When should you use separate URLs for facts and examples
Split them only when examples keep breaking retrieval
By Janis Plume, Founder, Outbound Pros · 9 min read · 2026-10-16
Quick answer
Use separate URLs when stable facts are getting buried inside stories, screenshots, case examples, or narrative detail that changes often. Keep facts and examples together when the example is necessary to interpret the claim or apply it safely. The test is simple, if an AI assistant should quote one clean answer without dragging in the anecdote, split the URLs. If the fact becomes misleading without the example, keep one page and structure it better.
What problem do separate fact and example URLs actually solve?
They solve retrieval conflict. A lot of pages try to do three jobs at once, define a rule, prove it with an example, and tell a persuasive story around it. Humans can usually separate those layers. AI systems often compress them together.
That is where citation quality drops. The model finds the vivid example, the quoted customer line, or the edge case, then treats it like the primary fact. You end up with answers that are directionally right but operationally sloppy.
A separate fact URL gives one canonical place for the clean claim, the qualifier, and the current status. A separate example URL lets you keep the supporting proof, walkthrough, or scenario without forcing the extraction system to untangle it every time.
This matters even more on sites that lean on client side rendering or interaction heavy layouts. Verified server log work showed AI crawlers fetch JavaScript files and never run them. So if your examples sit in toggles, tabs, or app style components, the system may see fragments instead of the intended hierarchy.
If rendering is part of your problem, start with this breakdown of crawler behavior.
When should you keep facts and examples on the same URL?
Keep them together when the example is doing interpretive work, not decorative work.
- Keep one URL when the fact is easy to misapply without a concrete scenario.
- Keep one URL when the example contains the qualifier that makes the fact true.
- Keep one URL when users arrive with mixed intent and need the rule and application in one session.
- Keep one URL when the page is short, stable, and structurally obvious.
- Keep one URL when splitting would create near duplicate intros and weak standalone pages.
A common mistake is splitting too early. Teams hear that concise fact pages are better for AI extraction, then they atomize the site into thin pages with no supporting context. That creates a different mess. Now the model has a clean statement on one URL, a necessary caveat on another, and a contradictory old example on a third.
If the example is required to stop misuse, keep it close. Put the direct answer first, then the example underneath. In practice, many pages do not need a split. They need tighter structure.
That is the same principle behind answer first page design.
When does splitting into separate URLs help AI retrieval?
Split when your stable claim and your changing proof have different update rhythms.
For example, a methodology page may contain a durable definition, but the screenshots, test cases, and examples change as tools change. If you keep editing the whole page to refresh examples, the durable statement keeps moving inside the document. That increases the chance of partial or stale retrieval.
A dedicated fact URL is better when you need one clean source of truth for a claim that should be repeated consistently across your site. The example URL can then expand, compare, narrate, and age naturally without destabilizing the source statement.
- Split when the fact should be cited verbatim.
- Split when examples are long and push the main claim far down the page.
- Split when examples are likely to age faster than the rule.
- Split when one fact serves many example pages.
- Split when the example includes brand specific context but the fact is general.
This is especially useful if your site has recurring educational patterns, such as one rules page and several scenario pages. You can maintain a single canonical fact and let multiple example pages point back to it.
How should you decide, split or restructure?
| Situation | Better move | Why |
|---|---|---|
| The fact is short, stable, and gets lost inside a long story | Split into separate URLs | The claim needs a clean canonical source |
| The example contains the caveat that keeps the fact honest | Keep one URL | Separating them increases misquotation risk |
| Examples change often but the rule rarely changes | Split into separate URLs | Different update cycles create retrieval noise |
| The page is already concise and well structured | Keep one URL | A split adds complexity without retrieval gain |
| Several pages repeat the same fact with different anecdotes | Create one fact URL and link example pages to it | This reduces contradiction across the site |
| The issue is hidden content or weak headings, not page scope | Restructure the existing URL | Architecture is not a substitute for basic extractability |
That last row matters. People often treat URL strategy like a magic fix for weak page composition. It is not. If the real issue is that the answer lives under vague headings, inside accordions, or far away from its qualifiers, splitting the page may just produce two mediocre URLs.
What should a fact URL include if you create one?
