient, and where a template is a starting point you are expected to break.

Spend the time the system gives back. Automation frees real capacity, and the strongest teams reinvest it in the work models cannot do. Stripe is the example here: with production handled in-house, the team funded on-location brand shoots, motion work and campaigns beyond templates, plus copy angles no outside writer would land. The efficiency paid for distinctiveness instead of replacing it.

Measure distinctiveness, not just compliance. Brand-review counts tell you whether the work is correct. They tell you nothing about whether it is recognizable. Jenni Romaniuk's work at Ehrenberg-Bass offers a ready metric: track the Fame and Uniqueness of your distinctive assets, and you can watch distinctiveness rise or fall on a dashboard the same way you watch review volume.

Write the brief that names what to avoid. A model reaches for the category defaults first. A single page listing the visual and verbal clichés your category is saturated with, the moves your brief explicitly rules out, does more for distinctiveness than another rule about logo spacing.

Canva's research points the same way. When the company asked marketers what AI can never replicate, the top answers had nothing to do with speed or consistency. Marketers named empathy, "human imperfection that sparks originality," and brand intuition. Those are the raw materials of the distinctive layer, and they are exactly what a system is built to standardize away. The system and the soul need different custodians.

Build the system, then add the nerve

Gendelman describes the creative team's new job as "designing the system that makes good output inevitable," which is a sharp way to put it. A system can make correct output inevitable. Making it distinctive is the part that still needs a person deciding, on purpose, what distinctive means and then defending it.

So build the infrastructure she describes. It is the right foundation, the wins are real, and brand teams that skip it will struggle. Then add the part no template can supply: the deliberate, slightly risky, unmistakable choices a model would smooth into the category average. Consistency is becoming free, for you and for every competitor at once, which makes it the floor rather than the edge. The edge is everything the system was built to make safe, kept deliberately a little dangerous. That part still takes nerve, and it always did.