Choosing the right Google Ads management for an online shop

Provider type is the second question

Agency or freelancer is the second question for me, because the same provider can be exactly right for one shop and money thrown away for the next, depending on what is actually missing. A shop without reliable margin figures will only be steered faster in the wrong direction by better campaign maintenance. A shop with clean measurement and an overstretched head of marketing, on the other hand, mainly needs hands on the account.

Before you read on, factor in my own interest. I sell Google Ads management myself: self-employed, focused on e-commerce in the DACH region (Germany, Austria and Switzerland), and I work on the accounts myself. A text from me will tend to look kindly on individual providers. That is why the checks below are built so that I can fail them too, and I say when someone like me is the wrong choice. I know of no independent data on which model performs better on average.

Which bottleneck leads to which model?

Name the bottleneck first, and back it with evidence rather than a feeling. The mapping in this table is my assessment.

Two rules help when several rows apply. Measurement always comes first, because otherwise every other improvement is judged on wrong numbers. And if what you mainly expect is cover for absences, or shop technology, creatives and channels from a single source, you should not hire an individual like me. With an individual provider you buy direct access to the person working on the account, along with the risk that this one person is unavailable at some point. The details of this trade-off are in agency or freelancer.

What does the management have to earn back?

A fee is only cheap if it brings in additional contribution margin. The check is a formula:

Required revenue growth = fee / (revenue from Google Ads × contribution margin rate before ad spend)

It applies with ad spend held constant. For the rate you need the same figures as for the break-even ROAS. A worked example with round, assumed values, once with a larger and once with a smaller budget:

Shop A has to grow by a little more than a tenth, which is a realistic brief. Shop B would have to lift its ROAS by almost 30% just to pay the fee. Savings count as well: ad spend that can be cut without losing revenue goes straight to the bottom line. To get there, though, A would have to cut 20% of its budget and B half of it, without revenue falling.

My rule of thumb, not a measured threshold: if the hurdle is above roughly 25%, do not buy full management. Buy a smaller scope instead, such as monthly sparring, for Shop B under €1,300, or a fixed-price audit, and implement in-house. For context, a German agency gives €800 to €3,000 a month excluding media budget as typical for small and medium-sized businesses.

The rule does not apply if the account demonstrably has major faults, such as broken tracking, because then the hurdle is cleared quickly. During a time-limited build-up phase, calculate with the target budget, provided the money for it is planned. The hurdle comes out too high if nobody maintains the account today and the baseline would fall anyway, or if repeat purchases make up much of customer value, because the formula only sees the first order.

The fee model shifts your budget ceiling

How the provider is paid shapes which budget recommendation is convenient for them. A German agency treats a fee calculated as a percentage of ad spend as a warning sign, because it creates an incentive to increase spend. I would not dismiss it out of hand, but I would do the maths. At a 35% contribution margin, an additional advertising euro pays off from a marginal ROAS of 2.86, that is 1 / 0.35. Marginal ROAS means the revenue that this specific additional euro brings in, not the account average. If a 15% fee is added to every euro, the threshold moves to 3.29, that is 1.15 / 0.35.

Suppose Shop A increases ad spend from €10,000 to €15,000, and the additional €5,000 achieves a marginal ROAS of 3:

The account's average ROAS afterwards is 4.33 and looks respectable in the report, revenue is up 30%, the provider earns more, and the shop loses money. Nobody has to act dishonestly for this to happen. The figures simply do not reveal the gap by themselves.

Every other model has its own built-in flaw too. With a flat fee, the provider earns most by doing little. With a share of reported revenue, brand revenue becomes attractive, even though it might have come without the ad. A percentage can therefore work if the budget decision demonstrably stays with you and marginal ROAS is reported alongside. With a flat fee, you need to be able to see the work that was done.

