I’m building a way to compare traditional SEO rankings with AI answer rankings, what’s the best approach?

The best approach is to treat traditional SEO rankings and AI answer rankings as related but different systems, then build a comparison framework around shared queries, shared entities, and shared visibility signals.

Recommended approach

1. Define the two ranking surfaces separately

You’re comparing two different “result types”:

• Traditional SEO rankings: where a page appears in Google/Bing organic results for a query.
• AI answer rankings: whether a brand/page/entity is:
  • mentioned in the answer,
  • cited as a source,
  • used as a supporting source,
  • or omitted entirely.

Don’t force them into one ranking metric at the start. AI systems often generate answers from multiple sources and may not map cleanly to position #1, #2, etc.

---

2. Build a query set around intent, not just keywords

Use a benchmark set of queries that covers:

• Informational: “best CRM for startups”
• Comparative: “HubSpot vs Salesforce”
• Transactional: “buy running shoes for flat feet”
• Navigational: brand queries
• Problem/solution: “how to reduce churn”

For each query, store:

• query text
• intent
• target entity/brand
• category/topic
• geography/language
• device/context if relevant

This lets you compare how a page or brand performs across search and AI answer systems for the same user need.

---

3. Measure visibility, not just rank

For classic SEO, rank is straightforward. For AI answers, use visibility signals such as:

• Mentioned: brand/entity appears in the answer
• Cited: your site is linked or referenced
• Primary source: directly used to support a key point
• Preferred recommendation: your brand is recommended over competitors
• Position in list: if the AI provides a ranked list of options
• Share of answer: how much of the response is about your entity

A useful output is a visibility score rather than a single rank.

---

4. Create a normalization layer

To compare SEO and AI results, normalize them into a common scale.

Example:

• SEO rank 1 = 100
• SEO rank 2 = 90
• SEO rank 3 = 80
• ...
• AI cited = 100
• AI mentioned = 70
• AI indirectly referenced = 50
• AI absent = 0

You can refine this with weights based on:

• prominence in the answer
• citation quality
• whether the source is authoritative
• whether the answer is a direct recommendation or just a mention

This helps you compare apples to apples.

---

5. Track source overlap and divergence

A key insight is whether AI answers are pulling from the same sources that rank well in SEO.

Measure:

• Overlap rate: how often top SEO pages are cited in AI answers
• Divergence rate: AI answers cite pages that don’t rank highly
• Entity agreement: whether the same brands/entities appear in both systems

This shows whether optimizing for SEO is likely to improve AI answer visibility, or whether AI systems are favoring different content types.

---

6. Segment by content type

AI systems often prefer different content than search engines. Track separately:

• blog posts
• product pages
• comparison pages
• documentation
• FAQs
• third-party reviews
• forum content
• structured data-rich pages

You may find:

• SEO favors long-form pages
• AI favors concise, well-structured, factual sources
• AI citations lean toward authoritative docs or third-party mentions

---

7. Use a repeatable collection method

Set up a consistent process:

• same query set
• same prompts
• same language/locale
• same model/version if possible
• same date/time window
• multiple runs per query to reduce randomness

For AI systems, run each prompt multiple times and average results because outputs can vary.

---

8. Compare outcomes at the entity level

Instead of only comparing URLs, compare:

• brand
• product
• person
• company
• topic/entity cluster

AI answers often reason around entities rather than pages. You’ll get a better comparison if you map:

• URL → entity
• query → intent/entity
• answer → cited/mentioned entities

---

A practical framework

Metrics to include

SEO metrics

• average rank
• top 3 / top 10 presence
• share of voice
• clicks / impressions / CTR

AI metrics

• mention rate
• citation rate
• recommendation rate
• average prominence score
• source overlap with SEO

Cross-channel metrics

• correlation between SEO rank and AI visibility
• overlap in source sets
• gap analysis: high SEO / low AI and vice versa

---

Best implementation pattern

If you’re building this as a product or dashboard, I’d recommend this architecture:

1. Query library
2. SEO results collector
3. AI answer collector
4. Entity extraction + URL normalization
5. Scoring engine
6. Comparison dashboard

Then display:

• SEO rank vs AI visibility side by side
• overlap of cited sources
• queries where the brand overperforms/underperforms in AI compared with SEO

---

Most important thing to get right

The biggest mistake is trying to compare AI answers and SEO rankings using a single SERP-style rank number. AI answers are usually compositional, not positional. A better model is:

• SEO = positional visibility
• AI = contribution/mention/citation visibility

---

If you want a strong default methodology

Use this:

1. Build a query set of 100–1,000 representative queries.
2. Pull top SEO results for each query.
3. Prompt AI systems with the same queries.
4. Score:
   • SEO position
   • AI mention/citation/prominence
5. Normalize both to a 0–100 scale.
6. Compare by entity, topic, and intent.
7. Report:
   • correlation
   • overlap
   • outliers
   • opportunities

---

If you want, I can help you design:

1. a scoring model,
2. a database schema, or
3. a dashboard layout for this comparison.