The hardest thing about hiring an agency is that the feedback arrives late.

You sign, you pay, and for a while very little visible happens. That silence is genuinely ambiguous – it looks identical whether the work is progressing normally or whether nothing is being done at all.

So owners wait, then get impatient, then cancel. A striking number cancel at almost exactly the wrong moment.

Where You Are Changes What “Working” Looks Like

One thing worth settling before judging anything: an agency’s realistic timeline depends on the market it is competing in, and metro geography decides more of that than most owners expect.

Kansas City is a clear case. The metro spans Missouri and Kansas – the Census Bureau builds metropolitan area definitions around economic integration rather than state lines, so Kansas City, Kansas, Kansas City, Missouri, Overland Park and Olathe are one market to a customer and four sets of paperwork to a business.

For an agency that means two things. Competition is metro-wide rather than city-wide, so the field is larger than the city name suggests. And there are two of everything available – two states’ chambers, two sets of local media, two licensing environments. An agency that only works the side you are registered in is leaving half the market untouched.

Ask early which geography they are actually targeting. If the answer is just your city name, they have not looked at the market.

The Problem With Judging Too Early

Marketing results lag the work that produces them, often by months.

That lag creates a trap. The third invoice tends to land right around the time the earliest real results are surfacing but before they are obvious enough to feel like proof. The owner has now paid three times and seen very little. Cancelling feels like the disciplined decision.

It is usually the expensive one, because the spend has already happened and the return was about to start arriving. Everything paid so far becomes sunk, and starting again elsewhere resets the clock to zero.

This does not mean every agency deserves patience. It means patience should be spent on evidence rather than hope – and evidence exists well before rankings and revenue move.

What Should Exist by When

You cannot judge an agency on results in month one. You can absolutely judge it on whether the work is real.

The most important row is month one. A baseline captured before changes begin is the single clearest sign of a competent engagement. Without it, nobody can honestly claim credit for improvement later, and any agency that skipped it has made its own work unprovable.

Reporting: What You Get Versus What You Need

Most agency reports are built to look busy. A useful report is built to be checked.

Things that sound impressive and prove little:

• Impressions, viewed alone
• Total backlinks acquired
• Social media reach
• Hours worked
• Rankings for your own business name

That last one deserves a flag. Ranking first for your own company name is not an achievement – you would rank there anyway. When branded searches are mixed into a performance report, every number inflates. The real scorecard is what happens on searches for what you sell, by people who have never heard of you.

Things that actually tell you something:

• Non-branded rankings, tracked over time against a stated baseline
• Calls, forms and bookings, not just visits
• Which specific pages gained visibility, and why
• What was shipped this month, listed plainly
• What is planned next, and what needs your input

Access: What to Hand Over and What to Keep

This one is quietly important, and most owners get it wrong in the same direction.

Agencies need real access to do real work. Withholding it guarantees a slow, hedged engagement. But access should be granted, never transferred – and the distinction shows up painfully at the moment a relationship ends.

The pattern to avoid is an agency building assets inside accounts it owns. It is rarely malicious; it is usually convenience. But when the relationship ends, the history, the audience data and sometimes the site itself walk out of the door, and the next agency starts from nothing.

A simple test before signing: if this ended tomorrow, what would I still have? If the honest answer is “not much,” fix that first.

Scope-Based Pricing Versus the Black Box

There are broadly two ways an agency can charge, and the difference predicts a great deal.

Black box: a flat monthly fee for “marketing.” No itemization. When you ask what you got, the answer is a summary rather than a list.

Scope-based: the deliverables are named up front. This many pages, this many links, these specific fixes, at this cadence.

Scope-based pricing is better for the client for an unglamorous reason: it makes the work checkable. It also makes patience easier, because when the results are still forming, you can at least confirm the agreed work happened.

If a proposal cannot tell you what you are actually receiving each month, that is worth resolving before signing rather than after.

The Questions That Separate Agencies

Most vetting questions get the same rehearsed answers. These do not.

• What will you measure, and what is the baseline today? Anyone without a baseline answer has not started thinking about proof.
• What have you decided not to do for me, and why? A real strategy excludes things. An agency that recommends everything is selling, not advising.
• What does month one look like specifically? Vague answers here reliably predict vague months.
• Who does the actual work? The person selling is often not the person delivering.
• What would make you tell me this is not working? The best answer is a specific one.

The last question is the most revealing. An agency willing to describe its own failure condition is an agency that expects to be measured.

Claims That Should Stop the Conversation

A few promises are not optimistic – they are misrepresentations. Treat these as stop signs:

• A guaranteed first position, on any timeline
• A “special relationship” with a search engine
• Results promised in weeks for a competitive market
• Pricing that only makes sense at volumes nobody could deliver individually
• Reluctance to name a single existing client

Nobody can guarantee a ranking, because nobody controls the ranking. Anyone claiming otherwise is either misinformed or counting on you being so.

Ask What They Are Doing About AI Search

This is the fastest way to find out whether an agency is current, and in 2026 it is a fair question to put to anyone taking a monthly fee.

AI assistants now answer a large share of the questions that used to produce a click. They pick which businesses to name using different inputs than a search engine does. Backlinks – the thing most agencies sell hardest – carry very little weight there. Mentions, review content, structured data and presence in communities carry most of it.

That produces a situation worth naming plainly: a business can rank first on Google and be invisible when someone asks an assistant for a recommendation. Those are two systems with two sets of inputs, and an agency reporting only on the first is reporting on half the market.

A current agency will have an answer to these:

• Is our business listed and complete on Bing, not just Google? Assistants have drawn on Bing’s index, and an incomplete listing there removes you from consideration.
• What structured data is on our pages? Assistants lean on it heavily.
• Are our reviews specific enough to be quoted, or are they all “great service”?
• Where are we mentioned without a link, and does anyone track that?

An agency treating these as a distraction is optimizing for how search worked years ago. The shift in how people locate information is not speculative – it is the same trend Pew Research has been tracking across a decade of online behavior, now arriving in search itself.

When Leaving Is the Right Call

Sometimes it is. The signals are behavioral, not numerical.

Leave when you cannot get a straight answer about what was done last month. Leave when the same report arrives with different dates. Leave when your questions are treated as a nuisance. Leave when six months have passed with no movement in anything, including the leading indicators that should have moved much earlier.

Do not leave because month two was quiet. That is what month two looks like.

There is a real cost to switching, and it is not just the new setup fee. The clock restarts, and whoever inherits the account spends their first month working out what the last one did. Two switches in a year can mean a year with no compounding at all.

Where a business is genuinely unsure whether the work is real, a second opinion is cheaper than a switch. Reviewing what has actually been delivered – before deciding anything – is a normal part of working with an advertising agency, and any competent firm will tell you plainly if the incumbent is doing fine.

The Shortest Version

Judge the work early and the results late.

In the first ninety days, you are checking that a plan exists, that a baseline was captured, that things are shipping on schedule, and that you can get a direct answer to a direct question. Those four are visible immediately and they predict almost everything that follows.

The revenue question is real, but it is a month-six question. Asking it in month two produces anxiety and nothing else – and acting on the answer is how businesses end up paying for the same first three months over and over.

And if you have replaced three agencies in two years, the common factor deserves an honest look first.