Why I left the Google Ads Partner programme, and why you might want to as well

I let the badge expire, and it was a choice

For years I was a certified Google Ads Partner. Not because I wanted the badge, but because I thought it gave my clients confidence. A label from the biggest name in digital advertising seemed to say expertise, accountability and a seal of quality.

Then I let it expire. Not for a lack of ad spend, and not for a lack of expertise. I let it go because staying a Google Ads Partner increasingly came at a cost, and the client was the one paying it.

That is a strong thing to say about a programme many good people belong to, so this article does it properly. I set out what the programme actually asks for today, what the optimisation score measures in Google's own words, how recommendations and auto apply work, and then the one argument at the centre of it all: the badge rewards a score, the score rises when you accept Google's recommendations, and a recommendation can raise the score while it lowers the client's profit. I show that last part with numbers, every step. I also say plainly what the badge does give you, because it does give you something.

The businesses and figures in the worked examples are invented so that every step can be shown. Nobody real, no client. The figures about the programme come from Google's own help pages, read in November 2026 and cited at the foot.

What the badge asks for, in Google's words

Google runs the Google Partners programme for agencies and other companies that manage Google Ads accounts for someone else. It has three tiers: Member, Partner and Premier Partner.¹ Anyone who signs up with a manager account is a Member. The Partner badge, the one you see on agency websites, has three requirements, and Google checks them every day.²

The first is performance. The company's registered manager account (the umbrella account an agency uses to run all its clients' accounts) needs an optimisation score of at least 70%. Google adds, in the same sentence, that you are in control, with the ability to apply or dismiss recommendations based on your assessment and your clients' goals.²

The second is spend. The manager account needs US$10,000 of ad spend across the accounts it manages over 90 days.²

The third is certification. At least half of the company's account strategists must hold a Google Ads certification, counted up to 100 people, and there must be at least one certification in each product area (Search, Display, Video, Shopping or Apps) where the accounts spend US$500 or more in 90 days.²

Premier Partner sits on top. A company has to meet all the Partner requirements and also rank in the top 3% of participating companies in its country, so for an Irish agency the top 3% in Ireland, assessed once a year. Google lists the factors that decide the ranking, and says the list is not exhaustive: growth in existing clients' ad spend, ad spend from new clients, client retention, the share of spend outside Search, and total annual ad spend.²

Read that last list slowly. Four of the five factors are measured in ad spend. The fifth, retention, is measured as the share of clients who are still spending a year later.² Nothing on the list measures whether those clients made money.

Worked example: a Galway agency checks its eligibility

An invented agency in Galway runs 12 client accounts from one manager account. Over the last 90 days they spent the equivalent of US$41,000 between them, so the spend requirement is met four times over.

It has six account strategists. Four hold a Google Ads certification, which is 67%, above the 50% line. Two clients run Shopping campaigns that spent more than US$500 in the period, and one of the four certified strategists holds the Shopping certification, so that box is ticked too.

That leaves the score. The manager account shows 66%. The agency is four points short of the badge, and every one of those points sits on the recommendations page. Keep that picture in mind. The rest of the article is about what happens on that page.

What optimisation score actually measures

Google defines optimisation score as "an estimate of how well your Google Ads account is set to perform". It runs from 0% to 100%, where 100% means the account "can perform at its full potential".³

It is calculated in real time from the statistics, settings and status of the account and its campaigns, the impact of the recommendations currently on offer, and the recent history of recommendations. Every recommendation carries a score uplift, from less than 0.1% to 100%, and applying or dismissing a recommendation changes the score.³ Google gives the example plainly: apply a recommendation with an uplift of 5% and the account's score goes up by 5%. An account reaches 100% by applying or dismissing all of them.³

The score has a focus, and Google takes that focus from the bid strategy. If a campaign runs on Target ROAS (return on ad spend), the score is "geared towards driving more conversion value at a similar ROAS, like expanding targeting and ensuring the budget isn't constraining additional high value conversions".³ If the campaign is not on a fully automated Smart Bidding strategy, Google derives the goal from the Smart Bidding strategy it would recommend.³

