I run an agency, so take this in the spirit it is written: most of what makes an agency good is invisible from the outside, and almost everything that makes an agency easy to sell is visible.
That is the whole problem. You are being asked to judge a service you cannot inspect, on evidence the seller chose, in a category where the difference between competent and incompetent is invisible for about ninety days and expensive by month six. The credentials do not help. Badges, awards and partner tiers are awarded for spend under management and certifications completed, not for whether the accounts work.
So the ten questions below are chosen for one property: they are hard to answer well without actually being good, and easy for you to check afterwards. Most of them cover who does the work, what happens when it goes wrong, and what you walk away with if it ends, rather than Google Ads itself.
I have also written what a good answer sounds like and what should worry you, because a question you cannot evaluate the answer to is not much use.
Key Takeaways
- Ask who is on the tools, not who is in the room. You are usually sold by the most experienced person in the business and handed to the least. Get a name, a title, and a weekly hour commitment in writing.
- Treat case studies as marketing, not evidence. Most published case studies are former clients, and the numbers are chosen after the fact. References you select yourself are worth more than any deck.
- "We have a process" means nothing. A real system has a name, stages, and monitoring. The most common cause of a failing account is work that quietly stopped being done.
- Agree what counts as a lead before anyone spends a dollar. Spam, warranty enquiries, existing customers and out-of-area calls all get counted as leads by somebody's report.
- Never let an agency run you inside their own ad account. You own the account, the conversion history, the website, the tracking numbers and the creative, or you are a hostage.
- A paid two to three week audit before a twelve month retainer is the single best de-risking move available, and a good agency will sometimes tell you at the end of it not to hire them yet.
- Ask what happens after two bad months. The answer should be a diagnostic process and a contractual remedy, not "we will optimise it".
- Below roughly $1,500 a month of media you cannot buy enough data to judge anything, and an agency willing to take a budget it knows cannot work has told you what it optimises for.
Before you ask anyone anything, check you are ready
Half of the failed agency relationships I have seen were failing before the agency was hired. Paid advertising costs money before it makes money, and it exposes whatever is already broken downstream.
The response time point is not a minor one. We have written up what it costs in the 60 second rule, and it is the cheapest improvement most businesses have available to them.
The ten questions
1. Who will actually run my account day to day, and how many hours a week will they spend on it?
You want a name, a job title, how many other accounts that person carries, and a weekly hour commitment written into the agreement. Ask to meet them during the sales process. That request is entirely reasonable and the reaction to it is informative on its own.
A good answer names the person, tells you their other client load, explains who reviews their work and how often, and offers to put you on a call with them before you sign.
Walk away from "our paid media team" with no names. That phrasing almost always means a junior runs the account and a senior glances at it monthly. Also be wary of heavy white labelling that is not disclosed. Outsourcing is not automatically bad, but you should know who is actually touching the account and be able to speak to them when something goes wrong, because something will.
2. What is your system called, and can you show me the diagram?
This is the question that separates an operation from a collection of good intentions, and almost nobody asks it.
Every agency will say they have a process. The word is free. A real system has a name, defined stages, someone monitoring that the stages happen, and a consequence when they do not. Ask what it is called. Ask to see it drawn. Ask how they know the work was done this week rather than skipped.
Here is why it matters more than strategy. Accounts usually decay because the account manager got busy, the search terms report did not get checked, negatives did not get added, tracking broke quietly, and performance slid for six weeks before anyone noticed. Drift is the default state of a managed account. A system is what resists it.
A good answer walks you through named stages, tells you what gets checked weekly versus monthly, and describes what happens internally when an account falls below their standard.
Walk away from "we have SOPs" with nothing behind it, or a process that turns out to be one person's habits.
3. Can you tell me about an account that did not work, and what you changed?
Anyone can produce a winner. The useful question is what they do when it goes wrong, and whether they have been wrong enough to have learned anything.
An agency that cannot talk openly about failure has usually not been tested. It also tells you something about how the relationship will go the first time a month goes badly, because it will.
A good answer is specific and slightly uncomfortable. A real account, what they assumed, what the data said, what they changed, and whether it recovered or ended.
