Choosing a Google Ads agency looks like a reversible decision. It rarely is. By the time you realize the current team is underperforming, you have usually burned two or three quarters of budget, trained Google's bidding algorithms on bad conversion data, and accumulated an account structure the next team will spend months untangling. The switching cost is not the contract. It is the compounding damage.
The frustrating part is that most agencies look identical in a pitch. Same slides about "full-funnel strategy," same logos, same promise of a dedicated account manager. The differences that matter only surface once you know what to ask: how they structure accounts, how they manage your money week to week, who owns your data, and how they measure.
This guide is the evaluation process we would want a prospect to run on us. It covers the types of agencies you will meet, the four dimensions that actually separate strong operators from mediocre ones, the pitch questions that expose weak agencies fast, how engagement models shape incentives, and the red flags that should end a conversation on the spot.
The four types of Google Ads agency (and who each suits)
Before evaluating anyone, know what category you are talking to. Each type has a legitimate use case, and most bad matches happen because a buyer picked the wrong category, not the wrong firm within it.
Freelancers and micro-agencies
One to three people, usually ex-agency or ex-in-house. At their best, you get senior hands directly on your account with no account-management layer in between. The risks are coverage and ceiling: holidays, illness, and churn hit you directly, and a solo operator rarely has deep expertise across Search, Shopping, Performance Max, YouTube, and measurement simultaneously. Sensible for small accounts with one clear objective.
Specialist PPC shops
Agencies that do paid search and little else. The good ones are genuinely excellent at the craft. The structural limitation is that Google Ads performance increasingly depends on things outside Google Ads: landing page conversion rates, first-party data quality, feed quality for Shopping, creative for Demand Gen and YouTube. A pure PPC shop will identify those bottlenecks but cannot fix them for you.
Full-service growth agencies
Teams that run paid media alongside SEO, CRO, data, and CRM. The advantage is that the levers interact: search query data informs SEO, CRO fixes raise the conversion rates that Smart Bidding feeds on, server-side tracking improves every channel at once. The thing to verify is depth: some full-service agencies are five shallow practices in a trench coat. Ask to meet the actual paid search lead, not the growth strategist who fronts the pitch.
Holding-company networks
Large networks suit large advertisers with procurement requirements, many markets, and complex governance. If you spend at that scale, you already know it. If you do not, the overhead is not worth it, and your account will be staffed accordingly.
Your business model matters as much as agency category. Lead generation and e-commerce are different disciplines with different failure modes, and the playbooks differ enough that this blog covers the B2B PPC agency and e-commerce PPC agency cases separately.
How to evaluate a Google Ads agency: the four-part grid
Once you have shortlisted two or three agencies of the right type, evaluate them on four dimensions. These are the ones that predict performance a year in, and conveniently, they are also the hardest to fake in a pitch.
1. Account structure philosophy
Ask every candidate to describe how they would structure your account, and why. There is no single correct answer, but there are correct reasoning patterns.
Modern Google Ads rewards consolidation: fewer campaigns with more conversion data each, so Smart Bidding has enough signal to work with. The old approach of fragmenting accounts into hundreds of single-keyword ad groups actively fights the algorithm now. A strong agency will talk about consolidation, signal density, and where they still deliberately break things out: brand versus non-brand, markets with different margins, product lines with different targets. A weak agency will either propose granularity for its own sake (it looks like work) or propose "one Performance Max campaign and trust Google" (it is barely work at all).
Listen for how they talk about Performance Max specifically. It is a genuinely useful campaign type and a genuinely convenient hiding place. Strong operators use it with structural guardrails (brand exclusions, feed segmentation, dedicated assets) and can explain what they monitor to know whether it is cannibalizing cheaper traffic. Weak operators treat its opacity as a feature.
Structure is not an academic concern. When we took over the search account for Adam & Ève, a Dutch e-commerce brand growing in the French market, the work started with rebuilding the entire search structure from scratch. That restructure cut the cost per acquisition by 10% while the business grew 1.8x year over year; the full case study walks through it. The point is not that restructures are always the answer. It is that an agency should be able to look at your account and tell you, specifically, what its structure costs you.
2. Budget management discipline
This is the least glamorous dimension and the one where mediocre agencies quietly bleed money. You are evaluating whether the agency treats your budget like a portfolio or like a bucket.