So the score is honest about what it is. It measures how closely an account matches Google's model of a well set up account, with the goal expressed as conversions or conversion value. It is not your profit, your margin, your stock position or your cash flow. Google does not claim otherwise. Its recommendations page says outright that it does not predict whether your ads will do well, it gives an idea of how each recommendation might improve performance, based on past data.⁴

The old version of this article said the score drops the fewer recommendations you follow. That was too simple and I have corrected it. Dismissing a recommendation also moves the score, and a fully dismissed list reaches 100% just as a fully applied one does.³ I come back to why that does not settle the matter.

Worked example: where the points come from

A Waterford stationery shop runs one Search campaign on Target ROAS. The recommendations page shows the account at 62% with three recommendations. These uplift figures are the example's assumptions, chosen to sum neatly.

The first is "Add broad match keywords", uplift 9 points. The second is "Raise your budgets", uplift 6 points. The third is a repair to a broken sitelink, uplift 2 points.

Apply all three and the score goes from 62% to $62 + 9 + 6 + 2 = 79\%$. Dismiss all three and Google records them as dismissed, and the page shows how much of the score is attributed to dismissed recommendations.³ Either route clears 70%. The sitelink repair is plainly worth doing. The other two are questions about the shop's margin that the score cannot answer, because the shop's margin is not one of its inputs.

How recommendations and auto apply work

Recommendations are generated from the account's performance history, its settings and trends across Google, and Google regularly launches new recommendation types.⁴ They come in families: ads and assets, bidding and budgets, keywords and targeting, and repairs.⁵

The bidding and budget family is the one that matters most here. A few examples, in Google's words. "Adjust your ROAS targets: get more conversion value by adjusting your ROAS targets", recommended because simulations show the campaign could receive more conversion value with a smaller relative decrease in the target. "Raise your budgets", recommended when the ads missed 5% or more of their potential traffic last week. "Add broad match keywords", to get more conversions at a similar or better return.⁵

Notice the shape. Each one promises more conversions, more conversion value or more traffic, at a similar return. None of them promises more profit, because Google does not know your cost of goods.

Auto apply is the switch that turns recommendations into changes without a person clicking. Once it is on, chosen recommendation types apply regularly. You can pick bundles such as "Maintain your ads" or "Grow your business", or choose types one by one, and you can review a queue and dismiss anything that does not fit your goals before it runs.⁶ Google states that auto applying will not raise your budget, and budget raising is not in the auto apply list.⁶ Bidding changes, broad match keywords, search partners and targeting expansion are on it.⁶

Two dated changes are worth knowing. From 26 January 2026 the "Add responsive search ads" recommendation no longer suggests or applies new responsive search ads automatically, and it has gone from the auto apply settings.⁶ And from June 2026 Google relabelled its bid strategies, so "Maximise conversion value with a Target ROAS" is now simply "Target ROAS", with the bidding behaviour unchanged.⁷

The flow below is the loop the badge sits inside.

The chart reads from the top. A recommendation is born, the agency makes a decision at the diamond, and either branch moves the score. The last step is the one people forget. Google says a dismissed recommendation can reappear after a certain period if it becomes relevant again, and new types keep arriving.³⁴ The decision is never made once. It is made again and again, and only one branch takes a single click.

The core problem: a score that can rise while profit falls

Here is the whole argument in one sentence. The badge needs the score, the score is built around more conversion value at a similar return, and more conversion value is not the same thing as more profit.

The gap between the two is the difference between an average and a margin. A campaign's ROAS is an average: all the revenue divided by all the spend. Profit depends on the next euro, the marginal return, and on what you keep from each euro of revenue.

