Like your weird uncle, no one knows exactly how AI engines choose the sources they cite. But experiments are starting to reveal ways you can get on their radar.
And as consumers increasingly turn to AI search for product and service recommendations, you definitely want to be on their radar. (Ironically, unlike your weird uncle, who you avoid.)
Today I have such an experiment contributed to a 642% increase in citations from AI tools like ChatGPT.
And to the delight of all word nerds, it’s all about semantics. But first, everyone’s favorite part: the disclaimer!
The sum vs. the parts
Before moving forward, it’s important to know that this tactic is just one part of a larger playbook that our growth team affectionately calls the “Everything Bagel Strategy.”
“Our experiments did not show this one Tactics are key to better AI visibility,” says Amanda Sellers, head of EN blog strategy at HubSpot.This is what we found out the sum of the parts is what is good for AI visibility.”
But if I covered all of these parts at once, this would be a novel, not a newsletter – so think of it more like Part 1.
A little bit of why behind the AI
“A human could probably tell you what the phrase ‘Paris is cool’ means,” Sellers says. “But an AI engine without (immediate) context wouldn’t know whether it’s Paris, France, or Paris Hilton.”
AI tools can sound very human, but the way they understand language is very different than we do.
To stick with Sellers’ example about Paris, before you read it, know from the start whether an article you clicked on is about travel tips or celebrity gossip. This context would be all you need to understand the word “Paris”. AI models require a little more handling.
One way to coddle their cold, metallic hands is a framework called “semantic triples.”
As simply as I can explain it: Semantic triples are a writing pattern that creates context based on the sequence Subject – predicate – object.
If you’ve also pushed third-grade English out of your brain to make room for Lord of the Rings trivia, here’s a quick summary of what they mean:
- Theme: Who or what is a sentence about?
- Predicate: Information about (or the action of) the subject.
- Object: The noun or pronoun that receives this action.
A real-world marketing example might look like this: “HubSpot (subject) can automate email marketing (object) (predicate).”
With just one sentence, I can quickly guide a bot to connect HubSpot to email automation. Why is this important?
“We want HubSpot to be associated with ‘marketing automation’ so that someone asks ChatGPT, ‘What is the best marketing automation platform?’ We are mentioned in this conversation.”
Semantics in action
During the experiment, Sellers’ team took key information on pages that AI models should understand and rewrote it from paragraph format to a list of semantic triples.
Below is a snapshot of Current INBOUND presentation from the sellers This highlights what the content looked like before and after the changes.
In conjunction with the other “everything bagel” ingredients (like schema, backlinks, etc.) This tactic helped increase mentions of HubSpot in AI responses by 58%.and the number of times HubSpot pages were cited by AI increased by 642%.
To some of you, this might just sound like very simple good SEO, and you’re not wrong.
“A stable SEO foundation is very important for good LLM visibility. But even though semantic triples are beneficial for SEO, they are necessary for AEO.”
To others, it may sound like a really annoying content for a person to read. And you’re not entirely wrong either. If implemented poorly, semantic triples can read like the over-optimized garbage that dominated early search engine optimization.
Luckily, Sellers has shared some practical tips on how to effectively use semantic triples without effectively annoying your audience.
Triple tips
1. A little goes a long way.
“We need to find the middle ground between content that is easy to understand (through AI) and content that is still enjoyable for people to read.” Laughing, Sellers advises using the benchmark: “As a human being, would reading this cause me to throw my phone into the pool?”
Instead of cramming semantic triples all over the page, She suggests introducing a triple for each core concept.
2. Target humans and bots with the same content.
You might think you can eliminate the need for the first tip by simply writing separate content for AI engines and for your human audience. Seller advises against it.
If AI or search engine crawlers discover your human-targeted content, they may decide to punish you both Parts of the content due to excessive similarity.
But what’s even worse is when your human readers stumble upon your bot content. A reputation for having bad content is hard to shake.
“We really try to feed two birds with one scone because we have a huge readership who are really interested in what we write.”
3. Use answer-first phrasing.
Both humans and bots love to skim, and your content, as amazing as it may be, is no exception. Your job is to make sure they get important information quickly as they skim.
To this end, Sellers recommends using the “answer first” phrase.
So instead of a sentence like, “According to recent research, pizza is delicious,” you could rewrite it as, “According to recent research, pizza is delicious.”
A warning: both human and software editors I absolutely hate this. Do it anyway. This is a structure I insisted on when I led the HubSpot blog user acquisition program.
4. Don’t bury the lede.
Similar to placing important information at the beginning of a sentence, you also want to ensure that your semantic triples appear early in the paragraphs.
This also makes it easier for human skimmers to quickly get the information they are looking for. However, this is even more important for bots, as they often take parts of content out of context.
“Writers must pay careful attention to the order of sentences If an LLM comes and takes this one paragraph, that is enough to represent the idea.”
4. Think about middle-of-funnel and bottom-of-funnel content.
Product reviews, product comparisons, and listicles are all great places to use semantic triples. Readers expect this type of content to be simple and direct, so semantic triples don’t seem out of place.
It’s also a natural opportunity to associate your brand with a product category, with specific features, or even with your competitors.
“You want your entity to be linked to similar entities. For example, we want HubSpot to be linked to Salesforce or MailChimp. So when an AI engine mentions a competitor, it would be remiss not to mention us in the same breath.”
How to check your AI visibility with AEO Grader
If you’re not sure where you stand in the eyes of answer search engines, you can easily find out using HubSpot free AEO grader.
I sat down to write a guide for you and found it to be so simple that it would almost seem insulting.
Just enter four simple answers and you’ll be ranked for the three most popular AI search tools in areas like brand awareness, sentiment, and share of voice. You will then have the option to provide your email address to receive a detailed report with insights and recommendations.