The numbers are permanent. You can write "question P1" in an email to a provider and they will know what you mean.
Not every question applies to every project: the right questions depend on what you are aiming for as an end result. Pick the ones that directly relate to the final product or the goal you want to achieve. Asking all of them without relevant judgement is more likely to confuse the person on the other side.
Ask these before you pay
P1
Whose name will the ad account be in?
Why you ask
The account holds the entire history of what worked. When you part ways it either comes with you or stays with them, and the next provider starts blind.
A good answer
The account is yours, they get partner access. This is standard practice and creates no inconvenience for a serious provider.
A bad answer
"We run it through our account, it is cheaper that way." Sometimes there really is a discount. Its price is losing the history and never seeing the real numbers.
How to check
Log in with your own email and see whether you can remove someone else’s access. If you cannot, the account is not yours.
P2
How exactly is your fee calculated?
Why you ask
On a percentage model the provider earns more when you spend more. Not necessarily a problem, but a conflict you should know about.
A good answer
Explains the model plainly and says what happens if the budget increases. Raises the conflict themselves if they work on a percentage.
A bad answer
One lump sum with ad spend and fee not separated. Then you know neither how much you spend nor how much you pay.
How to check
Ask for last month’s Google or Meta invoice. It shows the real spend and any gap with the report.
P3
How will we know which ads lead to orders, not just clicks?
Why you ask
A click is a cost. An order is revenue. A campaign optimised for clicks almost always spends more and returns less.
A good answer
Sets up tracking for the specific action before launching ads, and says explicitly that the first weeks are for gathering data.
A bad answer
Launches campaigns immediately. You will get a report of clicks and cost per click, because that is all the system knows without setup.
How to check
Ask them to show how many orders came from ads in the report, then compare with your actual orders for the same period.
P4
Who maintains the excluded searches, and how often do you review them?
Why you ask
In Google your ads also appear for searches you never chose. Without regular pruning you pay for people looking for something else. This is the quiet leak in almost every neglected campaign.
A good answer
Answers with a frequency and a person. Monthly is normal, weekly at larger budgets.
A bad answer
Does not understand the question, or says the system handles it. Partly it does. The rest is your money.
How to check
Ask for the list of actual searches your ads appeared for last month. Read twenty lines. It will be immediately obvious.
P5
What stays with me if we stop working together?
Why you ask
Campaigns, ad copy, creatives and collected audiences are an asset. Either it is yours, or it disappears the day you part.
A good answer
Everything is in your account, so the question is moot. That is precisely the right answer.
A bad answer
"We will export you a report." A report is not a campaign and cannot be restarted.
How to check
Check in the first month whether you can see the campaigns in your own account. Do not wait for the breakup to find out.
P6
How did you arrive at the proposed budget?
Why you ask
A budget should come from arithmetic: click prices in your industry and how many clicks an order takes, not from a round number.
A good answer
Shows the maths: expected cost per click, expected order rate, and how much must be spent before there are conclusions.
A bad answer
"Start with whatever you are comfortable with." Comfortable is often too little to learn anything, but enough for a fee.
How to check
Ask to see the calculation on paper. Three numbers and one logic. If they are missing, the budget is a wish.
P7
Who builds the page the clicks will land on, and is it in the price?
Why you ask
Ads bring people, but the page sells, and the best campaign on a bad page is an expensive way to prove it.
A good answer
Reviews the page before launch, says what on it must change and who will do it, with a clear boundary of responsibility.
A bad answer
Sends ads to whatever exists without looking at it. The weak result will later be blamed on "the market".
How to check
Open the page the ads will lead to on your phone and ask yourself one question: would I order from here.
P8
How many ad variants will you test, and who writes them?
Why you ask
Without variants there is no learning: one ad can be compared to nothing, and its text is half the result.
A good answer
Says how many variants launch, who writes the copy, and how it is decided which variant stays and which stops.
A bad answer
"The system optimises by itself." The system chooses among what it is given. Given one thing, there is no choice.
How to check
Ask to see the variants before launch. If they differ by one word, the test is a formality.
P9
Will I pay for clicks on my own name?
Why you ask
Clicks on your own brand are the cheapest and surest, and they easily fill a report with good numbers that would have arrived without any ads.
A good answer
Separates brand campaigns from generic ones and reports them separately, with an honest conversation about when defending the brand is worth it.
A bad answer
Everything sits in one campaign and the report shows one blended cost per order. The nice number hides where it comes from.
How to check
Ask for the results excluding brand searches. That number is the real cost of a new customer.
P10
What do I see in the account in real time, and what only in the report?
Why you ask
The difference between access and a report is the difference between control and trust on someone’s word.
A good answer
Full read access from day one, and a report that summarises rather than replaces the data.
A bad answer
"We will send you everything, the access would only confuse you." Confusion is preferable to blindness.
How to check
Log into the account on a random day and compare one number with the last report. The match is the test.
P11
How is cookie consent set up, and what do we lose without it?
Why you ask
Without properly configured consent, measurement loses part of the orders and the ads learn from incomplete data.
A good answer
Explains what is configured, how it affects the numbers, and where the data is legally incomplete, without hiding it.
A bad answer
"Those are technical details." Those details decide whether the numbers the budget is spent on are true.
How to check
Place a test order after refusing cookies. Then ask: is this order visible anywhere, and how.
P12
At what result do we stop or rebuild the campaign?
Why you ask
A campaign without a stop condition can spend for months on inertia, because nobody defined what "not working" means.
A good answer
Sets a threshold and a date in advance: if by this date the cost per order is above this number, we stop and rebuild.
A bad answer
"Let us give it a little more time." A sentence without a number or a date, repeatable forever.
How to check
Write the threshold and the date in an email after the first meeting. On that date, open it.
P13
How is the budget planned around my business’s season?
Why you ask
A flat monthly budget in a seasonal business means shortage in the strong months and waste in the weak ones.
A good answer
Asks when your strong months are and proposes a monthly distribution, with preparation before the peak, not during it.
A bad answer
The same budget every month, because that is easier to invoice.
How to check
Put last year’s sales by month next to the proposed budget. The two curves should resemble each other.
P14
What does the monthly report contain, and when do I receive it?
Why you ask
The report is the product you see most often, and if it is only charts without conclusions, you are paying for decorated data.
A good answer
Shows a sample report: what happened, why, what changes next month, and one recommendation that requires a decision from you.
A bad answer
An automatic export from the system, sent by email. Nobody looked at it before sending, and it shows.
How to check
Read the last report and look for one sentence that is not a description of a chart. If there is none, there is no analysis.
P15
Do you run Performance Max, and what inside it cannot be seen?
Why you ask
This campaign type delivers, but it hides where exactly the ads appear, and that should be said, not omitted.
A good answer
Explains honestly: part of the distribution is invisible, here is what we do to limit waste, and here is when this type is not suitable.
A bad answer
Presents it as magic without downsides. Opacity is only painless for the person not paying the budget.
How to check
Ask: what share of the budget goes to placements we cannot see. The honest answer starts with an admission.
P16
How many hours a month will you actually spend on my account, and on what?
Why you ask
The fee buys time and attention, and without hours behind it a small budget can receive five minutes a month.
A good answer
Gives an estimate in hours and breaks it down: search-term review, tests, reporting, and says when the account needs more.
A bad answer
Silence, or "as much as needed". Needed without a number tends toward zero with a busy provider.
How to check
Divide the monthly fee by the stated hours and compare with a normal hourly rate. The arithmetic shows whether the numbers are real.