Sloppy is easy. If somebody turns up late with your competitor’s name still in the deck, you already know.
The engagements that go badly almost never start that way. They start with a firm that seemed noticeably more professional than the others, and the things that made them seem more professional are, on inspection, the warning signs.
That is the uncomfortable structure of this market. Polish is cheap to manufacture and diagnosis is expensive, so a pitch process that rewards polish selects fairly reliably for the wrong thing.
Here are the signals that read as strength in the room and mean the opposite.
The strategy deck that existed before they saw your data
Forty slides. Channel plan, content calendar, projected results by quarter. Nobody has looked inside your ad account or your analytics yet.
It feels like enthusiasm and preparation. What it actually demonstrates is that the recommendations did not come from your situation, because there was no access to your situation. You are looking at a template with your logo on it.
The version that should impress you is smaller and less satisfying: two or three questions they cannot answer without seeing the data, and a clear statement of what they would look at first. I set out what a real first ninety days looks like separately, and the short version is that the finding comes before the plan.
A plan produced without evidence is a plan produced for somebody else.
The proposal that came back in forty eight hours
Same root cause, different tell, and this one is harder to argue with because responsiveness is genuinely a virtue.
A considered proposal for anything non-trivial takes longer than two days, because part of it involves going and looking. Speed here usually means assembly rather than thought.
Which is not to say slow is good. The distinction is whether the fast response is a proposal or a set of questions. Questions in forty eight hours is an excellent sign. A full strategy in forty eight hours means it was already written.
Nothing they asked could have cost them the deal
This is the one I would weight most heavily, and almost nobody notices it.
Somebody who has done this work seriously will ask at least one question whose answer might disqualify you. What is your gross margin. What happens when the leads arrive, and who handles them. Has anybody run this before, and what happened. What would make you cancel in month three.
Those questions risk the sale. Which is exactly why asking them signals that the person cares whether the engagement works more than they care about closing it.
If an entire pitch process passes without a single question that could have led them to say no thanks, you did not have a diagnostic conversation. You had a sales call with good manners.
Percentages without baselines
Three hundred percent increase in leads. Two hundred percent growth in organic traffic.
From what. Three leads to twelve is a three hundred percent increase and tells you nothing about whether they can operate at your scale.
Ask for the absolute numbers, the timeframe, and the spend level. A firm that has genuinely done the work will give you those or explain which client confidentiality prevents it, and will offer a reference instead. A firm that gets uncomfortable was relying on the percentage doing work the underlying number could not.
The same logic applies to the awards and the badges. Most agency rankings are partly a rate card, which does not make the listed firms bad and does mean the placement is not evidence.
The senior person who is not in the contract
The pitch is run by somebody impressive. The work will be done by somebody you have not met.
This is not inherently wrong, and pretending otherwise would be unfair to how agencies are structured. Senior people pitch. That is the model.
The red flag is the refusal to name the actual team in writing. Ask who specifically will do the work, how much of their week you get, and what else they are on. Then ask for it in the contract. The reaction to that request tells you more than the answer does, and it pairs with the ownership question that should be settled before anything is signed.
The contract terms that quietly decide everything
Two things are worth reading properly, and neither is the price.
Auto-renewal. Reported as the highest-frequency problem in agency contracts. The clause rolls you into a full new term unless you give written notice inside a window that typically closes 30 to 90 days before the renewal date. Miss it and performance becomes irrelevant for another year.
The ask that fixes it: replace automatic renewal with opt-in renewal requiring both parties to agree, or negotiate the notice period down to thirty days. A firm confident in the work rarely fights this, because they expect you to want to continue.
Initial term. Three to twelve months is normal, six is common. What matters is whether the length matches the work. SEO reasonably needs six to twelve because it is slow. Paid social does not. A twelve month lock on a channel that produces signal in three weeks is a financing arrangement rather than a strategy.
Inside the first sixty days
One signal, and it is the one that precedes most quiet failures.
The report only ever contains good news. Every month, a metric went up, and the metrics that went up are not always the same ones.
What you want instead is a report that names one thing that did not work and what is being done about it. Ask for that explicitly in the first review, as a standing item. The response is diagnostic in itself: a good operator is relieved to be allowed to tell you, because they have been managing the awkwardness of it alone.
Being fair about all of this
Every signal above has an innocent explanation, and I would rather say so than hand you a paranoia checklist.
The fast proposal might mean a slow month and genuine eagerness. The template deck might be an honest attempt to show how they think when you gave them nothing to work with. The unnamed team might be a firm that reassigns based on who is free, which is normal and not sinister.
So treat these as questions rather than verdicts. One is nothing. Three together is a pattern, and the pattern is that the sales process is more developed than the delivery process.
Where I come out
The most useful thing you can do in a pitch is stop evaluating how impressive people are and start noticing whether anybody is trying to find out what is actually wrong.
Polish is the cheapest thing to buy in this industry. Diagnosis is the expensive thing, and it is the only part you are actually paying for.
If you are still deciding what kind of help you need at all, the model question comes before any of this. My notes on what I work on cover the rest. If you’d like a second opinion on a proposal in front of you, drop me a line on email, WhatsApp or LinkedIn and we can have a quick chat. I’m contracted full time so this isn’t a pitch.