To answer it, we have to look at the interventions as well as the checks and decisions that come before them. A campaign may need a quick fix, a prepared test or keeping the setup as it is until there is enough data. Each situation calls for an explanation.
A lack of changes proves neither negligence nor good management on its own. Management is assessed through the agreed responsibilities, the quality of the analysis, justified interventions and the results tracked in the context of the business.
How long should a campaign run before major changes?
The interval depends on the type of change, the volume and maturity of the data, the time to conversion and the commercial risk. We establish what we want to evaluate and when the information becomes sufficient. An error in the offer, the destination or measurement may need immediate action, even if evaluating performance would take longer.
For adjusting targets in Search Smart Bidding, Google recommends evaluating after one or two conversion cycles. The cycle also includes the reporting delay for imported conversions. This reference point has a specific context and is not a universal deadline for every platform or campaign. Source: Google Ads, How our bidding algorithms learn.
Does passing the interval guarantee a conclusion? No. If volume stays low, the result can remain uncertain. The team has to explain what it knows, what it is still checking, how long it is justified to continue and which signal would change the decision. Waiting has a purpose and a limit, just like intervening.
What the change history shows
Google describes Change history as a tool for reviewing changes to the account, campaigns and ad groups. We can see, for example, changes to budgets or keywords and filter by period or type of change. Source: Google Ads, About change history.
It is a useful record of the interventions available in that report. By design, it is not a timesheet of the analysis or of the whole collaboration. Reading a report, investigating a fluctuation or preparing a recommendation is not the same as changing a setting.
That limit does not justify a vague answer. The team managing the account has to be able to explain the relevant work, and the client can ask for appropriate evidence: observations, recommendations, tests, materials or documented decisions, depending on the service contracted.
Monitoring, analysis and intervention
Monitoring looks for signals that deserve attention. Analysis tries to explain the situation. Intervention changes something for a defined reason. Sometimes all three happen on the same day; at other times, the analysis leads to a decision to keep watching.
A useful check can look at budget pacing, how the page is working, conversion measurement, the quality of inquiries and the gaps against the objective. Frequency and depth depend on the size of the account, its volatility and the agreed responsibility.
There is no universal number of weekly changes that proves competence. If the metric becomes “how many changes we made,” there is a risk of producing activity without a clear question. If nothing is documented, the client cannot tell a deliberate decision from a lack of attention.
When we intervene and when we need to understand first
| Observed situation | What we check | Possible direction |
|---|---|---|
| The ad points to an expired offer | Which materials and destinations are affected | We correct it or limit the affected exposure |
| Cost per result has recently increased | Incomplete data, changes in the mix, quality and technical causes | We investigate before making a broad change |
| Inquiries are numerous but unsuitable | The measured objective, the message, the audience and sales validation | We adjust what is feeding the problem |
| The setup is working within the agreed limits | Stability, opportunities and capacity to handle demand | We keep it or prepare a justified test |
These are directions for analysis, not automatic rules. A problem that spends budget on a broken destination should not be left running just to preserve an observation window.
Why the last day can tell an incomplete story
Someone can interact today and send the inquiry later. If the result reaches the platform with a delay, a recent period can have costs already recorded and conversions not yet reported.
Google recommends evaluating adjustments with the conversion cycle in mind, including delayed reporting. In its guidance for Smart Bidding in Search, it suggests evaluating after one or two conversion cycles for more complete data. This is not a fixed deadline that applies to every account. Source: Google Ads, How our bidding algorithms learn.
A fictional example. A service receives a large share of its inquiries a few days after the click. Comparing the last two days with a fully matured week can suggest an exaggerated drop. The team checks the delay and the quality of the inquiries before sharply cutting distribution.
If the results stay weak after checking, we need to act. Conversion delay explains a limit in the data; it cannot become a permanent justification for any result.
Not every change affects learning in the same way
A copy correction, a change of bidding strategy and replacing the conversion we optimize for have different implications. That is why the claim that “any change resets the algorithm” is too broad.
In its Search Smart Bidding documentation, Google states that changing a CPA or ROAS target does not erase accumulated learning. Even so, a large change can alter auction participation and results. The guidance on changing conversion goals and actions separately describes an adjustment period, usually one or two cycles. Sources: Smart Bidding learning and changing conversion goals.
The timing of an intervention depends on the problem and on what we are changing. We do not mechanically apply a maximum budget percentage or the same number of days to every situation. We check the relevant documentation, the account history and the commercial consequences.
Simultaneous changes make the explanation harder
If we change the offer, the budget, the conversion goal and the materials at the same time, the result that follows can have several explanations. Sometimes a broad change is necessary, for example for a new offer. In that case, we document the set of interventions and avoid attributing the whole effect to a single component.
When we want to find out what works, we choose a test question and a suitable method. Not every before-and-after comparison proves causation. Seasonality, competition or a price change can affect the same period.
The principle is developed in “If we change everything, what do we learn?”. In account management, it helps us keep the link between a decision and what we can conclude from it.
What a decision to keep the setup looks like
A useful explanation can say what was checked, what remains within limits, what data is still missing and when the situation will be reassessed. It can also state the signal that would trigger an intervention: a technical problem, a confirmed decline or a change in the commercial objective.
This kind of communication gives the client visibility without turning reporting into a list of button presses. Results and the quality of the service have to be assessed together; the existence of a log alone does not prove performance.
NOT MORE. can take on the promotion strategy, implementation, materials, measurement and agreed management, fully or alongside existing teams. Reporting links observations to decisions and next steps. The client contributes information about inquiry quality, sales, capacity and changes in the business.
If the assessment starts from an automated report, continue with “The report is ready. Is the conclusion right?”. The useful question is what was checked, what was decided and how well that work supports the agreed objective.