Google Ads 2026: Keeping Control With Performance Max and AI Max

Performance Max combines the images, video, headlines and data feeds you provide with Google's models and serves them across whichever surface fits, automatically. Channel choice, placement, bidding and budget split stay with the system; you only set the goal, such as sales, leads or store visits, and supply the assets.

Judging a new Performance Max campaign in its first days is misleading. Google recommends running a campaign for at least six weeks so the learning algorithm can gather enough data to optimise properly. For a generator manufacturer running its first Performance Max campaign, cost per lead is usually high in week two; the real call comes after week six.

AI Max bundles a few features into one layer: broader search term matching, automatic customisation of headlines and descriptions based on the landing page, and final URL expansion. The aim is to reach more relevant searches with the same bids and keyword set.

The difference is this: in a standard Search campaign, you largely decide which words trigger your ad. With AI Max on, that line blurs; the system can also show your ad for searches it judges relevant even if they do not directly match your keyword. That is why checking the search terms report in the first week after launch matters more than trusting the automation on day one.

As automation grows, a handful of settings still need a human decision, or the campaign can end up showing where the brand does not want it to.

• A negative keyword list: competitor brand names, irrelevant search patterns
• Excluded placements and content categories (brand safety settings)
• Conversion action selection: you tell the system which action counts as valuable
• A budget cap and bid strategy (a ceiling still matters even with automated bidding)
• A separate, low-budget Search campaign for brand terms, so Performance Max does not cannibalise it

Search term visibility in Performance Max is limited, but the asset group report, placement report and landing page report should still be read regularly. AI Max has its own reporting section that combines search term, keyword, asset and landing page performance.

A fifteen-minute weekly review usually covers it. For a residential property management firm's campaign, week two of the search terms report showed spend going to irrelevant searches such as 'careers' and 'job vacancy'; once those went on the negative list, budget shifted back to genuine demand.

Frequent changes in a campaign's first six weeks can reset the learning phase and delay results. During this period the job is to watch and to check the budget cap only if there is a large swing, such as daily spend suddenly doubling.

After week six the picture is clearer: if cost per conversion sits under target, budget can be raised in steps; if it sits above, the goal and the assets need a second look. Automation reduces the workload when it is fed the right signal; when it is not, it just speeds up the mistake.

Launching Performance Max on a small budget looks appealing because one campaign promises reach across every channel; but a low daily budget also slows down how fast the algorithm can gather enough data to learn. Without a daily budget running to several multiples of the target cost per conversion, a campaign struggles to reach a meaningful learning pace.

For a small workshop business targeting only a handful of conversions a month, a narrow-budget Performance Max campaign still had not gathered enough data after six weeks; the business either raised the budget or went back to a narrower Search campaign. A small budget does not rule Performance Max out, but it does stretch out the patience required.

As automation grows, the idea that 'the system handles it, so it does not matter who manages it' spreads too; but someone still needs to read the report, update the negative list, and check the brand safety settings. That person does not have to be an outside agency, but they do need to give it regular time.

In a small business, one person checking the account for even one hour a week consistently gets a better result than a team glancing at it randomly once a month; the real difference is not expertise, it is consistency. The advantage of working with an agency is that it guarantees that consistency and carries experience gathered across multiple accounts.

For a residential property management firm, the account was managed in-house at first, and three months in it turned out the report had never actually been read; it was handed over to an agency after that. The real problem was not a lack of expertise, it was a lack of time. The question to ask when deciding who manages it is whether that person can genuinely give it weekly time.