You log into Google Ads to check on the month and there is a number in the corner of the screen telling you your account is 68% optimized. Underneath it sits a list of things Google would like you to do about that. Add these keywords. Raise this budget. Lower that target. Turn on this campaign type you have never heard of.
It reads like a report card. Sixty-eight out of a hundred is a D, and nobody wants a D.
So the number does its job, which is to make you feel slightly behind. And somewhere in the next ten minutes a small business owner clicks Apply All, and Google starts spending differently tomorrow morning.
Here is the part almost nobody is told. That score is not measuring whether your account is any good.
The Number in the Corner Is Not Your Grade
Google calls it optimization score, and its own definition is more careful than the way the number gets used. Optimization score is an estimate of how well your Google Ads account is set to perform. It runs from 0 to 100%, and you will find it at campaign level, at account level, and at manager account level if an agency runs your ads.
Read that definition again, because the important word is set. Not performing. Set to perform. It is a measure of configuration, not results. Google is looking at your account, comparing it against the features and settings it believes accounts should use, and scoring the gap.
Every recommendation in that tab is tied directly to the score. Take a recommendation and the score goes up by a stated amount. Leave it sitting there and the score stays down. That is the loop the whole thing runs on, and it is why the tab feels like homework.
But there is a third option in that loop, and it is the one that gives the game away.
Dismiss Everything and Watch It Hit 100
You can dismiss a recommendation. Click the three dots, tell Google this is not relevant to your business, and it goes away.
Doing that also raises your optimization score.
This is not a loophole somebody found. It is documented behavior, stated plainly in Google's own help pages: applying or dismissing these recommendations changes the overall optimization score of your account. Both actions move the number in the same direction.
Aaron Young, who runs Define Digital Academy, demonstrates it on camera in about eight seconds. He opens a campaign sitting at 86.8%. There is one recommendation attached to it. He clicks it away with "I'll do it later." The campaign is now at 100%.
Nothing about the account changed. Not a keyword, not a bid, not a budget, not an ad. The account that was 86.8% optimized at the start of that clip and the account that is 100% optimized at the end of it are the same account.
Which means the number has two completely different meanings and no way to tell them apart. A 100% score might be an advertiser who took every piece of Google's advice. It might equally be an advertiser who read every piece of Google's advice and said no to all of it. Those are opposite businesses and they get the same grade.
Young has seen campaigns at 100% performing poorly and campaigns sitting around 50% producing excellent results. That is not a paradox once you know what is being counted.
There is a practical consequence to this and it is worth saying directly. If somebody managing your ads reports optimization score to you as evidence the account is healthy, that report is empty. Not dishonest necessarily, just empty. The number can be manufactured in a morning by dismissing things.
Google Is Not Lying to You. It Just Cannot See Your Business.
It would be easy to write the rest of this as an argument that Google is out to fleece you, and plenty of people do write it that way. That framing is satisfying and it is not very useful, because it leaves you with no way to tell a good recommendation from a bad one except suspicion.
The more accurate version is duller and far more actionable. Google can see your account. Google cannot see your business.
It can see that a campaign is limited by budget. It cannot see that you are already booked out for five weeks and another twelve leads would just be twelve people you disappoint. It can see that a shopping feed has 400 products and only 280 are being advertised. It cannot see that you pulled the other 120 because that supplier's margin is terrible. It can see that your ads have no price assets attached. It cannot see that you are a plumber and there is no such thing as a price for a job you have not looked at yet.
Aaron Young makes this point fairly, and he is worth quoting on it because he is not a Google apologist: the recommendations tab is not evil. It genuinely helps somebody new to the platform find features they did not know existed. It is simply working from an incomplete picture, and it never says so.
So the useful mental model is not "Google is lying." It is closer to a very capable consultant who has read your account inside out, has never met you, does not know what you sell it for, and gets paid on volume. Some of that consultant's advice will be excellent. Some of it will be confidently wrong in ways only you can spot.
Which brings us to sorting one from the other, and there is a clean line to draw.
The Recommendations Worth Taking Cost You Effort, Not Budget
The good recommendations have something in common. They ask for your time. They do not ask for a decision about money.
Add assets to a thin ad. If your responsive search ad has the bare minimum of headlines and descriptions, Google will tell you so. It is right. More headlines means more combinations to test and more chance of finding one that lands, which matters more than most owners think given that the customer only ever reads your ad. Sitelinks, callouts and structured snippets make your ad physically larger on the results page, which lifts click-through rate on its own. Ben Heath, who has spent more than $20 million on the platform by his own account, describes this one as having no downside at all. It costs a little effort and nothing else.
