Where Does AI Actually Belong in My Marketing?

Short answer: Adoption is not the constraint. Redesign is. Seventy-eight percent of organizations now use AI, and marketing uses it more than any other function, yet more than 80% report no enterprise-level profit impact. The companies seeing returns did not buy better tools. They rebuilt the work around them.

The tension: speed versus judgment

AI makes production nearly free. It does not make decisions better. Marketing leaders are being asked to capture the first without sacrificing the second, and most are doing it by adding tools to processes that were designed for a slower era.

What the data shows

Adoption is near universal. Returns are not. McKinsey's State of AI survey of 1,491 respondents found 78% of organizations using AI in at least one function and 71% regularly using generative AI, with marketing and sales leading every other function. More than 80% reported no tangible impact on enterprise-level earnings. Usage has outrun value.

The variable that predicts impact is workflow redesign. Of everything McKinsey tested, fundamentally redesigning workflows had the strongest correlation with bottom-line results. Only 21% of organizations using generative AI had done it. Four in five are running new technology through old processes.

The gains so far are efficiency gains, and oversight is thin. In Gartner's 2025 CMO Spend Survey, CMOs credited generative AI with time efficiency (49%) and cost efficiency (40%), while only 27% cited increased capacity to produce. Meanwhile McKinsey found just 27% of organizations review all AI-generated content before it is used. Teams are saving hours and publishing unreviewed work in the same motion.

The pattern: the bolt-on trap

Read together, the numbers describe what I call the bolt-on trap: AI attached to a workflow that was never redesigned to use it.

A bolted-on tool accelerates every step in the process, including the steps that should not exist. The draft arrives in minutes and then waits three days in the same approval chain. The hours saved are absorbed by review, rework, and meetings. Activity rises, output looks impressive, and the business result does not move.

What this changes

Redesign one workflow before licensing another tool. Pick a single process, map it end to end, and rebuild it assuming AI exists. Remove steps, do not just speed them up.

Assign AI to repeatable work and people to judgment. Drafting, reformatting, tagging, and research preparation are machine work. Positioning, final approval, and reading a specific market are not. The line should be explicit, and written down.

Measure outcomes, not usage. The number of prompts run or assets produced tells you nothing. Track hours returned to the team and what those hours were redeployed to.

The resolution

AI is a multiplier of process quality. Applied to a sound workflow, it compounds the advantage. Applied to a broken one, it scales the flaw. The strategic question is not which tool to adopt. It is which work deserves to be redesigned first.

The question to ask: Which of your marketing workflows was designed after you adopted AI, and not before?

If the answer is none, that is where the return is hiding. Reach out and we can map the first one.

Sources

McKinsey, The State of AI: How Organizations Are Rewiring to Capture Value. 1,491 respondents, surveyed July 2024, published March 2025.

Gartner, 2025 CMO Spend Survey. 402 CMOs and marketing leaders, surveyed February to March 2025.