Google Keeps Telling You to Spend More. When to Say No.

The email from your Google rep arrives with a number attached: 30% more conversions, right there, unclaimed. What the email leaves out is that the budget behind that 30% might have to increase fivefold, a detail Reva Minkoff of Digital4Startups documented in Search Engine Land on August 3, 2026. Google's recommendation is an ad for Google's revenue, not a plan for yours. The arithmetic that tells the two apart takes about five minutes.

Search for whether you should accept Google Ads recommendations and you will get pages telling you to weigh each recommendation carefully and line it up with your business goals. Not one of them gives you the division problem: what does one extra conversion actually cost under this specific prompt, and is that more or less than a customer is worth to you. They also all predate the 2026 recommendation engine, which now includes a "missed growth estimates" beta and an August 17 change to how Target CPA and Target ROAS campaigns bid. Target ROAS is the return you tell Google to aim for on every dollar of ad spend. I have run Google Ads accounts for 10 years and worked in search for 17, across 300+ businesses. The same accept-and-regret patterns show up in account after account, and what follows is the rule I use.

Google shows you the gain and never shows you the price

Accepting any recommendation Google surfaces moves money in one direction. Minkoff published the cleanest example I have seen of what that costs. Google recommended one of her clients double their daily budget. The projected return was 0.75 additional conversions per week. Run the division and you are paying, in her words, 14 times as much each week for a single projected conversion.

Notice what the interface never does. It never puts that 14 next to the recommendation. It shows you a gain and a green button and leaves you to work out the price on your own. Almost nobody does that at 9pm on a phone between a school run and payroll. In July 2026, Google went further and started testing "missed growth estimates" in the Recommendations tab, a number telling you how many clicks, conversions or how much revenue you are missing right now. Search Engine Land reported it on July 21. It is a well-built piece of persuasion. It converts an ordinary account into a leaking bucket in your mind, and it does that before you have checked whether the traffic on offer converts at all.

Translate that into your bank account. If a recommendation costs 14 times your current rate per extra sale, and your average customer is worth a few hundred dollars, you are not buying growth. You are buying volume at a price that only makes sense to the company selling the impressions. That is a decision, not a setting, and it belongs to you.

Your optimization score is a sales target with a progress bar

Owners accept these because the pitch is built to be accepted. You are shown conversions gained and never cost per conversion gained. You are shown a score out of 100 that drops when you decline, which is the same psychology as an unread badge on an app icon. And the company behind the email gets paid when you spend more, not when your margin holds.

The second reason is that saying no feels like admitting you are small. Owners accept budget increases they cannot justify because declining feels like telling Google the business is not ready to grow. Google is not judging you. Google is running a query against your account and firing a template.

The third reason is the one that costs the most. It is mechanical, not emotional. Minkoff makes the point plainly: once Google raises a budget, it does not lower it again on its own. There is no automatic reversal and no expiry date. The increase becomes your new normal and the overspend runs for months until a human notices. One accepted prompt in March is still charging you in August. That is the same slow leak I write about in the audit that finds the money already draining out of your ad account, except this one you approved yourself.

Two 2026 changes moved money, and one came with Google's own advice

On August 17, 2026, Google changed how Target CPA and Target ROAS campaigns behave. Brooke Osmundson covered the clarification in Search Engine Journal on July 8, 2026, and the mechanic matters for anyone whose campaigns are capped by budget. A budget-limited campaign frequently over-performs its own target: your target CPA says $50 and you are actually paying $35. A capped budget runs out before the day's most expensive auctions, which is how the average stays low. Since the change, bidding optimizes toward the target you set. A campaign that was buying leads at $35 can drift up toward $50 unless you lower the target.

Nobody is doing anything sneaky here. The system is doing exactly what the number in the box tells it to do. The problem is that most owners set that number once, years ago, as a guess. Google's own liaison was direct about the framing.

Google's liaison was telling owners to keep the reins tight. If your campaign has been over-performing at $35 against a $50 target, bring the target down close to your real result before the gap closes on its own. The number itself comes from your margin, and how to set your Target CPA or ROAS from your own margin covers the rule and the math together.

Minkoff flags four settings worth a closer look. I have added the two recommendations I accept in almost every account, because together they sort into two piles once you know what each one is really selling.

The pattern is not complicated. The recommendations that quietly repair your own mistakes are worth accepting the day they appear. The ones that expand where your money goes, or how fast it goes, need the math run first, every time, with no exceptions for a nice rep.

The five-minute test that answers every spend-more prompt

You do not need a spreadsheet or a certification. You need one division and one comparison, and you can do it on the phone in the car park.

Minkoff documents a second version of the same pitch in that August 3, 2026 Search Engine Land piece. A rep email promises 30% more conversions, and the budget behind it might have to increase fivefold. Run the same division on that one and the shape is clear. More conversions do arrive, but the denominator of spend grows several times faster, and the email quotes the numerator only.

Take a physio clinic with two therapists and a waiting list that runs about a week deep. A budget prompt arrives promising more bookings. Step four decides it. The clinic has to ask whether the two therapists have any hours left to sell. Enquiries it cannot fit in still get paid for. When each extra one costs about what a patient is worth in profit, the clinic ends the quarter busier, more tired and no richer.

If you have an agency, do not ask them "should we take this recommendation." Ask them one question instead: what is the cost per extra conversion on this prompt, and how does it compare to our current cost per conversion. A good partner answers in a sentence. If the reply is about optimization score or momentum, you have learned something about the partner as well as about the prompt.

One more job takes ten minutes. Open every campaign using Target CPA or Target ROAS and compare the target you set against what the campaign has actually been delivering over the last month. Where the real result is comfortably better than the target, move the target down close to the real result. The same logic applies to anything else in the account that hands decision-making to Google's automation. I keep telling owners not to hand the keys of the ad account to an AI assistant that has never met a customer.

Thirty days later, check the number Google will not put on the dashboard

Whatever you accept, put a reminder in your calendar for thirty days out. Google will not remind you and the ratchet Minkoff describes runs silently: budgets go up and stay up. The only thing standing between an accepted prompt and months of overspend is a human opening the account on purpose.

A missed-growth number can only ever go to zero if you spend without limit. It will never stop glowing at you. It is designed to be permanently unsatisfied. Treat it as a measure of how much inventory Google still has to sell you, which is what it is.

Some campaigns genuinely are starved, and the projected extra conversions genuinely do come in below your profit per customer. When the math clears, take the money and go. The rule is not "never spend more." The rule is that the number has to be shown to you and it has to survive contact with your own margins. If Google will not do that division for you, you do it yourself in five minutes.

Frequently Asked Questions

The instinct most owners have is that declining Google's advice is reckless. Run the division a few times and the opposite becomes obvious: accepting it without the division is the reckless part, because you are approving a price you were never shown. Google is not your partner in this. Google is a supplier with an excellent interface and a quota, and suppliers do not get to set your budget.

If you want me to check which prompts in your account are worth taking, book a call and we will run the math on your actual numbers together.