Repositioning your business helps you stay relevant as markets shift.

For example, maybe you're a consultant who used to specialize in one service and have since expanded into a broader practice. Or maybe you're a company that started out serving small businesses and have since moved upmarket to enterprise.

Either way, repositioning takes time and costs money. So once you've done it, you need people to actually know about it.

But here's the catch: More people now use LLMs to learn about products, and services and if a chatbot repeats your old positioning, prospects will leave with the wrong idea of what you do and who you serve. And they may never visit your website to get the right information.

I call this the LLM positioning lag. It’s effectively when your positioning has changed, but LLMs haven't caught up. And as you'll see from real examples, this lag can affect businesses for longer than you might expect.

Today, we'll look at the consequences of a positioning lag, what causes it, walk through some real examples, and cover how to avoid it.

What happens when your positioning in LLMs is outdated?

When your positioning in LLMs is outdated, users asking about your industry or business won't get accurate information when AI-generated answers discuss your brand.

That's a bigger problem than it might sound: The portion of consumers who replaced search engines with generative AI for product and service recommendations went from 25% in 2023 to 58% in 2025, according to Capgemini. That means a growing share of your prospects are forming their impressions of you based on information that doesn't necessarily reflect your latest positioning.

Let’s say you used to sell accounting software for small businesses, but you've since pivoted to enterprises. If an enterprise buyer asks AI for the best accounting software and your brand is listed as a small-business tool, they’ll assume you're not built for their needs and skip you.

And it's not just first impressions on the line. Agentic commerce (the process of AI agents making purchases on behalf of users) is emerging. Without accurate information, you might lose out on sales.

Why LLMs get positioning information wrong

LLMs get positioning information wrong about your business for three main reasons:

Repositioning not reflected in the model’s training data

When your repositioning isn't reflected in the model's training data, the LLM is working from an outdated picture of your brand.

Rebrands, new product lines, and market expansions that happened after the training cutoff may not be in there at all.

A model can sometimes close the gap by searching the web — but only if it decides the question needs it, or if the user explicitly chooses the web search function. Left on autopilot, many models still default to training data.

Inconsistent information about your business

If different sources online describe your business differently, LLMs won’t know which source is correct.

Your own site might reflect your current positioning while your social media profiles reflect different positioning. Plus, third-party lists, old interviews, or review sites could also have outdated information. When an LLM encounters enough conflicting information, it has to decide which version to trust. And it might not be your version.

Years of old positioning outweigh a recent change

Years of old positioning can outweigh a recent change as it can be difficult to overcome an extensive record of your old brand.

LLMs learn from what has been previously published about your business. If you described yourself one way for a decade, that's years of potential reviews and authoritative citations that you must now override.

The longer the old narrative ran, the wider that gap. Your new positioning can be accurate and well documented and still be outnumbered. Until enough new content builds up, an LLM may keep leaning on the version it has seen most often.

Real examples of the LLM positioning lag

I decided to see what types of companies the positioning lag affected, and I found that it ranges from large enterprise products to individual consultants.

As the following examples show, even well-funded visibility efforts don't guarantee that an LLM will catch up to your new positioning right away.

Hotjar

Hotjar was a behavioral analytics tool (think heatmaps and session recordings) that Contentsquare acquired in 2021 and officially merged with by July 2025.

Yet, when I asked ChatGPT for the best heatmapping software, Hotjar was second on its list with no mention of Contentsquare.

Hotjar's website redirects to a page on Contentsquare's site, but the table above has no clickable link. Users are unlikely to find a link elsewhere, like in the search results, so few will reach Hotjar's old website organically. That means anyone asking this question walks away thinking Hotjar is still an independent product, with no idea it's now part of a bigger platform.

That's a brand awareness problem for Contentsquare, since Contentsquare paid to acquire Hotjar's name recognition and customer base.

But if ChatGPT keeps recommending Hotjar as its own thing, none of that awareness transfers — prospects comparing heatmapping tools within an LLM don’t learn about Contentsquare and all of the other tools they offer.