The profit the ads bring in is

$$\pi = m \times R - S$$

Here $\pi$ is the profit from the ads in a month, after the cost of the goods but before fixed costs, $m$ is the contribution margin, the share of each euro of revenue the shop keeps after goods, packaging, delivery and payment fees, $R$ is the revenue the ads bring in, and $S$ is the ad spend. Revenue and spend over the same period make the ROAS a rate, see Foundations: ratios and rates.

Whether more spend is worth it depends on the marginal return, what the extra spend brings in:

$$\text{mROAS} = \frac{\Delta R}{\Delta S}, \qquad \text{mROAS}^{*} = \frac{1}{m}$$

Here $\text{mROAS}$ is the marginal return on ad spend, $\Delta R$ is the extra revenue and $\Delta S$ is the extra spend, and $\text{mROAS}^{*}$ is the break even point: an extra euro of spend adds profit only if it brings in more than $1/m$ euro of revenue, because only the share $m$ of that revenue is kept. Why the next euro matters more than the average is explained in Foundations: marginal versus average.

Worked example: a Kilkenny tea shop lowers its target

Nore Tea is an invented online tea shop in Kilkenny at noretea.example. Its contribution margin is 40%, so $m = 0.40$ and the break even marginal return is $1 / 0.40 = 2.5$. Every extra euro of ad spend has to bring in more than €2.50 of revenue, or it loses money.

Today the Search campaign runs on Target ROAS at 500%. It spends €4,000 a month and brings in €20,000 of revenue. Profit from the ads is $\pi = 0.40 \times 20{,}000 - 4{,}000 = 8{,}000 - 4{,}000 = 4{,}000$, so €4,000 a month.

The recommendations page shows "Adjust your ROAS targets", suggesting 350%, with a score uplift of 7 points that would lift the account from 65% to 72%. The uplift and the forecast below are this example's assumptions, chosen to match what the recommendation promises: more conversion value for a lower target.

The forecast says that at 350% the campaign would spend €8,000 and bring in €28,000. That is exactly what was promised. Conversion value up from €20,000 to €28,000, a rise of 40%, see Foundations: percentages. ROAS still a healthy 3.5.

Now the profit at 350%: $\pi = 0.40 \times 28{,}000 - 8{,}000 = 11{,}200 - 8{,}000 = 3{,}200$, so €3,200 a month.

Profit falls by €800 a month. The reason is in the marginal return. The extra spend is $\Delta S = 8{,}000 - 4{,}000 = €4{,}000$ and the extra revenue is $\Delta R = 28{,}000 - 20{,}000 = €8{,}000$, so $\text{mROAS} = 8{,}000 / 4{,}000 = 2.0$.

That is below the break even of 2.5. The extra €4,000 of spend earns €8,000 of revenue, of which the shop keeps $0.40 \times 8{,}000 = €3{,}200$, and €3,200 kept against €4,000 spent is a loss of €800. The average ROAS of 3.5 looks fine and hides it completely.

Over a year that is $12 \times 800 = €9{,}600$ of profit gone. In return the account score is 7 points higher, the revenue line is 40% higher, and Google's ad revenue from the shop has doubled.

Let me add a middle step, so you can see where it turns. Suppose a target of 400% would spend €6,500 and bring in €26,000. Profit would be $0.40 \times 26{,}000 - 6{,}500 = 10{,}400 - 6{,}500 = €3{,}900$, just €100 below today. The marginal return of that first step is $(26{,}000 - 20{,}000) / (6{,}500 - 4{,}000) = 6{,}000 / 2{,}500 = 2.4$, already a touch under 2.5. The step from 400% to 350% then has a marginal return of $2{,}000 / 1{,}500 = 1.33$, deep in the red.

The horizontal axis shows the three ROAS targets, from today's 500% to the recommended 350%. The vertical axis shows the monthly profit from the ads in euro, after the cost of goods and the ad spend. The bars fall from left to right while revenue rises, from €20,000 to €26,000 to €28,000. The takeaway: the recommendation delivers what it promises, more conversion value, and the shop ends up with less money.