Concrete things to probe:
Pacing. How do they track spend against the monthly budget, and how often? Weekly check-ins are not enough on accounts of any size; you want daily or automated pacing with alerts.
Reallocation. What triggers moving budget between campaigns? A strong answer references marginal performance — the next unit of spend goes where it returns most — not "we review it monthly."
Search term hygiene. Who reviews search terms, how often, and what has that produced recently? Broad match plus Smart Bidding is powerful, but only under active negative-keyword management. An agency that cannot show you recent negative-keyword work is not doing it.
Seasonality and cash constraints. Can they adapt when your finance team cuts the budget mid-quarter, or when stock runs out on a hero product? Ask for an example of a time they spent less than the client authorized because the incremental return was not there.
That last behaviour — recommending less spend — is the clearest available signal of discipline, especially from agencies paid as a percentage of spend. It is also where good agencies produce results that look like the ones from our work with Amelis, a home-care services company: the cost per lead was cut in half while lead volume grew by 160%, with 340,000 new users brought to the site. Numbers like that come from hundreds of small allocation decisions, not from a clever launch. The Amelis case is worth reading if lead generation is your model.
3. Transparency on fees and data ownership
Two non-negotiables, and both are binary.
First: you own the Google Ads account. The account lives in your billing profile, the agency accesses it through their manager account, and if you part ways, they unlink and you keep everything: history, conversion data, audience lists, the learning that Smart Bidding has accumulated. That history is a real asset. An agency that runs your spend inside an account they own is holding your performance hostage, whatever they say about convenience. Google's own documentation on manager accounts describes exactly how agency access should work (support.google.com/google-ads); there is no technical reason for an agency to own your account.
Second: you can see the fee. Whatever the model, the agency's compensation should be a line you can point to, separate from media spend. Be wary of bundled arrangements where fee and media blur, and of any reluctance to show the actual Google Ads billing. The same logic extends to data: your analytics property, your conversion data, your audience lists, your creative. All of it should be in accounts you control.
None of this is exotic. Reasonable agencies agree to all of it without friction, which is precisely why hesitation is informative.
4. Measurement setup
Ask each candidate what they would do in the first thirty days, before scaling anything. The right answer is heavily weighted toward measurement, because every bidding decision Google makes is only as good as the conversion data feeding it.
A serious agency will want to audit and usually rebuild some of: conversion tracking and its deduplication, enhanced conversions, consent handling for the markets you operate in, the import of offline or CRM outcomes so bidding optimizes toward qualified revenue rather than raw form-fills, and the difference between what Google Ads claims and what your business actually banked. For lead-gen businesses, feeding lead quality back into the account is frequently the single highest-leverage project available, because it redirects the entire bidding system toward leads that close.
A weak agency will skip this and start "optimizing" on day three. It feels fast. It means every subsequent decision is built on whatever tracking debris the last team left behind.
Pitch questions that expose weak agencies
You do not need to be a PPC expert to run a good evaluation. You need questions whose answers are hard to fake. These work:
"Walk me through an account you inherited. What did you change and why?" Strong agencies tell specific stories with trade-offs. Weak ones give you a generic audit checklist.
"When did you last recommend a client cut their budget?" If the answer is never, their incentives are running the account.
"What would make you tell us Google Ads is the wrong channel for part of this plan?" An agency with no answer sells Google Ads regardless of fit.
"Who exactly will work on our account, and how many accounts does that person handle?" Meet the operator, not the pitch team. Vague answers mean junior staffing behind a senior facade.
"How will you measure incrementality, not just what the platform reports?" You are listening for an honest account of attribution's limits (holdouts, geo tests, blended metrics as sanity checks), not a promise that the dashboard is truth.
"Show me a case where results went backwards. What did you do?" Everyone has one. An agency that claims otherwise is curating too hard to be trusted.
"What do you need from us to succeed?" Good agencies name real demands: access to margin data, fast landing-page changes, sales feedback loops. An agency that needs nothing from you plans to do nothing structural.
Then verify against evidence. Ask for case studies in your business model and check whether they name real numbers on real problems. We publish ours, more than two hundred of them, across our references, precisely because claims without receipts are the industry's default setting.
Engagement models: how the fee shapes the incentive
Do not fixate on which model is cheapest. Fixate on what each model pays the agency to do, because over a year, incentives beat intentions.
