How to know the next marketing provider will deliver

You have bought marketing before. The proposal was polished, the meetings were warm, and the reports arrived on time. What never arrived was the work the reports were supposed to describe.

So the next provider starts the conversation in a hole. When they say they are different, you hear a claim you cannot check.

Here is the useful part. Most of what went wrong last time was visible before you signed. It sat in the provider's economics, in plain sight.

It sits in the next provider's economics too. Promises are hard to weigh. Arithmetic can be checked.

Read the arithmetic and the worst case is that you walk away a week earlier. Skip it and you can spend another year buying reports. The fee will bother you now and then. The year never comes back at all.

The burn was arithmetic

Start with the model behind the last engagement.

A high-volume provider, the kind practitioners call a mill, runs on one equation. Many clients, junior hands, and a sales team replacing whoever leaves. The senior person who impressed you sells. Delivery goes to an account manager carrying a stack of accounts.

One agency owner has described the pattern from the inside. Shops hand 25 accounts to one coordinator, then wonder why the bucket leaks. He had been the burned client twice himself before writing it.

Now do the division.

Take the monthly fee you paid. Divide it by the number of clients your account lead carried. That is roughly the attention your account could get, before anyone got careless.

Nobody at the last shop set out to waste your money. The model spent your fee the way it spends every fee.

The shipping pattern follows the same equation. Month one is heavy, because onboarding is templated and cheap. Month three thins out, because your account has joined the stack. If you kept the old reports, reread them in order. You will likely watch the thinning happen on paper.

The model decided who answered your emails. The model decided what shipped in March. You were angry at a person, and the person was mostly a symptom.

The churn was priced in

A mill does not need you to stay.

In Setup's Marketing Relationship Survey, 40% of clients said they planned to switch agency partners within six months. At volume, that churn is not an emergency. It is a line item, covered by next month's sales calls.

Your departure was budgeted before your kickoff.

So the experience you had was common, and it was not personal. It was a model operating normally.

That reframes the question you carried out of it. How do I know the next one is different has a sharper version.

What does the next one's model need from me? A model that needs your retention behaves one way. A model that only needs your signature behaves another.

AI changed what a mill ships

For a while, monthly volume was a rough proxy for effort. A thick content calendar meant hours were spent somewhere.

AI ended that. A mill can now produce more in a week than it once produced in a quarter. The stack of deliverables tells you almost nothing about the attention behind it.

Google noticed. Its March 2026 core update named scaled content abuse as a target. SEO analysts tracked drops of 50 to 80% for sites publishing AI content at scale without editorial review.

Read that as a buyer. The modern mill does not merely underdeliver. Unreviewed volume now carries a penalty, so it can leave your domain worse than it found it.

The check has moved with the tools. What ships matters less than who reviews it, and whose experience is inside it. Volume is cheap now. Judgment is still expensive, and it is the thing you are paying for.

A practice reads differently on paper

By a practice, I mean a provider whose model needs your retention. Few clients. Senior hands on the work. Economics that stop working if clients leave.

You do not have to take any of that on faith, because each piece is checkable.

Ask who does the work, and you get a name you will meet. Ask the client count, and the division comes out in whole hours. Ask about terms, and the notice period is short, because retention is earned monthly.

Ask what you own, and the answer is a list with your name on it.

One more tell costs nothing to check: willingness to lose the sale. A provider who tells you what the work cannot do is showing you their incentives. So is one who names the part you do not need.

Recommending less only makes sense in a model built on staying.

Then ask what they need from you. This one deserves a pause. A provider who plans to use your experience has to ask for your time.

A provider who asks for nothing is planning to sell you the generic layer. That is the product that burned you.

All of it surfaces under a short set of questions put to any provider before money moves, including me.

Small providers carry their own arithmetic

Fairness requires the other half of this.

A one-person practice concentrates risk in one head and one calendar. If that person is unavailable, the work slows. If they vanish, whatever lived only in their head vanishes too.

Notice who writes the warnings, though. The pieces warning you about agency mills are mostly written by small providers. The pieces warning you about freelancers who disappear are mostly written by agencies.

Read both genres as bids for your fee. Then run the same kind of check on everyone.

For a small provider, the check is capacity and portability. Ask what happens when they are sick, or full. Ask what you hold if they stop answering: the accounts, the files, the documentation.

A one-person practice can answer this well. The good ones build the answer into the engagement. Everything they know about your brand then exists on paper you own.

What to do with the proposal on your desk

Take the proposal in front of you and do tonight what nobody told you to do last time.

Ask for the name and the client count, and run the division. Ask for the exit terms and the ownership list in writing. Ask how many hours they need from you, and worry if the answer is none.

The last provider's promises and the next provider's promises will sound alike, because promises are free. The arithmetic is where providers differ, and you can read it before a dollar moves.

Brett Wharton, looking ahead