This is not a trick of the example. Google's simulation forecasts conversion value. Your margin lives in your accounts, not in Google Ads, unless you feed profit in as the conversion value, which most shops do not. For a shop with a 70% margin, break even would be $1 / 0.70 = 1.43$ and the very same move would add profit. The recommendation is not wrong. It is blind to the one number that decides whether it is right.

Why the dismiss button does not close the gap

The fair reply is that the agency can simply dismiss the recommendation. Since 2021 Google has let partners apply or dismiss recommendations to reach the 70%, and when the requirement was first announced in 2020 Google clarified that partners are not required to adopt all recommendations or reach 100%.⁸⁹ The requirements page repeats it today.²

That is true, and it is a real improvement. It does not change the incentive, for three reasons, each one in Google's own documentation.

The first is that dismissing is not final. A dismissed recommendation can come back after a while if it becomes relevant again, and Google keeps generating new ones and launching new types.³⁴ The agency that said no to a lower ROAS target in March may be asked again in May.

The second is the cost of each answer. Applying is a click. A good dismissal is a judgement: checking the margin, the stock, the cash and the forecast, then writing down why. Both count towards the score. Only one costs an hour.

The third is the bigger badge. The Premier Partner ranking is built on clients' ad spend growth, new client spend, retention, spend outside Search and total spend.² An agency that says no to spend that loses its client money is, by those measures, a weaker agency.

Worked example: two agencies, one recommendation

Two invented agencies each run an account like Nore Tea's, and both see the 350% recommendation.

The Cork agency applies it. Its client's spend goes from €4,000 to €8,000 a month, €48,000 more over a year. That counts towards existing client growth and towards annual spend, two of the Premier Partner factors.² The client is €9,600 a year worse off.

The Limerick agency dismisses it, with the margin calculation as the reason. Its dismissal counts towards the score as well. Its client keeps the €9,600. Its spend under management is flat, so on every spend based factor it ranks below the Cork agency.

Neither agency did anything dishonest. They answered the same question under the same rules, and the rules gave the higher marks to the answer that cost the client money. That is what I mean by a cost paid by the client.

What I saw in my own accounts

The arithmetic above is invented. What follows comes from my own accounts, which were British, so the figures are in pounds, and it shows the cost I mean.

I have managed campaigns where manual bidding, narrow keyword targeting and controlled budgets outperformed every automated setup, yet the score stayed stuck at 55%, with red warning flags and an endless list of suggestions to "improve" it. It was not a performance score. It was a compliance score.

I have seen clients spending £10,000 or more a month pushed into recommendations that increased their daily spend by up to £5,000, not because performance justified the increase, but because it was one of the fastest ways to lift the score. These were not startups throwing money at growth. They were established businesses with strict ROAS requirements and cash flow plans.

Campaigns that had been built and refined over months, with strong targeting and sustainable returns, suddenly started flagging as underperforming. The culprit was a refusal to apply low quality recommendations, such as moving exact match keywords to broad match, or replacing structured assets with generated ones that lacked message control.

In one case a campaign on manual bidding returning 600% was flagged because it was not using Target ROAS. Activating that recommendation cut the return by more than half and pushed costs up with little strategic targeting.

That left me a choice: follow the list and keep the badge, or act in the client's interest and risk it. Many choose the badge. I did not.

This is a textbook case of what economists call the principal agent problem. The client is the principal, the agency is the agent, and Google pays the agency in badges for behaviour the client never asked for. I take the theory apart, with fees on spend and ROAS targets, in principal agent theory in agencies and retainers.

What the badge does give you

Leaving the programme does not mean the programme is worthless, and it would be unfair to pretend otherwise.