Walk away from a perfect record. Nobody has one.
4. What do I own if we part ways?
Have this answered before you sign, in writing, not after.
You want to own the Google Ads account itself and its entire conversion history, the website and wherever it is hosted, the Google Business Profile, the analytics property, the landing pages, the creative, the tracking numbers and the leads. The agency should hold manager-level access to your account, not ownership of it.
| Asset | Who should own it | Why it matters |
|---|---|---|
| Google Ads account | You, with the agency on manager access | Conversion history is what smart bidding learns from. Starting a new account restarts learning and costs months |
| Website and hosting | You | A site on a proprietary builder only that agency can edit cannot follow you anywhere |
| Analytics property | You | Without history you cannot tell whether the next provider improved anything |
| Landing pages | You | Often built outside your site and quietly retained |
| Call tracking numbers | You | Numbers you have advertised for a year are not the agency's to keep |
| Creative and ad copy | You | You paid for it |
| Lead data and CRM records | You | Obvious, and routinely not the case |
Walk away from any arrangement where campaigns run inside the agency's own ad account, or where the website sits on a builder nobody else can work in. Both are hostage structures, whatever the stated reason. A professional agency has offboarded clients before and has a process for handing everything over, because good relationships still end, sometimes because you succeeded.
5. How do we define a lead, and how will you handle the ones that are not?
Agree this before anyone spends anything, because otherwise you will have the argument in month three with no shared definition to settle it.
A report saying thirty leads can contain spam form fills, warranty enquiries, existing customers asking a service question, people outside your service area, recruiters, other agencies prospecting you, and jobs you do not want. Some of those are real people the campaign was correctly told to find. That is the uncomfortable part: junk leads are usually a targeting instruction, not a platform fault, and we went through the specific causes in why your Google Ads leads are junk.
A good answer defines a qualified lead in terms of your business, not the platform: someone who needs the service you want more of, in your area, at a size you will take. It covers how they will exclude the rest with negative keywords, match types and qualification on the page, and how they expect feedback from you about which enquiries were any good.
Walk away from any definition that is simply "a form fill or a phone call", and from anyone promising a guaranteed number of leads per month. Ask exactly how a guarantee would be honoured. Sometimes the answer is shared or resold leads from a vendor list, which is a different product entirely and one we have argued against in stop buying leads from vendors.
6. What will you report, and can I see a real sample?
Ask for an actual weekly and monthly report from a current client with the name removed. This is the single most revealing artefact in the whole process, because a report is the finished output of whatever the agency actually does all month.
| What a weak report shows | What a useful report shows |
|---|---|
| Impressions, clicks, click-through rate | Spend against target, and pacing for the month |
| Conversions with no definition | Qualified enquiries against a target, with the definition attached |
| A screenshot of the Google Ads overview tab | Cost per enquiry and, where possible, cost per signed customer |
| "Optimisations made: 47" | Two to four experiments run, what each tested, and the outcome |
| No mention of anything that went wrong | Flagged issues, who owns each one, and the next action |
| Same commentary every month | What to do next month and why, including a budget recommendation |
The number that settles whether any of this paid for itself is cost per signed customer, not cost per lead. Very few agencies report it, partly because it requires your cooperation to get the outcome data back. If they raise that themselves, it is a good sign. We built PixelRush HQ precisely because that loop is where most reporting breaks.
Walk away from a template that is a copy and paste of Google Ads' own default report. That is a printout rather than analysis.
7. What happens if the account underperforms for two months running?
The answer should be a process, not a reassurance.
A good answer describes a diagnostic sequence: what they check first, how they form a hypothesis, when it escalates to a senior strategist, and what the contract says about sustained underperformance. It also describes what they would expect from you in that period.
Walk away from "we will optimise it" or "we will figure it out". Also be wary of an agency whose answer is to make lots of changes, because changing several things at once in a smart bidding account is the reliable way to ensure nobody ever learns which change mattered. That is one of the more expensive habits in the industry and we covered it in costly Google Ads mistakes.
