AI systems have outdated information from a person’s job title to a brand’s leadership, a deal or offer available with a business, and incorrect product and service information. This is likely because there are multiple “trustworthy” resources out there giving conflicting information, so the AI system evaluates what they feel is the most accurate based on their findings.
While you cannot control outdated information like media coverage from one or five years ago, PDFs with citations, industry journals where they laid off the staff, and through abandoned niche sites; you can be proactive and work to surface accurate information. When AI has accurate information from “trustworthy” sources to share with users, your company or personal brand may see more accurate information surfacing.
When it does happen there’s less customer support complaints, a person is represented fairly for jobs or citations when the hiring manager asks about them, and products or services can be presented as solutions where others are not as strong.
There’s a few ways we’re attempting to do this.
Archives vs. Individual Pages and Posts
If you’re in the SaaS space, are publicly traded, or release a new products seasonally or annually, you have a ton of conflicting information. If they all exist as blog posts, individual pages within a series of folders, and as tags and categories with other topics, it can become hard to know exactly what the most recent and up-to-date information is. Even with schema on the pages and date modified in your sitemaps.
It applies to multiple situations like:
- Product updates and release notes
- Earnings reports for a publicly traded company
- Ownership records and pricing (real estate, etc…)
- Ride updates at theme parks
- Statistics like uptime/downtime, stock and currency prices, sports results
This is where creating an archive works to your advantage. Have the most recent version at the top, and then previous releases below and in archived pages. Make sure to not let them leak into other sections of your website so they stay siloed in the archive.
There’s also schema you can apply like ArchiveComponent to the pages as a subset of the original type. This specific library is not for mainstream usage yet. This can be combined with a proper meta robots, site structure like internal links, and other on-page SEO elements.
Pro-tip: Use meta robots with caution as you don’t want to block the person that asks “when did XYZ brand add ABC” from their query and get the wrong information.
Now it becomes easy for search engines, LLMs, creators, editors, and users to know the most recent updates while still being able to cite and reference things from the past.
Update Pages You Own
Set a reminder at least once a quarter to check and update all company information on your most important pages.
- Professional bios
- About us pages
- Author bios
- Product and service details and specs
- Service locations
- Customer support information including hours of operation, ways to reach you, and holiday information
While third parties are a good source of information, specific things like the people managing your company and their roles are in your control. The same making sure product pages say the product now works with X and Y models, but is no longer compatible with A and B.
Coming direct from the source may carry more weight than a third party saying it, and when you update a PDP (product display page) you may also update your datafeed which powers third party resources. This can be especially helpful if the LLM is pulling from a media company covering you a few years ago that uses the feed for affiliate monetization. The link from that page can be crawled to your PDP and the spider can access current and updated information.
More important is that LLMs have partnered with multiple ecommerce platforms like this one with ChatGPT with Shopify and ETSY to pull feeds in directly for instant checkout.
Pro-tip: This may become a huge advantage during peak seasons if you don’t have SEO traffic and cannot afford PPC if stock levels are full and your competitors are out. LLMs will likely be asked “where can I find XYZ in stock?” If they have your feed and you have stock but the sites in the top 10 of Google don’t, you can potentially get surfaced and the traffic while Google would still show the top 10 SEO results and you get nothing.
Third Party Signals and Controlling Narratives
Last is to constantly work with your affiliate manager, influencer lead, and PR professional to synchronize external talking points. Affiliates, influencers, and creators need to mesh your brand with their audience’s needs, that doesn’t change. What you can do is ask them to include specific talking points, and it is your job to say how it is relevant to their audience demographics.
By doing this the newest content that can be discovered by AI systems like ChatGPT, Gemini, and Claude will all be aligned with the same talking points on your own website. When this happens the RAG retrieval may realize the current known information is now out of date and it is time to update what it knows. It’ll be interesting to see how this evolves as these systems move away from RAG.
It is easy to jump into technical details and go ten layers deep. But in reality, this may be all that is needed. Think about how you can surface the most relevant and recent information so that LLMs will understand their own knowledge needs to be updated. It is keeping your website clean, making sure third parties are sharing the most accurate information, and when possible, having content changed on your own websites and others.
