AI made your marketers faster. Why isn’t your marketing better?

Generative AI is that new hire. It's faster and better read than anyone on your payroll, and it never gets past its first day. Speed, it turns out, was never what stood between your team and better work.

That's why so much of what it produces is competent and useless at once. Ask for a campaign brief and you get "Unlock efficiency at scale," a line that could sit on the homepage of any company selling software and therefore belongs to none of them. So you rewrite it, explaining your priorities, your customers and last year's failed launch all over again. Next week, you'll do it again.

A CMO at a $8.9 billion company told us recently that each person on her team is faster than ever, and yet she can't point to a single thing they make together that's actually better. She has plenty of company. In September of 2026, Bain found that only 6% of marketing leaders using AI report significant performance impact. The usual explanation is that it's early and teams need more training or a better tool. I'd argue the AI is doing exactly what we asked. We just haven't told it anything worth knowing.

Why hasn’t AI made marketing better?

Picture a Monday dashboard review. Website traffic fell 30% last month, and the demand team wants to move more budget into paid media to win it back. Nobody wants to be the person who watched traffic drop and did nothing, so the room nods along.

What the room doesn't know is that the media team switched that campaign off on purpose, because it was buying clicks rather than customers, and analytics has watched cost per lead fall 40% while leads held steady. Every fact needed to make the right call exists inside the company. It's just spread across people who don't attend the same meeting. Hand that dashboard to AI and it will produce a confident, well-organized plan to undo something that was working.

Companies have always run on knowledge nobody wrote down. That worked well enough when work passed through people, because people carry context with them. They hear about the killed campaign in the elevator. They remember why the last launch slipped. Organizations stored a surprising share of their judgment in hallway conversations and never noticed, because the hallway never sent an invoice. AI doesn't walk the hallways. It shows up to every meeting having missed every previous one.

What's missing has a name, and it's an old, plain word: context.

I mean something specific by it. When it comes to AI, context is your data (what customers do, what campaigns produced, what research revealed) plus your institutional knowledge (what the brand values, what you prioritize, what you've learned the hard way), applied to the decision in front of you. The data tells you traffic fell. The institutional knowledge tells you why that's fine.

The irony is that the companies most proud of their institutional knowledge may be the least prepared to use it. When a business runs on shared instincts, shorthand and relationships, nobody feels much urgency to document how decisions get made. Organizations stored a surprising share of their judgment in hallway conversations and never noticed, because the hallway never sent an invoice.

AI doesn't leave the gap empty when context is missing. It fills it with plausible guesses in the same assured tone as everything else, which is how output ends up polished and wrong at the same time. Context is confidence. Without it, you're reviewing guesses.

You can watch the difference happen with that campaign brief from earlier. Give a model nothing but an audience and a vague goal, and you get "Unlock efficiency at scale." Tell it four things the company already knows (the real priority is landing three enterprise pilots, buyers are drowning in disconnected tools, security concerns stall deals, and the company works with a customer's existing stack rather than replacing it), and the same model with the same prompt leads with "Make the AI you already have work together, governed, coordinated, proven on one operation."

Your competitors have the same models you do. What you feed them is the one part they can't copy.

Faster people, same old silos.

This is where most AI rollouts quietly stall. Every team adopts its own tools and feeds them its own version of the story. PR has one set of prompts, social has another, the web team has a third, and each person spends the first ten minutes of every chat re-explaining the business from memory. Individuals get faster. The team doesn't, because the gains never travel past the person who typed the prompt.

Everything is disconnected, so the business impact doesn’t compound.

We've been here before, more than a century ago. When factories first electrified in the 1890s, most of them pulled the big steam engine out of the basement and installed a big electric motor in its place. Everything else stayed put: the same drive shafts running the length of the building, the same belts, the same floor plan designed around one central source of power. Productivity barely moved for decades. The economist Paul David traced the eventual surge to the 1920s, when manufacturers realized electricity let them put a small motor on every machine and redesign the factory around how work actually flowed from one station to the next. What finally paid off was rearranging the work around the new power, a shift that took the better part of forty years.

Most marketing organizations are in the engine-swap phase of AI. We've installed a faster motor at every desk and left the floor plan alone: the same handoffs, the same alignment meetings, the same knowledge stranded in separate departments. Then we wonder why the results look like 1900.

In marketing, redesigning the floor means building one shared context behind every tool your teams already use: a single, governed source of what the company knows, captured once and kept current, that the brief writer, the media planner and the social team all draw from. Forcing everyone onto a single platform is tempting and almost always a mistake, because teams have reasons for their tools and ripping them out is a fight nobody wins.

With shared context in place, every brief starts from everything the company knows instead of whoever happened to be in the room, and every finished piece of work adds to what the next one starts from. That's the point where AI stops making individuals faster and starts making teams better.

Where should business context live for AI?

Right now, it lives in a dozen places at once. Slide decks, research folders, the CRM, your agency’s memory, a Slack thread from March, the head of whoever ran the last launch.

Where it should live is one place: a centralized hub that every AI tool your teams use can draw from. A single source of truth, in other words, though the phrase undersells the job. Facts are the easy part, and any competitor can assemble their own. What has to make it into the hub is the nuance, the hard-won judgment about your customers, your brand and your market that makes your business better than the one down the street.

Of course, knowing everything a company knows doesn't mean knowing everything it should. Organizations have bad assumptions, outdated beliefs and sacred icons that survive long after anyone remembers why they're sacred. Giving AI access to institutional knowledge makes it more useful. It doesn't make that knowledge right.

Which is why the harder questions are human ones: which beliefs still hold up, what's true when two teams disagree, who owns it, and how what you learn this month makes it back in before next month's work starts. That makes this an operating question as much as a software one.

See it for yourself in a week

Before you overhaul anything, run a small test. With one workflow and the AI tools your team already has, you can see two things in about a week: how much better the work gets when AI knows what your company knows, and how much effort it takes to pull that knowledge together.

The side-by-side shows you what context is worth. The tally shows you where it’s hiding. Doing that once is easy. Doing it across every brief, plan and campaign, and keeping it current as the business changes, is the part worth getting help with.

Writing down what your company knows has always been worth doing. It would have made onboarding faster and Monday meetings shorter, and most organizations skipped it because the hallway covered for them. AI has removed the hallway. The companies that get real value from it will be the ones that finally do the homework they've been putting off for decades, and they'll discover, perhaps slightly to their annoyance, that the humans benefit too.

Where We Come In.

It took factories decades to rebuild around electricity. Your marketing team doesn’t have to wait that long to rebuild around AI.

This is the work we do with marketing teams at Traction.

We start by finding where knowledge gets lost between your teams, including in the AI tools nobody officially approved. Within weeks you have a map of what context to capture and which workflows to start with. Then we build it alongside your team, one workflow at a time, connecting the tools you already use to the context they’ve been missing.

The payoff shows up where the handoffs used to be: less rework, faster time to market, work that sounds like your brand, and budget decisions made with the whole company’s knowledge.

If you’d like help building shared context for your marketing team, call us at 415.962.5800 or email futureproof@tractionco.com.