8. How do you charge, is media separate, and what is your minimum?
Get two numbers, never one. A bundled quote of "$3,000 a month all in" can mean $1,500 of media and $1,500 of fee, or it can mean $2,700 of media and $300 of attention, and those are completely different products.
The three models are a flat retainer, a percentage of spend, usually 10% to 20% once spend is meaningful, or a hybrid with a floor. All three can be fair, provided the model is stated rather than negotiated case by case, and that you know what happens to the fee if you cut spend by half.
On the minimum: below roughly $1,500 a month of media you cannot generate enough clicks in most industries to make a decision inside a reasonable window, and in expensive categories like personal injury or criminal law that figure is not close. At $33 a click, $1,500 buys about forty-five clicks. The full picture by industry is in how much Google Ads costs, and the fee bands are in Google Ads management costs.
A good answer includes a minimum, and a willingness to tell you your budget is too small rather than take it anyway.
Walk away from a fee so low it cannot buy meaningful attention. A $350 a month management fee, after the agency's own margin, funds roughly three hours on your account. Three hours a month buys maintenance rather than management, and small budgets are harder to manage than large ones rather than easier, for reasons we set out in why a small budget is harder.
9. Can I speak to two current clients and two former clients?
Current clients tell you whether the pitch matches the delivery. Former clients tell you how the agency behaves when things end, which is the part you will never see in a case study.
That second half is the question most agencies flinch at, so it is worth asking. You are not looking for a bitter ex-client. You are looking for whether the handover was clean, whether the assets came across, and whether the relationship ended honestly.
Be sceptical of published case studies generally. They are carefully constructed and the numbers are selected after the outcome is known. A genuine example from a well-known software company claimed a 300% conversion rate improvement for a client, and the improvement turned out to be a single day inside a six month engagement, with everything flat either side. Most agencies also only publish clients they no longer work with, for the sensible reason that they do not want competitors targeting their best accounts. So what you are reading is usually a highlight from a past relationship, not evidence of current work.
Our case studies have the same limitation and you should treat them the same way. Ask us for references instead.
10. What is your internal standard for a winning account, and what share of your clients are hitting it?
This one catches people off guard, and the reaction is the answer.
You are asking whether the agency measures itself. Not whether clients are happy, which is a customer service metric, but whether they have a defined benchmark for what success looks like in an account and know what proportion of their book is currently meeting it.
A good answer gives you a number, admits it is not 100%, and describes what happens internally when an account falls below the line.
Walk away from "most of our clients are happy" or a blank look. An agency with no internal standard has no mechanism for noticing that your account has stopped working.
Key Point
Ask question 10 of every agency you shortlist and compare the answers side by side. The distance between "most clients are happy" and "about eight in ten are hitting our benchmark, here is what we do with the other two" is the distance between buying activity and buying outcomes.
The single best thing you can do before signing
Buy a small piece of work before you buy a big one.
A paid audit or discovery sprint, two to three weeks, priced somewhere between a few hundred and a few thousand dollars depending on the account, should produce a full account audit, a prioritised list of fixes with estimated impact, a ninety day plan with milestones, a named senior person assigned to you, and a recommendation on whether to proceed.
It costs a fraction of a twelve month retainer and it tells you more than every sales call combined, because you are buying the thing itself rather than a description of it. You see how they think, how they write, whether they found things you did not know about, and whether they are willing to tell you something you did not want to hear. The strongest signal available is a paid audit that concludes you should fix something yourself first, or wait, or spend the money elsewhere.
It also inverts the relationship. In a pitch, they are selling. In a paid audit, they are working, and you are watching them work.
Red flags, collected
| Red flag | What it usually means |
|---|---|
| Guaranteed leads, guaranteed rankings, guaranteed results | Either meaningless, or fulfilled with resold and recycled leads |
| Everything compressed into one call and a proposal at the end | Optimised for closing, not for fit |
| Campaigns run inside the agency's ad account | You cannot leave without losing your conversion history |
| Website on a proprietary builder | You cannot leave without rebuilding your site |
| No named person on the account | A junior runs it and a senior reviews it occasionally |
| Reporting is the Google Ads overview tab | No analysis is happening |
| A perfect track record | You are being shown a selection, not a record |
| Fee far below everyone else | Fewer hours, and shortcuts to keep you quiet |
| Will not introduce former clients | The endings are not clean |
| Takes your budget without comment when it is clearly too small | They optimise for signing you, not for your result |
| Long lock-in with no performance terms | The contract is doing the work the results should do |
On contracts specifically: a twelve month term is not automatically a red flag, because some work genuinely takes that long to pay back. A twelve month term with no defined deliverables, no underperformance clause and no exit process is. Month to month sounds safer and sometimes is, but it also encourages short-term thinking on both sides. Read what the agreement says happens when it is not working.