Contentsquare even made the acquisition highly visible across its website and communications. Since 2021, some of their publishings include:

Multiple press releases announcing the acquisition
Multiple blog posts detailing the acquisition
Dedicated support documents for Hotjar customers
Social media content like YouTube videos
A landing page for those searching for Hotjar

However, ChatGPT still references Hotjar as an independent product with no mention of Contentsquare more than one year after the official merger.

Blue Apron

Blue Apron is a meal delivery service that was purely subscription based for over a decade, but they reframed in August 2025 and now include a pay-as-you-go model along with a subscription to appeal to a broader market.

However, ChatGPT didn’t mention the pay-as-you-go model when I asked about Blue Apron.

To announce the end of requiring a subscription, Blue Apron:

Made the official announcement on a national talkshow (Good Morning America) that was then picked up by major news outlets like ABC News, Yahoo, and CNET
Produced commercials to promote the new model
Provided interviews to the press
Repositioned the messaging on their homepage

That's a national press push, paid media, and a homepage rewrite — about as much visibility as a repositioning can get. Yet ChatGPT doesn’t mention that subscriptions are no longer required, which will alienate potential customers who don’t want to subscribe.

Duolingo

Duolingo is a learning app that initially focused solely on teaching users new languages. Today, users can learn math, music, chess, and over 40 different languages.

I asked ChatGPT for the best apps to learn something new. It recommended Duolingo only for language learning, and suggested other platforms for subjects like math and science.

Duolingo introduced math in 2022, music in 2023, and chess in 2025. During that time, Duolingo has:

Made product announcements at Duocon (their free virtual event)
Published blog posts about their new products
Updated their app description in the app store to include mentions of math, chess, and music
Reworked their homepage to mention chess and math
Published press releases for new products

That's an ongoing campaign across multiple product launches. Yet ChatGPT still describes Duolingo as a language app only, the same positioning it held a decade ago.

Peep Laja

Peep Laja is an entrepreneur who founded CXL, Speero, and Wynter. Peep became widely known as a CRO expert through CXL, but he left CRO to focus on business strategy around 2016.

When I asked ChatGPT who I should follow to learn about CRO, Peep was the first person on the list, with no mention of his current company, Wynter.

Peep has also posted about this on LinkedIn. In fact, I was inspired to dig deeper into the positioning lag after seeing Peep’s post.

Entrepreneurs might not have the budget for an orchestrated PR campaign like Blue Apron, but Peep's digital footprint still shows a consistent, long-term effort toward repositioning:

His LinkedIn profile and personal site both list Wynter, not CXL, as his current company
He consistently publishes LinkedIn content on B2B strategy
He has published a newsletter and hosted a weekly podcast on B2B strategy — not CRO — for years
He's talked openly about moving on from CRO since 2016

Despite all of that, ChatGPT still leads with Peep’s decade-old reputation.

How to see what LLMs associate with your brand

See what LLMs associate with your brand by using a tool that analyzes AI answers at scale.

Semrush’s Perception report gives you a detailed analysis, so you know how different AI platforms perceive your brand.

Enter your website into the Perception report and scroll to the “AI Feature Descriptions” section to see what features AI associates with your brand. You can evaluate both non-branded and branded questions.

Click the number in the “Answers” column for a given row to read specific answers and uncover more details.

You’ll want to focus on correcting any information that isn’t accurate. We’ll show you how to do that in the following section.

How to keep brand information accurate in LLMs

Keeping brand information accurate in LLMs takes coordinated action across PR, content, and brand monitoring.

Step

What it fixes

Announce changes with PR

Creates a citable news event

Update old content

Removes outdated claims from your site and external sites

Keep messaging consistent site-wide

Minimizes contradictory signals

Keep business listings current

Aligns external sources with your new positioning

Structure company pages for LLMs

Makes new positioning easy for AI to parse

Monitor what LLMs say

Catches drift before it spreads

Announce changes with PR

Announcing changes with PR creates a citable news event that gives AI platforms a fresh, dated source to reference.

Your announcement should make the change explicit. Clearly state what your company does now, who it serves, and how that differs from your previous positioning. This gives AI systems a clear record of both the old and new versions of your business.

