We were asking this question in 2018
After a presentation in Helsinki in 2018, I discussed how machine learning was changing ad creation and bidding. The practical question was whether automation could make better use of an advertiser’s resources.
The original piece described analytics features and third-party case studies from that period. Those are not current setup instructions or a promise of the same result for your account.
Check what the system is learning from
Before relying on automation, establish what counts as a useful outcome. A conversion signal should represent something the business actually values, and the tracking needs checking against the underlying enquiries or orders.
Ask whether the available evidence is sufficient for the decision you are making. Sparse or misleading data is not improved merely by feeding it into an automated process.
Keep control of the experiment
Set the budget boundary, monitoring plan and conditions for intervention before a change. The original anecdote about rising costs followed by improvement does not justify waiting indefinitely for an account to recover.
Review the campaign, landing page and commercial result together. Automation can change how the work is done; it does not remove the need to decide whether the work is worthwhile.
See our Google Ads management and AI consulting.
This 2018 article has been updated to focus on conversion-signal quality, budget boundaries and accountable automation.