A few weeks ago an owner asked me a question I get some version of constantly. He had been paying an agency four thousand a month for eleven months. Leads were coming in. The reports arrived on time and looked busy. And he could not tell me whether any of it was working.
He was not confused about his own business. He knew his close rate. He knew what a customer was worth. What he could not do was look at what he was paying for and form an opinion about it, because the thing he was paying for was invisible to him.
That is the real problem, and almost nobody says it out loud. You are buying something you cannot inspect. Every instinct you have developed for judging a vendor over twenty years of running a business is calibrated for work you can see. This work you cannot see.
You Are Not Qualified to Judge the Work. That Is Not an Insult.
Think about how you feel at the mechanic. The car makes a noise. Someone in a shirt with his name on it says the radiator is going. You do not argue, because you have no basis on which to argue. You have no idea what a radiator costs, how long one should last, or whether the noise you heard has anything to do with it. So you nod, and you pay, and you drive away with a faint feeling that you may have just been handled.
The trade knows you feel that way. Some of them build a business on it.
Advertising is worse, because at least the car either makes the noise afterward or it does not. With ads, the feedback is slow, noisy, and easy to narrate around. A bad month has forty honest explanations and one dishonest one, and from where you sit they all sound identical.
So let me take one thing off your plate straight away. You are not going to learn enough about campaign management to evaluate the craft. You should not try. An owner who spends his evenings learning enough Google Ads to second-guess a media buyer has spent a lot of expensive hours to arrive at a weak opinion.
What you can do is evaluate the things that do not require expertise. There are four of them, they are not technical, and anyone competent can produce all four inside ten minutes. Most of what follows is about those four. But first, two traps.
The Person Who Sold You Is Not the One in Your Account.
At most agencies above a handful of people, the person on the sales call and the person in your account are different humans. That is not automatically sinister. It is how a company with a pipeline works.
It does mean something uncomfortable, though. Whatever read you took on competence during that first call, you took on somebody who will never touch your campaigns. You liked them. They were sharp, they asked good questions, they knew the acronyms. None of that is evidence about the person who actually shows up on Monday.
It also explains one of the most common ways these relationships go wrong. The salesperson promises a number. The salesperson is not the one who has to hit it, and was often never in a conversation with the person who does. So the promise gets made at sales altitude and lands on somebody who would have said no. Six months later you are frustrated, and the account manager is quietly frustrated too, and nobody is lying. The promise was just made by the wrong mouth.
Ask before you sign, and ask again if you are already in it: who is actually in my account, and can I meet them. It is a completely normal question. The answer, and how comfortable the answer makes everyone, tells you a lot.
Effort Is the Cheapest Thing to Manufacture.
Here is the second trap, and it is the one I see swallow the most money.
When you cannot judge results, you judge effort. It is a reasonable substitute and it is almost entirely wrong. A monthly report is the easiest artifact in this business to inflate. Impressions are up. Click-through rate improved. We added ninety-four negative keywords, refreshed the ad copy, and built out three new audiences. Fourteen pages of it, and a client who reads all fourteen pages feels looked after.
Nothing in that report is a lie. It is just that none of it answers the only question that matters, which is whether the money you put in came back out bigger. Activity is not achievement. A report that leads with what was done rather than what it produced has told you about the agency's month, not about your business.
The tell is simple and you can check it tonight. Open the last report you received. How far down the page do you get before you hit a number that exists in your own accounting, rather than a number that exists only inside an ad platform? If the answer is that you never do, you have been reading a work log.
Four Things Anyone Competent Can Show You in Ten Minutes
None of these require you to understand a bidding strategy. All of them are things a good operator has already prepared, and a bad one has to invent on the spot.
One: The Number You Agreed On Before the Money Moved
This is the big one and it makes the other three easier.
Somewhere before your first dollar went out the door, somebody should have worked out what a lead is allowed to cost you. Not what leads cost in your industry. What one is allowed to cost in your business, given your margin, your close rate, and what a customer is worth to you over the life of the relationship.
