SparkToro's January 2026 research found AI tools recommending defunct real estate offices, inactive social accounts, and businesses that had closed. Every one of those recommendations occupies a slot in an answer, and nobody is defending it.
Category answers are short. Three to five names, sometimes fewer. Which makes it worth knowing whether one of those names belongs to a company that shut down eighteen months ago.
When it does, you are looking at the cheapest opportunity available in AI visibility, and at a preview of what happens to your own record when you stop maintaining it.
Key Takeaways
- Category sets are built from accumulated discussion: which lags reality by years.
- Defunct entries take real slots: in answers that name only a handful of options.
- The slot has no defender: nobody is publishing, correcting, or earning coverage for a closed company.
- Check who is named, not just whether you are: most audits skip the competitor list entirely.
- It works in reverse too: the same mechanism preserves your discontinued products and old positioning.
Why Models Recommend Companies That Closed
A model assembling a category list is not consulting a registry of active businesses. It is drawing on accumulated discussion, and discussion about a company does not stop when the company does.
The archive keeps working. Review threads, comparison articles, forum recommendations, and roundup posts all persist, frequently with more specificity than anything written about a newer entrant. A company that operated for six years and closed last year has a deeper footprint than one that launched eight months ago and is thriving.
Nothing in that footprint carries a signal that the company stopped operating. Shutdown announcements are brief, poorly covered, and rarely propagate through the same channels as the years of discussion that preceded them.
Why This Is Worth Checking
Every established competitor in an answer is defending its position with content, coverage, and customers. A defunct one is defending nothing, and its slot is held entirely by inertia.
Displacing a live competitor means outperforming an organisation actively working to stay there. Displacing a ghost entry means being the more current, more specific, more corroborated answer to the same question, against an opponent that stopped competing.
The practical implication is about sequencing. If your category answers contain one or more defunct entries, that is where the first move sits, ahead of any attempt to displace an entrenched incumbent.
How to Find Them
Most visibility audits record whether you appear. Fewer record who else does, and almost none check whether those companies still operate.
Extract the full competitor list
Run your core category questions several times and record every company named, not only your own presence. Outputs vary between runs, and SparkToro's research suggests genuinely stable results require asking repeatedly rather than once, so a single run will miss entries that appear intermittently.
Check each name against reality
For every company you do not recognise or have not encountered recently, check whether it still operates, whether it was acquired, and whether it still serves the segment the answer implies. Acquisitions matter as much as closures, since a company absorbed into a larger one is frequently still described as independent.
Note which sources produced them
When a defunct company appears, identify the source. That page or thread is usually still ranking, still cited, and still shaping the category set. Sources differ by platform, since engines draw on largely different sets, so an entry that appears in one may be absent in another.
What the Findings Usually Mean
| What you find | What it tells you |
|---|---|
| A closed company named | Slot held by inertia; cheapest displacement available |
| An acquired company named as independent | Category set built from pre-acquisition sources |
| A company that pivoted out of the category | Framing based on their old positioning, not current |
| Companies you have never encountered | Either category framing is wrong, or your competitive map is |
| Only current, active competitors | Category set is well maintained; expect a harder contest |
The fourth row is worth pausing on. Consistently seeing companies you do not recognise usually means the model is answering a slightly different question than you think, which is a positioning problem rather than a competitive one.
How to Take the Slot
The mechanism is the same as any citation work, with the difference that nothing is pushing back.
The sources naming the defunct company are usually roundups and comparison articles. Those pages age, and their authors are frequently open to updating them, particularly when informed that a listed company no longer exists. This is unglamorous outreach with an unusually high response rate, because the correction serves the publisher's interest as much as yours.
Where the source is a community thread rather than an article, a dated reply noting the closure and offering current alternatives becomes part of what models retrieve. Removal is neither available nor necessary.
And where no current alternative to the defunct company is well documented anywhere, that gap is the opportunity. A specific, current, well-structured resource answering the question the closed company used to answer has very little competing with it.
The Same Mechanism, Pointed at You
The reason ghost entries persist is the reason your own outdated record persists, and the second is more expensive.
Discontinued products, superseded pricing, abandoned positioning, and old customer lists survive in exactly the same way, because the sources describing them were never updated and nothing signals that anything changed. A company that pivoted two years ago is frequently still described by its original category.
This is the practical case for treating the external record as something you maintain rather than something that maintains itself, which is what our work on citation freshness examines in more detail.
Adding It to a Standard Audit
This is a small addition to work most teams already run, and it produces a different kind of finding.
- Record every company named, not only whether you appear.
- Verify current status for any name you do not recognise.
- Trace the source behind each unfamiliar or defunct entry.
- Prioritise displacement targets by how actively they are defended.
That last point reframes what a competitive analysis is for. Difficulty in AI visibility is not primarily about how good a competitor's product is; it is about how much current, corroborated material exists supporting their position, and that is what visibility tracking should be recording alongside your own presence.
Why It Gets Missed
Two reasons, both structural.
Visibility tooling reports on the brand you configured, because that is what it was built to do. The competitor list is context in the interface rather than the subject of the report.
And the finding requires manual verification. No dashboard knows that a company shut down last spring, which means someone has to look at the names and check. It is an afternoon of work that most programmes never schedule, and it occasionally surfaces the easiest win available.
FAQs
1. Why do AI models recommend companies that no longer exist?
Because category lists are assembled from accumulated discussion rather than from a registry of active businesses. Years of reviews, comparisons, and forum threads persist after a company closes, while shutdown announcements are brief and rarely propagate through the same channels.
2. How do I find out if this is happening in my category?
Run your core category questions several times and record every company named, not only whether you appear. Then verify current status for any name you do not recognise, checking for closures, acquisitions, and pivots out of the category.
3. Why is displacing a defunct competitor easier?
Because nothing is defending the position. A live competitor publishes, earns coverage, and acquires customers who discuss them. A closed company's slot is held entirely by archived material that will not be reinforced.
4. What do I do about a source that lists a closed company?
For roundup and comparison articles, contact the author; response rates are unusually high because correcting a defunct listing serves the publisher too. For community threads, add a dated reply noting the closure and current alternatives, which becomes part of what models retrieve.
5. Does this mean my own outdated information persists the same way?
Yes, through exactly the same mechanism. Discontinued products, old pricing, and abandoned positioning survive because the sources describing them were never updated and nothing signals that anything changed.
References include SparkToro's January 2026 research into recommendation consistency. Prevalence of defunct entries varies considerably by category, and is generally higher in sectors with high business turnover or recent consolidation.