Keep it brutally plain. State the fact early. Add the qualifier in the same sentence or the next one. Make the scope obvious. Say what changed, if anything. Then point to examples as supporting material, not as the container for the main claim.
- A direct answer near the top of the page
- The exact qualifier attached to the claim
- A clear date or current status when recency matters
- Visible language that distinguishes fact from opinion
- Links to example pages that show application, edge cases, or proof
Do not stuff the fact page with decorative intro copy. Do not force the user through a brand story before the answer. Do not make the examples the only place where the practical caveat appears.
On this topic, llms.txt is not the lever. Google states it is not used by Search, and the SE Ranking study found 10.13% adoption across a large domain set, with 0% among the top 1,000 sites and no citation lift after controls. If your fact and example structure is confusing, a separate text file will not rescue it.
If you are still weighing that file, read our plain view on llms.txt.
What should an example URL do differently?
An example URL earns its keep by showing application, not by restating the full rule in a messier way.
Use it for before and after structure, worked examples, screenshots, edge cases, decision logic, and failure modes. Link back to the canonical fact URL when the reader needs the stable statement. That relationship tells both humans and machines what is foundational and what is illustrative.
This is where founder honesty matters. If the example is unusual, say so. If it came from a specific platform or content model, say so. If it fails outside that context, say so. A lot of citation errors start because publishers present one vivid case like a universal law.
Where does this advice fail?
It fails on small sites with weak authority and thin editorial discipline. Splitting pages creates more maintenance work, more internal linking decisions, and more opportunities for contradiction. If you cannot keep the fact page and its example pages aligned, one page is safer.
It also fails when the main issue is not URL design at all. If your site is hard to crawl, client rendered, full of hidden text, or inconsistent across duplicate statements, fix those foundations first. Separate URLs cannot compensate for invisible content or conflicting claims.
And it fails for topics where interpretation is the product. If you publish nuanced expert judgment, separating facts from examples too aggressively can flatten the thinking. The result may be more quotable but less true.
Who should not follow this advice? Teams publishing light editorial content with no recurring fact base. Also teams that will split pages for SEO volume rather than clarity. If the goal is to manufacture more URLs instead of cleaner retrieval, skip it.
What is the practical workflow I would use?
I would start with an extraction audit, not a sitemap brainstorm. Pick the claims you want cited. Check whether they currently appear in one stable, visible, directly worded location. Then inspect the examples around them.
- Identify the claims that must remain stable across the site.
- Mark which surrounding examples change frequently.
- Test whether the claim is understandable without the example.
- If yes, create a fact URL and keep examples separate.
- If no, keep one URL and move the answer plus qualifier higher.
- Review internal links so example pages consistently point to the canonical fact source.
If your team also needs outbound execution after the demand capture side is fixed, that belongs with our sibling brand, not here. Outbound Pros covers managed outbound execution in detail. Inbound Pros should stay on extractability, retrieval, and citation mechanics.
If you want operator help reviewing your AI retrieval issues, you can book a working session here.
Common questions
Should every fact get its own URL?
No. Only separate a fact when it needs a stable canonical home and the surrounding examples are making retrieval worse. Many pages work fine as one URL if the answer is clear and the qualifier stays attached.
What is the biggest sign that I should split facts from examples?
The strongest sign is that assistants keep pulling the anecdote, edge case, or proof detail instead of the core claim. That usually means the page is mixing stable facts and variable illustration too tightly.
Can schema solve this instead of separate URLs?
Usually not by itself. Schema can add clarity, but if the visible page structure is messy or the key claim is buried in narrative, the retrieval problem remains. Clean copy and clear page roles come first.
Should the fact page link to every example page?
Only when it helps navigation and understanding. A fact page should stay concise. Link to the most useful examples, not every possible variation.
Does llms.txt help models understand which page is the fact page?
Not in a way you should rely on. Google says llms.txt is not used by Search, and the available study found adoption but no citation lift after controls. Use page structure and linking, not wishful signaling.
Last updated: 2026-10-16
Talk through your AI visibility
with people who measure it
30 minutes. We will look at what assistants can actually retrieve from your site and tell you plainly what is worth fixing first.
30 minutes, no obligation. The calendar shows real availability.