Why Google Ads Still Wins (And What AI Gets Wrong About It)
AI tools are good at Google Ads execution and poor at strategy. Here's the three questions every advertiser needs to answer before touching the account, and why they matter more now.
What is happening
Every few months someone publishes a piece declaring that AI has made Google Ads obsolete. The algorithm manages bids. Smart campaigns write the ads. Performance Max does everything automatically. The argument is that if the machine handles execution, the strategist is redundant.
At the same time, AI tools like Claude and ChatGPT are getting genuinely good at PPC operations: pulling search term reports, writing ad variations, auditing account structure. I use them regularly. They are useful.
Neither development has made Google Ads strategy less important. They have made it more important.
What I learned from this
Automation optimises for the objective you give it, not the objective you meant to give it. Most advertisers do not know the difference until the results come in wrong.
The example I see most often: target ROAS set to 4x. The campaigns hit 4x. Revenue flatlines. When I dig in, the algorithm has learned to prioritise easy, low-value conversions from brand search terms - technically hitting the target, practically useless for growth. That is not an automation failure. That is a strategy failure that happened before anyone touched the account.
What AI tools are good at is everything that happens after you know what you are trying to do. Pulling reports, writing variations, checking structure. What they cannot do is the conversation you need to have before you open the account. The one where you figure out that the business’s real problem is not conversion rate - it is that they are targeting the wrong customer segment entirely. Or that their ROAS target is based on what they want rather than what their margins support. Or that their landing page would cause any sensible person to leave immediately.
Those problems require understanding the business, the market, and what success actually looks like for this specific company at this specific moment. AI can help you execute once you know that. It cannot tell you what to know.
What I recommend for your business
Before you touch the account, answer three questions in writing.
What does a successful customer actually look like for your business? Not the broadest possible audience - the specific person who buys, comes back, and refers others. That is your targeting brief.
What does a realistic ROAS or CPA target look like based on your actual margin and your actual conversion rate from the last 90 days? Not the number you want. The number the data supports.
What is the one thing you want someone to understand about your business after seeing one ad? Not three things. One.
If you cannot answer those three questions clearly, no amount of AI-assisted optimisation will produce the results you are looking for. The hard part has never been execution. It has always been knowing what to execute. Figure that out first and let the tools handle the rest.