SEA in the AI Era: Why Account Managers Are Not Going Away
AI agents able to run a Google Ads account now promise fully automated advertising: campaign creation, bidding, budgets, reporting. So the question comes up in many companies: do we still need an SEA account manager?
The answer is yes, for three fundamental reasons: AI assumes its data is right and knows nothing about the business; it is only as good as the person using it; and it amplifies the good as much as the bad. That does not mean doing without it: used well, AI transforms the way account managers work. The point is knowing where it helps, and where it does not.
1. AI assumes the data is right, and has no business context
An algorithm learns from what it is given. It has no way of knowing that a conversion tag stopped firing, that a sale was counted twice, or that a week of leads was spam. It treats that data as reality and optimises accordingly.
It also knows nothing about business events: a stock-out, a price increase, a product launch, a trade show, a strike, a website outage. None of this exists in the ad account.
Google acknowledges this itself, through two features:
- Data exclusions, to tell Smart Bidding about a period when conversion data was wrong or missing (About data exclusions, Google FAQ; Use data exclusions for conversion data outages, Google FAQ);
- Seasonality adjustments, to warn it about an expected change in conversion rate, for example during a promotion (About seasonality adjustments, Google FAQ).
These tools exist because the algorithm cannot guess. Someone has to know what is happening in the business, spot the anomaly and step in. An AI agent running an account inherits exactly the same limitation: it reads the same data, with the same confidence.
2. AI is only as good as the person using it
AI is an assistant. The person using it asks the questions, sets the goals, defines the constraints and validates the results.
A badly framed question produces a convincing but wrong answer. And judging whether a recommendation makes sense takes expertise: should you really raise this budget, switch these keywords to broad match, or merge these campaigns? Without domain knowledge you do not see the mistake; you apply it.
The account manager remains the one who sets the direction and the guardrails. AI executes, suggests and speeds things up; it does not replace judgement.
3. AI amplifies the good and the bad
AI works faster and at a larger scale than any human. That is its strength, and its risk.
- With a sound strategy, reliable conversions and realistic goals, it multiplies the account’s efficiency.
- With a wrong conversion goal, an incoherent structure or an unrealistic target, it damages the account faster and further than any manual mistake.
In other words, automation does not fix poor management: it accelerates it. Hand an account to an AI under a weak account manager, and the catastrophe simply arrives sooner.
Where AI really changes the game: your own tools
AI is, however, remarkable for building custom management platforms: scripts, dashboards, alerts, audit tools, automated checks. Developments that used to take weeks are now within reach of an account manager, building on the Google Ads API (Introduction to the Google Ads API, Google documentation) and Google Ads scripts (Google Ads Scripts: Product overview, Google documentation).
It also helps you understand, and finally use, API features that were never exploited for lack of development time: bulk changes, anomaly detection, control rules specific to your business.
The division of roles is clear: AI writes the code, the account manager defines the business logic and checks the result.
But AI is not a reporting tool
You should, however, avoid using AI as a new way to produce reports. A language model generates plausible text: it can get a calculation wrong, mix up two periods, or state a wrong figure with confidence. A report requires figures that are exact, reproducible and traceable.
The right approach is to use AI to build the reporting pipeline (API queries, scripts, dashboards), not to produce the figures. The data must come from the platform, never from the model.
Conclusion: AI is an accelerator, SEA managers are here to stay
- AI assumes its data is right and knows nothing about the business: someone has to provide that context.
- It is only as good as the person using it.
- It amplifies the good and the bad.
- It excels at building custom tools, not at producing reports.
AI is a remarkable accelerator. But SEA managers are not going away: their role is shifting from manual tuning to steering, control and building their own tools.
To go further on steering Smart Bidding: Mobile Traffic Eating Your Budget? Portfolio Bidding, Max CPC and Value Rules.
Want to review the automation of your accounts? Get in touch.