Here are some tips for :

Publish the press release on your own site. This gives you a source you control and ensures there's a canonical version of the announcement that clearly reflects the change.
Distribute your press release through a wire service. This gives other publications the opportunity to pick up and republish the announcement.
Ask trade publications to publish a story about your pivot. Reputable publications can reinforce your new positioning and carry topical authority for AI systems.
State your new positioning clearly and consistently. Don't make readers or journalists piece together what changed. Use the category, audience, and description you want associated with your brand.
Explain what changed and when. A clear timeline helps distinguish your current positioning from the one that appeared in older coverage.
Treat PR as the starting point, not the entire strategy. A new announcement creates a fresh signal, but it still has to compete with older content and third-party sources that reinforce your previous positioning.

Update old content

Updating old content removes outdated claims LLMs might still rely on.

When updating content to reflect your new positioning, make sure you:

Audit content for mentions of your old category, product name, or service model
Update or redirect pages that no longer reflect current positioning and add a short clarifying statement near the top of high-traffic legacy pages to ensure both readers and AI systems see it
Find authoritative sites still referencing your old positioning and request an update

One way to find outdated, authoritative mentions on other websites is by reviewing your backlinks (links from external sites pointing to your site).

Start by entering your domain into Semrush’s Backlinks tool.

Click “Backlinks.” Find any links that use outdated anchor text (the clickable part of a link), such as incorrect product names, by scanning the “Anchor and Target URL” column or using the search bar.

Prioritize updating links from sites with the highest authority first, since these have the greatest impact.

Keep messaging consistent site-wide

Keeping messaging consistent site-wide minimizes conflicting signals that can push LLMs toward outdated information.

Your website should give AI systems a clear, repeated description of what you do today. Not several versions of your business depending on which page they find.

Follow these tips:

Check your homepage, about page, product pages, and footer for outdated descriptions
Use consistent terminology for your category, products, and services. Don't use different terms for the same thing across different pages.
Update internal links and anchor text. Old terminology can continue reinforcing your previous positioning even after the destination page has changed.
Update your meta descriptions (text that describes your page) and title tags. Inconsistent metadata can send the same mixed signals as inconsistent copy.

Keep business listings current

Keeping business listings current matters because LLMs often reference external sources to get information about brands.

Make sure your business information is accurate everywhere you have a presence — Google Business Profile, LinkedIn, G2, Crunchbase, Wikipedia, social media, and any industry-specific directories.

Semrush's Listing Management tool handles this at scale. It syncs your business information across directories from one dashboard, so a positioning update doesn't mean logging into dozens of platforms separately.

Structure company pages for LLM extraction

Structuring company pages for LLM extraction makes your positioning easier for AI systems to understand, quote, and cite.

Apply these practices to your homepage, about page, product pages, and any other pages that contain information about your brand:

Keep sections self-contained — AI systems retrieve excerpts rather than full pages, so a section shouldn't depend on context from elsewhere on the page
Use headings that match how people actually ask about your business and products or services
Answer headings with the copy that follows

Here's a before and after of how heading plus body copy could look on an About page:

The second version names the category, states the current audience, and dates the shift — three things an LLM can pull directly into an answer. The first gives it nothing to cite.

Monitor what LLMs say about you

Monitoring what LLMs say about your brand on a recurring basis helps you spot inconsistencies before they spread.

Semrush's AI Visibility Toolkit lets you monitor what LLMs say about you and your competitors, so you know what’s incorrect and where competitors are earning mentions where you aren’t.

For example, the Visibility Overview report shows you how your mentions and citations are growing within LLMs. And the topics/prompts you appear for.

You can also use this report to see your “Topic opportunities” (the topics where your competitors are appearing but you aren’t).

Missing opportunities is just as bad for your brand as inaccurate information. It means you won't even be part of a relevant conversation.

Click “Monitor all prompts” to add the prompts you find to the Prompt Tracking tool, so you can track your performance.

For brands with multi-market or multi-brand needs, Semrush's Enterprise AIO offers monitoring that’s scalable and suited to complex needs.

Ready to show up accurately in LLMs?