If you sell a service where an average customer is worth four thousand dollars in margin, and you close one in five of the leads you get, then five leads produce four thousand dollars. That sets a ceiling on what those five leads can cost before the whole exercise stops being worth doing. Where you set the target under that ceiling is a business decision. But the ceiling is arithmetic, and it comes from your numbers, not from anyone's benchmark.
Here is the diagnostic. Did anyone ever ask you for those three inputs? Your margin, your close rate, your average customer value. If nobody has ever asked, then nobody has ever managed your account toward an outcome, because nobody knows what the outcome is. They have been managing toward the only thing left, which is making the platform numbers look better than last month.
That number should be written down somewhere you can find it, and every conversation about performance should start there. We wrote more about how to derive it in Your First Ad Budget Isn't an Investment. It's Tuition.
Two: Your Name on the Account
You should be able to log into your own ad accounts today, at admin level, without asking anyone's permission. Not a dashboard someone built for you. The actual account.
There is no legitimate reason to withhold it. Nothing inside a Google Ads or Meta account is so proprietary that the person paying for the clicks cannot look at it. When an agency will not give you access, the reason is almost never protecting a method. It is that access makes leaving easy, and it makes the work visible.
The harder version of the same question is what happens when the relationship ends. Does the account go with you, along with its history, its conversion data and its learning, or does it belong to the agency and vanish? Accounts built inside an agency's own business manager, under their own billing, have a habit of turning out to be theirs at exactly the moment you need them to be yours. That history is worth real money, and rebuilding it costs you months. We covered the wider version of this in Your Ad Account Is Not Yours.
Three: A Human Being, Named
Who does the work, how long have they been doing it, and roughly how many hours a month land on your account.
The hours question makes people uncomfortable and it should be answerable anyway. If you are paying two thousand a month, you are not buying a full-time person and you should not expect one. You are buying a slice of several specialists, which is the honest case for using an agency at all. But you are buying a slice of a specific size, and an operator who will not put any shape on it is telling you something.
Experience matters here more than credentials. Somebody a year into the job can do good work with real oversight above them, and can do expensive damage without it. Both answers are acceptable. Vagueness about which one you are getting is not.
Four: A Reason for the Last Three Changes
Ask what the last three meaningful changes to your account were, and why each one was made.
Notice what this question is not. It is not asking how many changes were made, and it is not asking you to evaluate whether the changes were correct. You cannot evaluate that and you do not need to. What you are listening for is whether each change connects to a reason, and whether the reason connects to your number.
A good answer sounds like a decision. We were losing impression share on the terms that actually convert, so we moved budget from the campaign that generates volume but not customers. A bad answer sounds like a list of chores. We refreshed the ad copy and added negatives and adjusted some bids. The first one has a thought behind it. The second is what you say when there was no thought and there was a report due.
Three Warning Signs That Are Wasting Your Time
Now the other direction. Some of the most repeated advice about spotting a bad agency is wrong, and it costs owners good relationships.
"They Barely Changed Anything Last Month."
You will find plenty of advice telling you to open your change history and treat a quiet month as neglect. You will also find experienced practitioners arguing the exact opposite, that constant changes are the actual warning sign, because every change resets the platform's learning and most of them are somebody figuring things out on your money.
They cannot both be right, and the reason they disagree is that change count is not a competence signal in either direction.
What matters is the shape. A new account should be busy early, because early is when the unknowns get resolved. A mature account that is hitting its number should get quieter, because the work has moved from fixing to protecting, and there is genuine skill in leaving a working thing alone. An account that has been running two years and still shows daily churn is not being tended. It is being fiddled with.
So the honest version of this check is not how many. It is: does the activity pattern match the age and state of the account, and can somebody explain the last few changes without reaching. Which is question four, again.
"They Would Not Guarantee Results."
Somewhere in this industry you will meet an agency that guarantees performance, and somewhere else you will read that a guarantee is a red flag from someone who is about to over-promise. Both positions are argued sincerely by people who know what they are doing.
