The 4 decisions that actually move the needle
GEO optimization can turn into an endless to-do list. What actually changes your visibility is four decisions — everything else is accessory or, worse, noise.
- Define your entity precisely and unify it across every channel. Without a clear entity, the model doesn't know who to recommend.
- Reformat your key pages into self-contained question-answer blocks. This is what decides whether you get quoted literally or not.
- Implement advanced schema (FAQPage, HowTo, Article, Organization). The lever with the best impact/effort ratio.
- Build external authority with real citations in sector outlets. The hardest one for competitors to imitate, which makes it the most defensive.
Blacklist: tactics that do NOT move the needle
If anyone's selling you any of these, they're selling smoke:
- Invisible text or "hidden prompts" in your HTML. Detected and penalized.
- Buying mass mentions on low-quality sites. Dilutes authority and damages your entity.
- Optimizing for vanity prompts ("the best company in the world"). Doesn't bring customers.
- Paying obscure "AI directories" to appear. 95% bring nothing.
- Generating bulk AI content and publishing it unedited. Models already detect low-value generative patterns.
- Monthly strategy pivots. GEO needs 3-6 months to show real traction. Changing earlier is measuring nothing.
How GEO is really measured (vs. vanity metrics)
| Serious metric | Vanity metric usually offered |
|---|---|
| Weekly mention rate | One-off screenshots of "we appeared in this prompt" |
| Share of voice vs. competitors | "We are mentioned by ChatGPT" |
| Sentiment of the cited fragment | Aggregate volume with no context |
| Coverage of real commercial prompts | Coverage of irrelevant prompts |
| 8-12 week trend | Result from a single week |
Tools: what to pay for, what to avoid
The market is saturated with tools labeled "GEO" — most are classic SEO with a repainted dashboard.
- What to pay for: a tool that monitors real prompts against several engines, weekly minimum, with competitor comparison. Reasonable cost: €79-600/mo SMB, €1,500-3,000/mo mid-market.
- What to avoid: €2,000 one-off "GEO audits" that ship a PDF. GEO is a continuous cycle; the PDF isn't a deliverable, it's an Excel with a cover.
- Red flags: "proprietary framework", "AI score" with no methodology explained, pricing that doesn't match the actual compute cost.
Realistic results timeline
| Timeframe | What to expect |
|---|---|
| Weeks 1-4 | Baseline + first technical fixes. Zero visible impact yet. |
| Weeks 5-12 | First sporadic mentions. Mention rate 5-15%. |
| Month 4-6 | Visible traction. Rate 15-30%, first top-3 positions in category. |
| Month 6-12 | Consolidation. 30-50% mention if the category is accessible; less if highly competed. |
| Month 12+ | Maintenance + defense against competition. |