What a good agency will ask you
The reverse test, and it is a fast one. If you get through an hour and they have not asked most of these, they are selling a package rather than solving a problem.
A related point worth sitting with: when a campaign underperforms, the cause is often downstream of the ads entirely, in how enquiries are answered and followed up. That is an uncomfortable conversation for an agency to open with a prospect, and the ones willing to have it early are usually the ones worth hiring. We made that argument at length in stop blaming your marketing agency.
A scoring sheet for three agencies
Talk to three. Not one, and not seven. Three is enough to calibrate and few enough that you actually complete the process.
Score each question out of three as you go, straight after each call while it is fresh: 3 for a specific answer with evidence, 2 for a reasonable answer with no evidence, 1 for a vague answer, 0 for a deflection or a red flag.
| Question | What a 3 looks like |
|---|---|
| 1. Who runs the account | Named person, other account load disclosed, offered on a call before signing |
| 2. What is your system called | Named, drawn, with monitoring and a stated consequence for drift |
| 3. An account that did not work | Specific, slightly uncomfortable, with what changed |
| 4. What do I own | Everything, in writing, with an offboarding process they have used before |
| 5. Definition of a lead | Defined in your terms, with a plan for excluding the rest and a feedback loop |
| 6. Reporting | A real sample report with targets, experiments and flagged issues |
| 7. Two bad months | A diagnostic sequence, escalation, and a contract term |
| 8. Pricing and minimum | Media and fee separated, a stated model, a stated minimum |
| 9. References | Two current and two former clients offered without hesitation |
| 10. Internal benchmark | A real number below 100%, and what they do about the rest |
Anything at 24 or above out of 30 is a serious candidate. Below about 18, keep looking. A zero on question 4 or question 8 should end the conversation regardless of the total, because those are the two that are expensive to discover later.
Then, finally, use your judgement. The scoring sheet exists to stop you being persuaded by whoever presents best, not to replace the read you get on whether these are people you want to talk to every month for a year. Clear explanation without jargon is a real signal, because an agency that cannot explain its approach simply usually cannot explain its results either.
If you want to run these questions past us, get in touch. We would rather be scored than pitch, and if the answer is that your budget is not ready or the work belongs somewhere other than Google Ads, we will say so.
Want us to implement these strategies for you?
Book a free strategy call and let's discuss how we can grow your business.
Book Your Free CallContinue reading
More on Paid Ads
Is Google Ads Worth It? What $7 Million of Spend Says, and When It Isn't
Google Ads is worth it when a client is worth more than it costs to buy one, and most businesses asking the question cannot fill in either number. What a click, an enquiry and a client cost by industry across $1.65 million of recent Australian spend, the break-even maths with three worked examples, the five situations where it is not worth running, what $10 a day actually buys, and how to know in 30 days.
7 Lessons From Optimising 400+ Google Ads Landing Pages (With the Before and Afters)
What four hundred landing page rebuilds taught us about why paid traffic does not convert, with the real pages, the before and afters, and the results: 3% to 8% conversion, 60% more direct bookings, 150% more organic conversions.
Google Ads Management Costs in Australia, and What the Fee Actually Buys
What Australian agencies and freelancers really charge to manage Google Ads, the three pricing models compared, how many hours a fee actually funds, and how to tell whether you are overpaying.
Costly Google Ads Mistakes Small Businesses Make in 2026 (and How to Audit for Them)
The mistakes that actually drain small business Google Ads budgets, what each one costs, how to check for it in your own account, and the AI prompts that let a non-specialist run the same checks accurately.



