Platform · Campaign structure
Every campaign needs around 30 conversions a month for Google's algorithm to learn on it. The portal counts how many conversions your ads bring, says how many campaigns that is enough for, and goes through each of the ones already running.
Sign in with Google in one click. The portal changes nothing in your account — no campaigns, no bids, no budgets.
Why campaigns don't learn
Google's algorithm needs at least 30 conversions per month for every campaign to learn properly and reach its targets; ideally more than 50. In the account below all ten product campaigns together give exactly 30, and not one of them reaches the minimum bar. The question is not the size of the budget, but how many parts it is split into.
Google's learning threshold — 30 conversions per month
The same conversions in one campaign — and the learning threshold is met. Merging campaigns adds not a single conversion — it stops you splitting them. A campaign that reaches the norm learns more precisely and comes closer to its targets — and that is where ROAS and revenue come from.
An account in the portal: 10 product campaigns, a 90-day window. The numbers of one shop, not a market average.
Every campaign has a row of its own: how much it spends and what share of spend it takes, how many conversions per month it brings against the norm of 30, ROAS, how many of its products really had impressions over the last two weeks — and a verdict. Expand the row and you will see the strategy, the target ROAS, the budget, the action the campaign learns on, the trust in its conversions and every problem with a ready prompt on what to do.
The wording is the same one you will see in the portal: the site and the product call things by the same names.
there are problems that need urgent attention
the campaign works, but there is room for improvement
working as it should — we found no problems
running for less than 45 days — too early to judge by conversions
younger than two weeks, spend already noticeable, no conversions — we keep it in view
enabled, but effectively off: not a single impression in 90 days
A rung, not a sentence
The section starts not with “rebuild everything”, but with what you must not do right now. The portal looks at how many conversions your ads bring and says how many active campaigns that matches and how many there could be. On the “too few conversions” rung, splitting campaigns is forbidden — there wouldn't be enough data for each. And only then does it recommend a structure: one where every campaign gathers enough conversions to learn, to scale and to improve its numbers.
The rung is set by your numbers, not by our opinion. First the portal looks at whether your conversions can be trusted, then it counts how many of them the product campaigns bring: 30 a month opens rung 2, 170 opens rung 3. Below is what the portal allows and what it forbids on each of them, in its own words.
0 · no trust
the bottom of the ladder — start here
safe to do now
not yet
rebuild the structure on wrong data — Google will learn from garbage
1 · too few conversions
entry: correct conversions
safe to do now
not yet
split into more than 2 campaigns — there won't be enough conversions for each
2 · time to expand
entry: from 30 conversions per month
safe to do now
not yet
go beyond 2–3 campaigns — Google will lose data
3 · scale
entry: from 170 conversions per month
safe to do now
not yet
fundamentally change the structure — rebuild it from scratch
A rung can be lost as well: if conversions disappear, the portal moves it down and tells you to check the assortment first, and only then the structure.
Five passes over 90 days of Google Ads data. For each one — what it found in the account on the screens above.
The main thing is product campaigns, Performance Max and Shopping. Search the portal looks at separately and compares only with the product ones: if they perform noticeably worse, it lowers the section score and says by how much. Video, Display and Demand Gen it lists with their spend but does not assess — their job is not conversions, and measuring them by the same yardstick would be a lie.
found in the account
10 product campaigns, 2 Search
How many conversions per month each campaign brings. We count not by the calendar, but over the time the campaign has been running without a break: Google doesn't give the launch date, and a campaign could have been stopped and started again — then an “average over the period” would show something untrue. The norm is 30 conversions per month, the ideal is 50.
found in the account
not one of the 10 campaigns reaches 30 conversions per month
Overlaps: the same product serving from two campaigns at once for more than three days — they compete with themselves. Campaigns that are too narrow: fewer than 100 products, too little data on them to draw conclusions. And what part of the catalogue gets no impressions at all, because no campaign picked it up.
found in the account
overlaps — 65% of spend: 1,360 € over 30 days goes on campaigns competing with each other
Whether the campaign's strategy fits the data coming into it. The most common breakage is a campaign optimising for ROAS while the order amount isn't passed to Google: the algorithm learns on the default value. The portal checks two more things: which action the campaign learns on, and whether the target ROAS is set too high against what the campaign actually delivers.
found in the account
8 problems with optimisation settings
How many conversions the product campaigns bring — that sets the maturity rung, and the rung sets what you can and cannot do with the structure right now. Not “the right structure in general”, but the one your conversion volume can carry today.
found in the account
rung 1 — one campaign, two at most
We always count over 90 days: the period selected on screen does not affect this calculation.