A guarantee is a pricing structure. It is a statement about who carries the risk, and it is priced accordingly, which is why the ones offering it usually cost more. It is not a statement about skill. Plenty of excellent operators will not guarantee an outcome that depends on your offer, your close rate and your competitors' budgets, because those are three things they do not control.
What you should insist on instead is a forecast with its assumptions showing. Not a promise. A projection that says: at this budget, in this market, we expect to be roughly here at three months and here at six, and here is what we are assuming about your conversion rate to get there. That is a document you can hold somebody to, and unlike a guarantee it tells you how they think.
"This Month Was Down."
One month is not a verdict. On the budgets most small businesses run, one month is barely a sentence.
If your account produces thirty conversions in a month, the difference between a good month and a bad one is a handful of events, and a handful of events is noise. Judging on that is how owners kill campaigns that were working and, just as often, how they get talked into panic changes that make the next month genuinely worse.
The unit of judgment is conversions, not weeks. You need enough of them to tell signal from randomness, and on a small budget that takes longer than a calendar month. This is also why it is worth knowing whether your slow month happened to contain a holiday week, a supply problem, or your own phones going unanswered. And if the numbers you are arguing about came from three different platforms that each claim credit for the same customers, that argument was never going to resolve itself. We wrote about why in You Will Never Know Exactly Which Ad Made the Sale.
Twenty Minutes in Your Own Account
If you have access, there are a few things you can look at yourself without being an expert. None of these prove anything on their own. They are conversation starters, and the quality of the explanation you get is the actual test.
Does the account match the business you described? If you told them you only want commercial work, and the campaigns are full of residential terms, something got lost. This is the check that needs no technical knowledge at all. Read the campaign names and the keywords the way you would read a stranger's description of your business, and see if you recognize yourself.
Is the target in the account the target you agreed to? Bidding targets are visible in the campaign settings. If you said you would happily pay eighty dollars for a lead and the account is set to chase them at thirty, that setting is quietly strangling your volume. This is one of the most common expensive misalignments there is, and it usually comes from nobody ever writing the real number down.
Are you paying to appear in places you did not choose? Two settings matter here and both are visible in your campaign settings. Google can show your search ads on partner sites beyond Google itself, which you can opt out of at any time. And the default location setting targets people who are physically in your area or who have merely shown interest in it, which for a plumber who only serves one county is a meaningfully different audience than the one he thinks he bought. Both settings are defensible in specific circumstances. Neither should be news to you.
Is Google editing your account for you? There is a setting that lets Google automatically apply its own recommendations. A deliberate operator has a position on that setting. If it is on and nobody knew, your account has a co-manager nobody hired.
Where is the money actually going? Sort your campaigns by spend and look at the top two or three. That is your account. Whatever is happening in the little ones matters less than whichever campaign is eating most of the budget, and the conversation should be mostly about those.
When It Is Actually Time to Leave
Most owners in a bad relationship stay too long, and the reason is not stupidity. It is that leaving means starting over, losing the account history, spending a month onboarding somebody new, and admitting that the last year was expensive. Those are real costs and they make waiting feel prudent.
Here is a cleaner way to think about it. You are not deciding whether the last year worked. You are deciding whether the next six months look different from the last six. If you have raised the same concern three times and heard three versions of the same reassurance, that is your answer, and the next six months are the last six repeated.
The other honest possibility, and I would rather say it than not, is that they are not the problem. If your follow-up is slow, your offer is undifferentiated, or your margins cannot support what a customer costs to acquire in your market, no agency on earth fixes that from inside an ad account. Switching agencies when the constraint sits somewhere else just buys you a new onboarding period and the same result. We wrote about that harder conversation in Some Businesses Should Not Be Running Ads Yet.
Give it a fair window, defined in advance and in conversions rather than weeks. Then hold the line you set.
Ask Us the Same Questions
Everything above applies to us. If you talk to us about running your Google Ads, we will ask you for your margin and your close rate before we quote you anything, because without them we would just be guessing at what good looks like. The account is yours and stays yours. You will know who is in it.
We do this all day long, so if you would rather get back to running your business, that is what we are for. And if you would rather keep doing it yourself, take the four questions with you anyway. They work on anybody.
