Why Faster Drafting Didn't Make You Faster

By Midya U · Midya U Advisory

In brief: The Content Marketing Institute found 95% of B2B marketers use AI, but only 39% see better performance, because the tools sped up drafting rather than the approval queue that determines how fast work actually goes out the door. Before your next renewal, count the calendar days a piece of copy spends in review and remove the reviewers you can't justify.

A year in, the writing assistant is doing what it promised. First drafts that used to take an afternoon take 15-20 minutes. And the campaign still went out three weeks late, the same as last year, which is the part nobody can explain in the quarterly review.

That gap shows up in the survey data now, and it is wide enough to be worth a conversation before your next renewal.

Why isn't the AI showing up in your results?

Because it made the fastest part of the work faster.

The Content Marketing Institute reported in 2026 that 95% of B2B marketers use AI tools in at least one part of their workflow, while only 39% say it is improving performance. Supermetrics, in a survey of 435 marketing professionals across five countries fielded at the end of 2025, found 6% had fully embedded AI into their workflows.

So adoption is close to universal and the returns are not. The usual explanation is that the tools are overhyped, and I do not think that holds up - the tools mostly do what they claim. The problem is where they were pointed.

Every process moves at the speed of its slowest step. Drafting was rarely the slowest step in an institutional marketing team. If a piece of copy takes three weeks to get out the door and four hours of that was writing, then cutting the writing to thirty minutes gives you back three and a half hours out of three weeks.

What is slowing the work down?

Approvals, in most institutions.

A single piece of program copy can pass through the faculty that owns the program, an associate dean, the international office if it touches admissions, occasionally the registrar, and sometimes legal. Each of those is a queue, and each queue is a person with a full calendar and no service standard attached to their review.

None of that is touched by a writing assistant. You can generate the draft in twenty minutes and still wait eleven days for the second reviewer, and no amount of additional licence spend changes it.

This is the diagnosis that tends to land badly in a meeting, because it moves the conversation from a purchase, which is easy to authorise, to a governance change, which is not.

Who decided you needed the tools?

In most cases the decision came from above you.

The same Supermetrics survey found 80% of marketers feel pressure to adopt AI, and 89% of those said it came from executive leadership or the board. Meanwhile 37% said leadership had given them no clear AI strategy to work to.

That sequence explains a lot. A president reads something over the summer, asks the marketing lead what the institution is doing about AI, and a licence is the fastest available answer to a question that was really about strategy. The tool arrives before anyone has written down which problem it is solving, and a year later the honest answer to "what did it change" is "drafting."

A second finding in that survey is worth flagging. 52% said decisions about data strategy happen outside marketing, and only 31% said the CMO is meaningfully involved. In higher education that usually means IT owns the student CRM, so an AI plan that depends on data you do not control carries a dependency nobody has scheduled.

What would you change instead of buying?

Pick the piece of work that matters most and redesign the path it travels, before adding anything to the stack.

Write down every step between request and publication, and put a name and a working-days target against each one. Most teams discover two or three reviewers whose sign-off nobody can justify, added years ago after something went wrong once. Removing a reviewer is free and returns more calendar time than any tool on your invoice.

Then decide what the AI is for, in a sentence, against a step you have measured. Speeding up drafting is a legitimate goal if drafting is your bottleneck, and in most institutional teams it is not.

I would name the limit here. The Supermetrics and CMI research covers brand and agency marketers, mostly B2B, and neither surveyed higher education. The adoption-without-returns pattern is well evidenced; the claim that approvals are the specific bottleneck in your shop is mine, drawn from how institutional sign-off is structured, and it is the first thing I would test rather than assume.

What would I do first?

Take the last five things your team published. For each one, find the date it was requested and the date it went live, and mark where the calendar days actually went.

Most teams have never done this, which is why the AI conversation keeps happening at the level of tools. Once you can see where three weeks goes, the conversation changes on its own. And if it turns out drafting really was your constraint, you have just justified the subscription with evidence, which is a better position than the one most teams are in at renewal.

That exercise is where the Marketing Function Audit & Reset starts, and it pairs with the renewal-season triage in Five AI Tools Your Marketing Team Pays For.