The Creation Gap Is a Production Problem
The marketing industry has spent a decade and a fortune building the pipeline that moves a message from data to delivery. Customer data platforms, marketing automation, AI agents, the whole machine for targeting and sending at scale. Most of it works. The part nobody bought a tool for is the part upstream of all of it: turning a campaign idea into something the machine can actually send. In the State of Martech 2026 interview, Knak co-founder Brendan Farnand gave that overlooked space a name, the creation gap, and argued it has been hiding in plain sight. He is right, and the sharper version is that the creation gap is really a production problem.
What the creation gap actually is
There are two things people mean when they say "creation," and the conversation blurs them. One is making the asset: the design, the copy, the visual idea. The other is the end-to-end process of getting that idea produced, the brief, the build, the QA, the legal review, the localization, the approvals, and the handoff into a marketing automation platform for sending. The first kind of creation is the glamorous 5%. The second kind is the other 95%, and it is where enterprise campaigns go to wait.
That second kind is production, and the distinction matters because the tools have diverged. Generative AI handles asset creation now. A marketer can describe an email and get a draft in seconds. Production is a different animal. It involves field marketers, demand generation, product marketing, and a marketing operations team that owns the execution platform, all touching the same campaign on its way out the door. Mapping that journey from ideation to execution surfaces a dozen handoffs and a long approval loop. Calling it a creation gap undersells it. The asset was never the hard part.
Why the bottleneck moved from speed to scale
The reflex is to say enterprise marketing is slow, that a single email takes weeks. That used to be true and mostly is not anymore. Email production time has collapsed: in 2023, 62% of teams needed two weeks or more to produce an email, and by 2025 only 6% did. Templates, builders, and AI drafting solved the easy half of the problem.
So the constraint moved rather than disappearing, relocating to where production actually breaks at enterprise scale: volume, personalization, localization, and governance. Producing one good email is fast. Producing 15 versions of it for different segments, localizing each into a dozen languages, routing every one through brand review, and doing that across regions every week is not. The drag is rarely the keystrokes. Approvals are the primary bottleneck for a third of teams, which makes this a coordination problem at its core. Speed at the unit level is solved. Throughput at enterprise scale is the open question.
Feeding the beast: When platforms outrun content supply
Follow the money and the gap turns into a budget problem. Enterprises have invested heavily in powerful execution platforms, and those platforms are hungry. They can run nurtures, journeys, and highly targeted segments, but only if there is enough content to fill them. There almost never is. 59% of marketing leaders say their teams lack the bandwidth to meet current content demand, 71% have to produce significantly more content each year, and 64% point to fragmented workflows as a major barrier to production. The demand curve and the capacity curve are pulling apart.
Farnand described a pharmaceutical customer that lived this exactly. They re-platformed onto Marketo on the B2B side and Salesforce Marketing Cloud on the B2C side, stood up the full stack, then looked at everything the platforms could do and realized they could never produce enough content to use it. They called it feeding the beast. Their conclusion is worth sitting with: they did not have a sending problem, a segmentation problem, or a data problem. They had a creation problem. The most expensive part of the stack was the part starving for input.
The bottleneck funnels onto marketing operations
Watch where the work piles up and you find the same team every time. Campaigns originate everywhere, in field marketing, in demand gen, in product marketing, and they all funnel down to the one group that operates the execution platform. Marketing operations is built for segmentation, data analysis, and running the marketing automation platform. That is the sweet spot. Instead, ops gets pulled upstream into production because they are the only people who know the technical intricacies of getting a campaign from a marketer's request into something sendable. The bottleneck forms fast.
This is why throwing another tool at the problem rarely helps. Marketing's inefficient processes tend to predate the tech meant to fix them, and automating a broken handoff just moves the delay. The fix is structural: route the routine work so it flows without a specialist, and reserve the operations team for the exceptions that genuinely need their judgment. Tiered governance, where routine campaigns flow on their own and exceptions get review, gives back the capacity that a single shared queue quietly consumes.