Two tests before you sign

Test 1: the target. Give the provider your contribution margin rate and ask what target they derive from it and when they would advise you to reduce the budget. A useful answer names conditions, for example a marginal ROAS below break-even or products that do not cover their costs. An answer that only talks about ROAS benchmarks or more reach is of no use. In the same conversation, ask who works on the account and how many accounts that person looks after. One calculation makes the conversation concrete: 100 accounts in a 40-hour week leave 24 minutes per account per week.

Test 2: the exit. Ask what happens if you give notice in six months. The account should be in your name, with your admin access, and the provider is only linked to it. According to Google, linking a manager account does not automatically make it the administrative owner. If it is the owner, however, its admins can invite and remove other admins in the client account, transfer ownership and unlink other manager accounts. Those are rights you want to keep in your own hands.

Paying the invoice is not a reliable way to secure this. One consultant reports disputes in which the party paying did not automatically get its account back. That is a single observation, but clarifying it before the contract costs you no more than an email. I would apply the same rule to Merchant Center, Tag Manager and Analytics. The process is described in detail in who owns the Google Ads account.

What the Partner badge measures

The badge shows volume and formalities, not profit for the retailer. At its core, Google requires three things: an optimisation score of at least 70% in the manager account, $10,000 of ad spend in 90 days across all managed accounts, and certifications for at least 50% of account strategists. So the person on your account does not have to be certified. According to Google, the optimisation score is an estimate, and it rises when recommendations are applied.

Premier Partner status goes each year to roughly the top 3% per country, measured partly by growth in ad spend. That is Google's view of a sales partner. Contribution margin appears in none of these requirements.

I am a Google Partner myself and treat the badge as an entry signal. It carries little weight in the selection, and its absence says little about a specialised provider with a small portfolio. More on this in what Partner status tells you.

Which warning signs can you spot without account access?

Google supplies a checklist itself. The third-party policies apply to everyone who buys or manages Google advertising for clients, and they tell you what you can insist on:

• Guaranteeing top ad placement, or putting heavy pressure on you to sign or renew, is explicitly not allowed.
• A management fee must be disclosed to you in writing before the first purchase and shown on the invoice.
• If the provider reports Google costs, they must state the exact amount charged by Google, without their own fees.
• On request, they must give you the customer ID of your account.
• Every advertiser they manage needs a separate account.

On top of that come signals I would take seriously from a retailer's point of view, even though they break no rule:

• There is a ROAS forecast before anyone has looked at the account.
• The person selling is not the person who will do the work, and that person is not named even when you ask.
• The report does not separate brand from non-brand and has no notion of contribution margin.
• A high optimisation score is presented as proof of performance.
• There is no arrangement for handing over the account when the contract ends.

A breach from the first list is enough for me to say no. From the second list, a single point is no reason to rule a provider out, but with several I would stop negotiating.

How the first 90 days should run

The start determines what the management will later be measured against, so it belongs in the contract or at least in a written agreement. This sequence is my suggestion, adjust it to your size:

1. Before the start: settle ownership and admin access, record the fee and notice terms including handover, and agree on a named person for the account.
2. Weeks 1 to 2: reconcile conversion data with the shop system and record the baseline from your own figures, not from the Google report alone.
3. Weeks 3 to 4: an inventory with a prioritised list and a target based on contribution margin. You sign off the target.
4. Months 2 and 3: implementation in the agreed order, with a short monthly report that also says what did not work.
5. Day 90: a meeting with three questions. Is the measurement right now? What has happened to contribution margin? Does continuing pay off according to the calculation above?

Do not expect final proof after 90 days, because seasonality and demand overlay the result. What you can expect is reliable measurement, a target you can follow and an honest list.

You can check the work done yourself, provided you have access. Change history lists changes from the past two years and, for changes made through the interface, shows the person's email address. Use it as a basis for conversation, not as a tally: many changes prove nothing about quality, and in an account that runs well, restraint can be the right call. Several weeks without a single entry and without an explanation still warrant a question. You should put me through this check as well.