Put your actual keywords in your headlines. When somebody searches for emergency drain unblocking and your headline says emergency drain unblocking, they click. This is one of those things everybody knows and half of accounts still get wrong, usually because ad groups grew loose over time. If Google flags it, it has caught a real mistake.
Make generic headlines less generic. Weak recommendation on its own, but the underlying point is sound. Most advertisers write their ads by looking at competitors' ads, so entire industries end up sounding identical. You have fifteen headline slots. Spending two of them on something nobody else in your market would write is a cheap bet with real upside.
Fix disapproved assets. Always take this one. A disapproved headline is a headline not working for you, and letting disapprovals pile up in an account is not a habit you want.
Upgrade your conversion tracking. When Google prompts you to move to a newer tracking setup or a data-driven attribution model, this is usually worth doing. It touches how you measure rather than how you spend.
Clear out genuinely dead keywords. Keywords that have not served an impression in months are clutter. Removing them makes the account easier to read.
The test underneath all six: none of them hand Google a decision that belongs to you.
The Ones That Spend Your Money for You
Now the other pile. These share a family resemblance too, and once you see it you can sort new recommendations yourself without anyone's cheat sheet. Every one of them widens the funnel. Widening the funnel always costs money, and the question of whether that trade is worth it depends entirely on numbers Google does not have.
Raise your budget. This is the most common recommendation in the platform and it is not automatically wrong. If you are genuinely profitable at your current spend and want more of the same, scaling up is the correct move, though it is worth being sure the account does not simply need a cleanup first. The problem is the size of the jump. On one account Young walks through, the campaign was running $24 a day and Google's recommendation was to go to $81. That is more than tripling spend in a single move. His own rule is closer to 20% every seven days, which lets bidding adjust instead of throwing it into a market it has not learned yet.
Heath adds a check worth doing every single time. Look at the projected extra sales next to the projected extra spend. He has seen Performance Max recommendations where the extra budget Google was asking for was larger than the extra revenue it was projecting. The math was on screen and it did not work. Scale from a position of strength, when the current spend is already returning well, and not as a rescue for spend that is not.
Change your bid target. This is where the gap between Google's goal and yours is easiest to see. Austin LeClear at Grow My Ads shows an account running a Performance Max campaign at a 900% ROAS target, because that is the number the business needs to clear to make money on those products. Google's recommendation was to drop the target to 774%, with a note that this would bring in more conversion value.
That is true, and it is also useless. Of course a lower target buys more volume. The question is whether 774% is profitable, and Google has no idea, because nobody ever told it what those products cost to make. It is optimizing a number it can see against a constraint it cannot. We have written separately about why your bid target is an instruction rather than a wish, and this is the same lesson arriving through a different door.
Add broad match keywords. Google recommends this constantly. Broad match is not the villain it is sometimes made out to be, but dropping it into a tightly run phrase-and-exact campaign can undo months of work, because the searches it opens you up to are chosen by relevance rather than by your judgment. Our full case for how to use broad match without setting fire to the budget is worth reading before accepting this one.
Expand into Display, or into Search partners. Both take a search campaign, which is aimed at people actively looking for what you sell right now, and spread its budget across inventory where nobody is looking for anything. If Display belongs in your strategy it belongs in its own campaign with its own budget and its own measurement, not bolted onto a search campaign as an expansion setting.
Start a Performance Max campaign. Sometimes right, often premature, and long enough a subject that we have given it its own article on why Performance Max is Google's smartest campaign rather than yours.
"Conflicting Negative Keywords" Is Google Calling Your Best Work a Bug
One recommendation deserves singling out, because it reveals the whole logic more clearly than any other.
Google will tell you that you have conflicting negative keywords. What it means is that a negative keyword in your account is blocking a keyword you are also bidding on, so some searches you could be showing for are being filtered out.
Google presents this as an error. It is not an error. It is the system working.
A negative keyword list has two jobs. The first is obvious: block searches that have nothing to do with you. The second is the one that separates a good account from an average one: block searches that are related to what you sell but that you have learned do not turn into customers. The tire-kickers. The wrong-sized jobs. The searches that generate enquiries you always end up politely declining.
That second job means a mature negative list should conflict with your keywords. If it never did, it would not be doing anything.