Three situations that make people open campaign structure.
Every rebuild wipes out what has been accumulated. Burned once, you don't want to try again, even when you can see something is off.
What the portal does
A structure is not reshuffled every week: it is changed for the season or once every few months, when demand shifts. And the portal doesn't call for rebuilding everything from the ground up — it strengthens what already sells, carves new products and those going out of season into separate tests, and watches that every campaign has enough conversions. On the bottom rung it directly forbids splitting campaigns. The point is not that changes are forbidden — it is that changes have to come in careful steps and on sufficient data.
Spend goes out, sales come in, the report is not empty. That each campaign on its own is short of conversions is not visible from the Google account: there is no line there saying “this campaign doesn't have enough to learn”.
What the portal does
It counts conversions for each campaign separately and puts the norm next to it. “Everything works” and “works at full strength” are different things: a campaign short of conversions has nothing to learn from, and it will not get better on its own. For those that fall short, the portal names the neighbours worth merging them with, and how many conversions they would have together.
One contractor built it by category, another proposes a single campaign for the whole catalogue. Both argue from experience.
What the portal does
The portal argues differently: your data and one threshold — 30 conversions per month per campaign. Under every recommendation it says where it came from and how many points it returns. That can be verified, not taken on faith.
Access
Google Ads. Sign in with Google in one click — we change nothing in your account without your consent. Merchant Center is not needed for this section
Plan
The structure breakdown is on every plan, including the free one
Window
90 days. Independent of the period selected on screen
Automation
Auto campaign structure — the Scale plan, currently in development
The screens come from an account in the portal.
The rung depends on conversions, and conversions depend on whether they can be trusted and whether there is anything to sell. Both answers are right next door.
Whether what you count is what gets counted, and whether the figures can be trusted
See the section →
Which ones earn, which burn budget, which never show at all
See the section →
A grade, a score and a breakdown across four areas — where to start
See the section →
No. The portal changes no campaigns, no bids, no budgets: it counts, highlights what the numbers say will pay off, and gives a recommendation — whether to apply it is up to you. Nothing in your account changes without your consent. About the access itself: Google Ads has no separate read-only access, the same access is granted for both reading and writing, which is why the list of permissions on Google's screen sounds broader than what the portal does. If you'd rather not hand that over — sign in with a Google account that holds the Read only role in your account: the portal inherits its permissions and cannot change anything even in theory.
No, it is Google's learning threshold: the algorithm needs about 30 conversions per month per campaign, ideally 50. We see the same on our own data — across the 150+ shops we have analysed: as soon as a campaign steadily gathers 30–50 conversions, it comes noticeably closer to its targets. What is ours here is that we measure the pace honestly, over the campaign's current run rather than over a calendar month. The launch date in Google's export is empty, and a campaign could have been stopped and started again — and then “averages over the period” lie.
Not today. The section shows what to change, and you make the changes. Automation is coming — auto-structure and applying changes through the Google Ads API are part of the Scale plan, which is still in development — but even then it will only fire on your own Apply button. Nothing in your account changes by itself.
Search — separately and relative to the product campaigns: if it performs noticeably worse, the portal lowers the section score and says by how much. Video, Display and Demand Gen are listed with their spend but not assessed — their job is not conversions, and measuring them by the same yardstick would be a lie.
That the rebuild is visible, and what came out of it. The portal recognises it by the age of the campaigns, and checks the settings and composition straight away — those read from day one — and once data accumulates in a couple of weeks it will say how well the rebuild worked and highlight the places worth touching up.
Sign in with Google in one click. The portal changes nothing without your consent.
Analyse your campaignsTell us about your store and your question — we reply during business hours. If a call is easier, leave a phone number.
GetProfit analyses your Google Ads and Merchant Center: it shows which products bring sales and which spend budget with no result, checks the feed and the conversion setup, and gives an account score with recommendations.
By continuing, you agree to the Terms of Service · Privacy Policy