Where AI fits in the production workflow
Generative AI looks like the obvious answer to a content shortage, and used carelessly it makes the gap worse. The pattern Farnand calls vibe contenting, the marketing version of vibe coding, is a marketer spinning up a campaign in a chatbot and tossing the output over the wall. The asset might look fine. The problems land downstream. Raw AI-generated HTML jammed into a marketing automation platform breaks in ways a marketer never sees: an email over 102KB gets clipped in Gmail, preview text runs long, images ship without alt text. Email on Acid's research found that 99% of emails analyzed carry serious or critical accessibility issues, and fewer than a quarter of marketers write alt text at all. The model does not know the rules, and the marketer does not either.
The brand cost is quieter but just as real. When campaigns route around the established process, skipping approvals and brand review, the result is drift: copy and design slowly wandering off-brand, multiplied across every team that found the shortcut. Call it shadow creation. It is not hypothetical. More than 70% of marketers have already hit an AI-related incident, a hallucination, a bias problem, or off-brand output that made it further than it should have. Access to AI is nearly universal now, but only about 6% of companies are capturing real value from it, by McKinsey's estimate. What separates the two is governance: whether the model runs inside a process with controls.
The enterprises getting this right are not handing marketers a chatbot and calling it a workflow. The most AI-forward companies in the world are the most cautious about exactly that. What OpenAI sees in AI-driven marketing is AI embedded in the production workflow: a marketer's request, structured automatically, kicks off generation of an on-brand asset built from pre-approved content, then drops the marketer back in to refine it before it moves through QA, collaboration, approval, and localization. The AI does the heavy lifting inside the controls. The human stays in the loop, and the workflow stays intact. Translation is the clearest version of the win, turning a multi-day localization cycle into minutes without sending a single asset to a developer.
What closing the creation gap unlocks
Solve production and two things show up on a CMO's dashboard. The first is speed to market, which sounds unglamorous until a competitor launches and the response takes six to twelve weeks. Every week of that lag is lost pipeline, customers won by the other campaign while yours is still in build. Enterprise teams that compress the production workflow report dramatic gains: 95% less time to build an email at Amazon, 59% faster time to market at Vanguard, and at FTI Consulting a 316x jump in output capacity. Those are throughput gains, measured across regions and teams.
The second is performance, which follows from speed in a way that is easy to miss. When an email takes weeks, testing is a luxury; when it takes minutes, teams can run 3 to 15 versions of a send, pick the winner, and fold what worked into the next campaign. Personalization works the same way, because personalization is just content at higher resolution. When every variant has to pass through a developer or an ops specialist, four audiences is ambitious. When production scales, personalization down to the segment, or the individual, stops being a someday project.
Constraint | Where it used to bind | Where it binds now |
|---|---|---|
Speed | Building a single email | Producing volume across regions and languages |
Testing | One version, ship it | 3 to 15 versions per send, iterate |
Personalization | A few broad segments | Segment or individual level |
AI | Access to the tools | Running the tools inside a governed workflow |
Constraint | Speed |
|---|---|
Where it used to bind | Building a single email |
Where it binds now | Producing volume across regions and languages |
Constraint | Testing |
|---|---|
Where it used to bind | One version, ship it |
Where it binds now | 3 to 15 versions per send, iterate |
Constraint | Personalization |
|---|---|
Where it used to bind | A few broad segments |
Where it binds now | Segment or individual level |
Constraint | AI |
|---|---|
Where it used to bind | Access to the tools |
Where it binds now | Running the tools inside a governed workflow |
The return to relevance
The most interesting consequence is not efficiency for its own sake. For 15 years, enterprise marketing optimized the mechanics: faster workflows, tighter processes, more volume. The inbox is now forcing a different question. 85% of commercial messages are never read, and Gmail has entered its Gemini era, with AI summarizing and ranking mail for roughly 3 billion users. After Gmail's move to relevance-based sorting, the brands that win the top slot see higher opens and fewer unsubscribes, while everyone else gets buried. Volume is no longer free.
When sending more stops working and the gatekeeper rewards relevance, the payoff shifts from producing more to producing better. That is the renaissance worth betting on. Once production stops being the constraint, and a campaign can be created and personalized in minutes, the scarce resource becomes attention to the audience again: what they actually want to hear, what engagement actually looks like. The efficiency gains are real. The point of them is that they free marketers to do the part that was always supposed to be the job.
Closing the creation gap is what makes that possible, and it is a production problem with a production answer. A platform like Knak runs the production engine in the middle of the stack, coordinating the workflow from brief through approval and localization to deployment, so the marketing automation platform finally gets fed. See how it works.