Heath calls this the recommendation he ignores more than any other, and his reasoning is the clearest statement of the difference between your goals and Google's that you will find anywhere. His job as an advertiser is to get the most out of the budget. It is not to show his ads to as many people as possible. That might be Google's objective. It is not his.
If you are running out of volume and genuinely need to open up, then yes, revisit the list. Most businesses are not in that position, and for them this recommendation should be dismissed on sight. Which, conveniently, also raises the optimization score. Our piece on paying Google for searches you would never choose covers how to build that list properly in the first place.
Auto-Apply Is the Part That Actually Costs You
Everything above assumes you are the one clicking Apply. There is a setting where Google clicks it for you, and this is the part of the tab with real money attached.
It is genuinely hard to find, which does not help. You cannot see it from inside a campaign. You have to click into view all campaigns first, then Recommendations, then the auto-apply option. Miss that first step and the screen simply is not there.
What you will find is roughly two dozen toggles, grouped into two bundles Google labels "maintain your ads" and "grow your business," covering ads and assets, bidding, keywords and targeting, and measurement. The exact list changes. Google adds and removes options over time, and the two experienced practitioners I drew on for this piece counted 23 and 25 respectively, which tells you how much it moves. Do not go looking for a fixed number. Go looking at what is currently switched on in your account.
Google documents auto-apply as opt-in, and says it can be turned on or off at any time. Worth noting fairly, because the folklore around this feature runs hot. What agencies consistently report, though, is finding it enabled in accounts where the owner had no idea. LeClear puts that figure at around 90% of the accounts he encounters with it on, which is his experience rather than a published statistic, but it matches what we see when we take over accounts.
The sorting rule is the same as the one above, and it is simple enough to apply from memory. Anything that changes what your ads say or how you measure, you can consider. Anything that changes what you spend, what you bid, or who sees your ads, switch off.
In practice that means the ones reasonably left on are optimized ad rotation, removing redundant and non-serving keywords, and upgrading conversion tracking. Everything touching bidding goes off, without exception, because bidding decisions depend on margins and seasonality and capacity that live inside your business and nowhere in the account. Everything touching keywords goes off, because a broad match keyword added automatically to a campaign that was already struggling is how accounts quietly get worse. Display expansion and search partners go off. Store visits is worth having on if you have physical premises and irrelevant if you do not.
Your Agency Is Being Graded on This Too
There is one more piece to this, and it explains a lot of otherwise puzzling behavior.
Optimization score is not just a nudge aimed at advertisers. It is wired into the Google Partners program. To hold Partner status, an agency's registered manager account has to maintain a minimum optimization score of 70%.
Be careful with what that does and does not mean, because this gets overstated online. Google is explicit that partners are not required to adopt all recommendations or reach 100%, and that agencies keep the ability to apply or dismiss based on their own judgment and their clients' goals. Nobody is being forced to wreck an account to keep a badge.
But an incentive that points one direction is still an incentive. There is now a number, attached to a badge an agency displays on its website, that goes up when the agency does what Google suggests. Darren Taylor, who runs an agency himself and is therefore criticizing his own industry here, puts the worry plainly: it pushes agencies toward Google's way of doing things regardless of what the client's results look like.
You do not need to be cynical about this. You just need to know it exists, so that a particular kind of reporting stops impressing you.
The questions that actually tell you something: what did we spend, how many real enquiries did it produce, what did each one cost, and how many turned into work. If you want a second opinion on whether your account is being run for your numbers or Google's, that is a large part of what we do when we take over a Google Ads account.
What to Actually Do This Week
Twenty minutes, and most of it is one screen.
Open Google Ads, click view all campaigns, then Recommendations, then auto-apply. Screenshot what you find, because you will want the record. Switch off everything touching bidding, budgets, keywords or targeting. Leave the ad rotation, dead-keyword cleanup and conversion tracking options on if you want them.
Then go back to the recommendations list itself and work down it once, slowly. Take the ones that ask for effort. Dismiss the ones that ask for money, and dismiss them without guilt, because dismissing is a legitimate answer that Google's own system counts as a resolution.
After that, change what the tab is to you. It is not a to-do list and it was never a report card. It is an inbox from a well-informed stranger who does not know what you charge. Read it monthly, act on maybe one message in five, and stop letting a number in the corner of the screen tell you how your business is doing.
Your account will probably end up somewhere in the eighties, with a handful of things you have deliberately said no to. That is what a well-run account looks like.