A Partner badge tells you three true things. The company manages a meaningful amount of Google Ads spend, at least US$10,000 every 90 days. At least half its account strategists passed Google's certification exams, including one in each product area it spends real money in. And its accounts are kept current against Google's recommendations, one way or the other.²

It also gives the agency things. Google lists the benefits in three groups: education and insights (product education, certifications, customised account recommendations, an annual industry report), recognition and rewards (competitions, promotional offers, the badge and a listing in the partner directory) and access and support (Google Ads support, plus executive experiences and dedicated account support by invitation). What each company gets depends on its tier, and benefits may not be available in every country.¹¹⁰ A direct line to Google support is genuinely useful when something breaks on a Friday afternoon, fair play to Google for that.

What the badge cannot tell you is whether the agency puts your margin above the score. That needs a different question. The guide to picking the right growth or digital marketing agency treats the badge as one input among many, and shows how to score a shortlist on the things that matter to you.

Worked example: a Cork founder reads a badge

An invented founder runs a homeware shop in Cork and is choosing between two agencies. One has the Partner badge. One does not.

From the badge she learns that the first agency spends at least US$10,000 every 90 days across its clients, that at least half its strategists are certified, and that it can reach Google's support team through the programme. All useful.

Then she asks both agencies the same three questions. What is my break even marginal ROAS? How do you decide whether to apply a bidding or budget recommendation? Can I see the change history and the reasons for the last ten dismissed recommendations?

If an agency answers the first with $1/m$ and her own margin, the second with a margin check, and the third with a list, it will serve her well, badge or no badge. If it answers the second with "we follow Google's best practice", the badge has told her all she needed to know.

What I do instead

I still run Google Ads campaigns. I run them on my own terms, for the client's business rather than for Google's internal metrics.

• I evaluate every recommendation critically and apply only the ones backed by data, strategic fit and commercial logic.
• I optimise towards defined business outcomes, whether that is leads, sales, ROAS or profit, never towards an abstract health marker.
• I track conversion quality, not just volume.
• I keep budget control at the centre and scale only when the data shows consistent returns.
• I compare manual and automated bidding with controlled tests in context, rather than trusting the claim.
• I fit every campaign to the client's market, product lifecycle and cost structure, not to industry benchmarks.

I document every change and explain it, so the client knows exactly what is being spent and why. No black boxes, no blind trust in an algorithm.

Partner status is not the same as partnership. If you keep being told to "just trust the system" while your returns shrink, the badge on your agency's homepage was never the thing that mattered. The margin was.

Footnotes

1. Google, About the Google Partners programme, Google Ads Help, read November 2026, https://support.google.com/google-ads/answer/9029203?hl=en-GB

2. Google, How to become a Google Partner or Premier Partner, Google Ads Help, read November 2026, https://support.google.com/google-ads/answer/9702452?hl=en-GB

3. Google, About optimisation score, Google Ads Help, read November 2026, https://support.google.com/google-ads/answer/9061546?hl=en-GB

4. Google, About recommendations, Google Ads Help, read November 2026, https://support.google.com/google-ads/answer/3448398?hl=en-GB

5. Google, Types of recommendations, Google Ads Help, read November 2026, https://support.google.com/google-ads/answer/3416396?hl=en-GB

6. Google, About applying recommendations automatically, Google Ads Help, read November 2026, https://support.google.com/google-ads/answer/10279006?hl=en-GB

7. Google, About Target ROAS bidding, Google Ads Help, read November 2026, https://support.google.com/google-ads/answer/6268637?hl=en-GB

8. Barry Schwartz, Google Lets Agencies Dismiss Some Partner Program Recommendations, Search Engine Roundtable, 24 February 2021, https://www.seroundtable.com/google-dismiss-partner-program-requirements-30984.html

9. Barry Schwartz, Google Clarifies New Partner Program Performance Requirements, Search Engine Roundtable, 18 February 2020, https://www.seroundtable.com/google-clarifies-new-partner-program-performance-requirements-29018.html

10. Google, Benefits of the Google Partners programme, Google Ads Help, read November 2026, https://support.google.com/google-ads/answer/9029198?hl=en-GB