Percentage of spend. The classic model, and administratively simple. Its known flaw: the agency earns more when you spend more, which taints every scaling recommendation. It can work well when paired with the discipline signals above: an agency on percentage-of-spend that has demonstrably told clients to spend less has earned some trust.
Flat fee. Decouples the agency's revenue from your media budget, which cleans up the scaling conversation. The failure mode is effort: once the fee is fixed, a complacent agency does the minimum to retain the account. Flat fee works when the scope is explicit and reviewed.
Performance-based. Superficially attractive: they only win when you win. In practice, the agency optimizes hard toward the exact metric in the contract, including by claiming credit for conversions that would have happened anyway. Brand traffic is the classic vehicle. If you use this model, define the metric with extreme care and audit attribution yourself.
Hybrids. A base fee plus a performance component is often the most honest structure: the base funds real work, the variable aligns direction. The details matter more than the label.
Across all models, insist on notice periods short enough to keep everyone honest. Long lock-ins protect agencies from the consequences of underperformance, and agencies confident in their work do not need them.
Red flags that should end the conversation
Some signals are not weaknesses to weigh. They are exits to take.
They own the account, or the account "stays with the agency" if you leave. This is the single most damaging arrangement in the industry. You lose your history, your conversion data, and your negotiating position simultaneously.
Guaranteed results. Nobody controls auction dynamics, competitors, or your conversion rate from a pitch deck. A guarantee is either a lie or a metric defined so narrowly it is meaningless.
Set-and-forget behind a management veneer. Ask to see the change history of an account they run. Google Ads logs every edit. An account "managed" with a handful of changes a month, framed as "letting the algorithm learn," is an account being billed, not managed.
Reporting that leads with impressions and clicks. Vanity-metric reporting is how underperformance hides. You want reports built on cost per acquisition, revenue, and progress against targets you set together.
No questions about your margins or sales process. An agency that can run your account without knowing your unit economics is optimizing numbers, not your business.
Certification badges as the proof. Partner badges reflect spend thresholds and product adoption as much as skill. Treat them as table stakes, never as the argument.
A pitch that promises to rebuild everything on day one, sight unseen. Confidence before diagnosis is a sales tactic. The honest version of ambition is "here is what we would audit first."
Make the choice with a process, not a gut feeling
A structured evaluation — right category, four-dimension grid, hard questions, incentive-aware contract — takes perhaps two extra weeks. Against the cost of a year with the wrong agency, it is the cheapest insurance you can buy.
If you want to see how we run paid accounts, our Google Ads and paid media team works exactly the way this guide recommends: your account, your data, visible fees, measurement first. And if you would rather pressure-test your shortlist than start one, talk to our team: we will tell you what we would ask.
Frequently asked questions
How long should I give a new Google Ads agency before judging results?
Judge process immediately and performance progressively. Within the first month you should see measurement work, a structural diagnosis, and a plan with dates. Meaningful performance comparison usually needs a full quarter, longer for long sales cycles, because bidding systems need conversion volume to stabilize. An agency that shows spectacular results in week two is usually harvesting brand traffic.
Should I run Google Ads in-house instead of hiring an agency?
In-house makes sense when the channel is core enough to justify senior full-time talent and you can keep that person challenged and current. The honest trade-off: an in-house marketer knows your business deeply but sees one account; an agency team sees patterns across dozens. Many strong setups are hybrid: an internal owner for strategy and context, an agency for execution depth and cross-account pattern recognition.
What access should I give an agency, and what should I keep?
Give working access, keep ownership. The agency links to your Google Ads account through their manager account with standard access; billing and admin remain yours. The same principle applies to analytics, tag management, and your product feed: agency users inside properties you own. Any request to move assets into agency-owned accounts should be refused, politely and permanently.
Is a Google Ads agency worth it for a small budget?
Sometimes not, and a good agency will say so. Below a certain scale, the fee consumes too much of the total for the math to work, and a well-configured simple setup run in-house can be the better answer. The threshold depends on your economics rather than a universal number. If an agency takes a very small engagement without hesitation and without discussing this trade-off, that is itself a signal worth reading.
Founder and CEO of Junto
Founder & CEO of Junto, Étienne has been an entrepreneur and digital marketing consultant for over 15 years. An expert in Paid Media, SEO, Data, Automation, AI, Growth and Performance, he helps ambitious companies build high-impact growth strategies — generating lasting results and helping brands move forward in a constantly evolving digital environment.
