- GBP Category: Not what is technically accurate, but what is driving discovery in this market right now. That’s no small lever, either: Whitespark’s 2026 survey ranks the primary category as the single highest-weighted relevance factor in the entire local algorithm.
- City and service area: The largest nearby city is rarely the right answer.
- Homepage intent: What people actually come for, not just what the company technically is.
- Internal pages for secondary markets: Don’t let the homepage compete with itself.
- Content based on real customer language: extracted from reviews, not generic texts.
- Prioritization corresponds to reliable local ranking factors, not what is easiest to automate.
Success must also mean something different. Metrics must reflect business results, GBP actions, calls, directions requests, conversions, not just traffic. If your system doesn’t report them, it’s optimizing for the wrong thing.
Use AI to augment analytics, not replace decision making. It is excellent at finding models in thousands of locations. It won’t tell you why a specific restaurant needs to focus on breakfast. That’s not a knock on the tools. OpenAI research found that models are trained and classified to reward a confident guess rather than admit uncertainty, so a wrong answer often seems exactly as confident as a right answer.
The real problem is poor suggestions and poor process design, not the tools. Tools do whatever you design; if the project removes judgment at the wrong step, the tools perform that removal at scale.
The building systems that rely on your judgment follow the same form every time:
- Do it manually first
- Apply your why
- Automate only clear logic
- Set a confidence threshold
- Validate against your own judgment on a regular basis
The last step is the one that people skip and it’s the one that catches the drift before the customer does.
If you can’t evaluate a step independently, you can’t automate it responsibly. This is the criterion by which I judge it, a test borrowed from a Wired series in which physicists explain quantum physics to a child, a teenager, a university student and a peer. If you can’t explain a decision at that range of levels, you can’t automate it, because you won’t see where it’s being misunderstood until it already has been.
