AI ate my lunch: what to do in your next campaign
Key Insights
- A vendor view, not a measurement: AI now produces content faster than most review processes can check it, so the constraint has moved from generation to approval.
- 51% of UK B2B marketers spent three hours or more last week explaining things others could not find for themselves (Jam 7 survey of 100 UK B2B marketers, October 2026, self-reported, opt-in sample).
- Marketers who lose hours re-explaining context were twice as likely to lose hours checking AI's work too.
- More production capacity without orchestration creates more work in progress, larger approval queues and a new form of Coordination Tax.
- Start your next campaign with the hypothesis it must prove or kill, not with the assets.
If AI now writes the drafts, builds the variants and resizes the creative, what is left for you to do? The answer is the part that decides whether qualified pipeline appears: direction, shared knowledge and judgement. This article shows how to plan your next campaign around those three, one step at a time.
What is left for a marketer when AI does the production?
The work that is left is deciding what to prove, supplying the knowledge AI works from, and judging what ships. Production was never the part that created qualified pipeline. It was the part that took the most time.
When production gets cheap, the weak points show up elsewhere. One governance vendor, CoreMedia, argues that AI now produces content faster than most review processes can check it, so the constraint has moved from generation to approval. That is a vendor view, not a measurement. It still matches what many teams report: the drafts arrive quickly, then wait.
So the feeling behind "AI ate my lunch" is fair, but the diagnosis is partial. Your task list shrank at the production end. It grew at the direction and review end. Your next campaign should reflect that shift rather than ignore it.
Why does faster production create a new bottleneck?
Faster production creates a bottleneck because every extra asset still needs a decision, a check and an approval. Output scales with software. Review scales with people.
Content governance is the set of rules that settle who can use which AI models, how output is approved before publishing, and what record shows a piece was checked (CoreMedia, enterprise AI content governance). Without those rules, each asset gets judged from scratch.
Our own position is that generative AI has made content production easier but has not solved marketing fragmentation. More production capacity without orchestration creates more work in progress, larger approval queues and a new form of Coordination Tax. Coordination Tax is the overhead of briefing, handoffs, revisions and approvals that keeps marketing slow even as the tool stack grows.
Where does the time actually go?
A lot of it goes on re-explaining. In our October 2026 survey of 100 UK B2B marketers, 51% spent three hours or more last week explaining things others could not find for themselves: brand, audience, positioning or past decisions.
The same group was twice as likely to lose hours checking AI's work. That fits how AI behaves. It only works from what it is given, so missing context comes back as rework.
Read this with care. The sample is self-reported and opt-in, so it shows direction, not cause. It does not prove that lost context drives AI rework. It does suggest that the fix is upstream of the tool you pick.
How do you plan a campaign when AI does the drafting?
Plan it in the order of the decisions, not the order of the assets. Four steps work:
- Write the hypothesis. One sentence: what do we believe, which ICP segment, and what result in qualified pipeline would prove it?
- Set the kill criteria. Agree the number and the date at which you stop. Every campaign should either prove a hypothesis or kill one.
- Load the knowledge. Put brand voice, positioning, proof points and past decisions where people and AI can both read them.
- Decide the review path. Name who approves what, and how fast, before anything is generated.
Only then ask AI for assets. The drafts will be closer to done because the brief is no longer in someone's head.
What should you put in front of AI before it writes?
Give it the same briefing you would give a new senior hire. That means your brand voice, your ICP, your product facts, the proof you can show and the claims you must avoid.
A Marketing Brain is a shared store of that knowledge, so each campaign starts where the last one finished instead of from a blank page. It keeps the brief out of individual heads and out of scattered documents.
This also protects your team's judgement. A reviewer who checks output against a written standard is faster and fairer than one who checks against memory. It turns "does this feel right?" into "does this match what we agreed?".
How do you review faster without lowering the bar?
Set the bar before the work, then review against it. Three habits help:
- Check against the hypothesis. Does this asset help prove or kill what we set out to test?
- Use a fixed checklist. Voice, claims, evidence, next action. The same four questions every time.
- Keep a human on approval. AI can draft and pre-check. A person still owns what goes out.
As a Head of Marketing, you also want a pilot, not a big-bang change. Run the new approach on one campaign, with a clear success measure, over one to two quarters. Keep what moves qualified pipeline and drop what does not.
Which fix works: hire, buy a tool or change the operating model?
Only a change to the operating model removes the coordination work itself. Hiring and tools both leave it in place. This comparison comes from Jam 7's article "The Coordination Tax: Where B2B Marketing Time Really Goes".
| Approach |
What it does |
What it does not do |
| Hire more people |
Adds creation capacity |
Reduces coordination overhead |
| Add another tool |
Automates individual tasks |
Connects strategy to execution |
| Redesign the operating model |
Addresses the structural cause |
Avoid the upfront investment |
Adding a third writer adds another handoff. It does not remove the context lost at the ones you already have.
What does coordination cost, and what does fixing it cost?
Coordination cost is the salary time spent briefing, chasing approvals, re-explaining and reformatting. It rarely appears on an invoice, so most budgets do not name it. Measure it first. Ask your team to log one week of time in two groups only: creation work and coordination work.
On the fix side, our Product List sets out the price points for AMP (Agentic Marketing Platform®), our Growth OS. AMP Solo is GBP 1,000 a month. AMP Team is GBP 2,500 a month and covers up to five Growth Agent seats. AMP Enterprise is GBP 15,000 a month. Compare those figures with your measured coordination hours, not with a headcount line.
What does this look like in practice?
In the Evolve Business Group case study, Jam 7 built messaging foundations first, then trained AMP on them. A full website refresh launched on schedule within the first three months. First drafts landed close to done, with technical and sales stakeholders reviewing for accuracy. This is one client, a managed networking provider, and the result describes speed of delivery, not a revenue figure.
What should you do next?
Pick the one campaign you are about to brief. Write its hypothesis and kill criteria today, list the context a new hire would need, and name the approver. That is a morning's work, and it moves your role up the chain from producing to directing.
If you want to see where your own team loses context, read the full Jam 7 research and score your own team.
FAQ
Will AI replace marketers?
AI replaces some production tasks, not the accountability for pipeline. Someone still sets direction, supplies company knowledge and approves what ships. Those tasks grow in importance as production gets cheaper, so the role moves toward judgement and ownership of results.
What is the first thing to change in my next campaign?
Write the hypothesis and the kill criteria before any asset is briefed. State the ICP segment, the expected qualified pipeline result and the date you will decide. This single step stops AI from producing volume that nobody has agreed to measure.
Why does AI output need so much checking?
AI works only from the context it is given. When brand, audience and positioning are missing or scattered, the output drifts and reviewers fix it by hand. Our survey found that marketers losing hours to re-explaining were twice as likely to lose hours checking AI's work.
How do I speed up approvals?
Agree the standard first. Name the approver, set a fixed checklist for voice, claims and evidence, and review each asset against the campaign hypothesis. Faster approval comes from shared rules, not from lowering the bar.