Why 37% of AI Systems Disagree on Brand Recommendations: Insights from 3,221 Prompts

Why 37% of AI Systems Disagree on Brand Recommendations: Insights from 3,221 Prompts

Here’s a simple question: can your business trust AI to put your brand in front of customers? Now complicate it, because it’s not just one AI system making those recommendations. It’s at least four major players, all with different strengths, behaviours, and biases. And according to our data, they disagree 37% of the time on which brands to recommend. That’s huge.

It’s tempting to think, “Well, surely Google AI Overview and ChatGPT will give me roughly the same results, right?” Nope. If your brand is visible on one, it’s far from guaranteed to appear on the other. Worse, you might be strong on a system customers aren’t using and entirely absent on one they rely on daily.

This is a commercial risk, plain and simple. But it’s also a massive opportunity if you’re prepared to optimise strategically. Let’s unpack what’s going on, and what you need to do about it.

The 37% Divergence Problem

First, some context. Across 3,221 prompts tested in our network, we saw a 37% divergence rate between ChatGPT and Google AI Overview. That means more than a third of the time, the two systems gave different brand recommendations for identical queries. For businesses that rely on AI visibility to capture attention at key decision points, this should be a massive wake-up call.

Here’s the breakdown across our dataset:

• ChatGPT had the highest overall mention rate at 74.5%.
• Google AI Overview lagged behind at just 45.9%.
• Perplexity and Claude sat in the middle, with mention rates of 53.8% and 60.4%, respectively.

That’s not just variance; it’s a visibility gap. Related reading: our article on the cross-provider visibility gap: why 39% of ai systems .... Related reading: our article on the psychology of ai recommendations: how cognitive bias .... If you’re leaning on one AI system and ignoring others, you’re leaving a lot of potential customers in the dark.

What’s Driving the Disparity?

It’s easy to blame the usual suspects: algorithms, training data, language models. But let’s dig deeper into how these systems work and why they might give different answers to the same query.

1. Different Training Data

AI systems are only as good as the data they’re trained on. ChatGPT, for example, benefits from OpenAI’s expansive training dataset, which includes diverse sources across the web, literature, and more. Google AI Overview, on the other hand, leans heavily on Google’s existing ecosystem, so it’s more likely to pull from indexed websites and prioritise SEO-heavy brands.

Here’s the problem: not all brands put equal effort into every channel. If your website ranks well but you’ve neglected broader AI-friendly content like FAQs, whitepapers, or conversational materials, you might do well on Google AI but struggle on ChatGPT. And vice versa.

2. Varying Prompt Sensitivity

Not all queries are created equal. Our testing lab ran 58 experiments across 14 businesses, and one of the standout findings came from our buying-intent tests: brand visibility changes dramatically based on how questions are phrased.

For example, a query like “What are the best fuel polishing solutions?” might recommend one set of brands, while “Who sells reliable fuel polishing machines?” highlights completely different ones. You can visibility tracking tools to see for yourself. You can how AI visibility monitoring works to see for yourself. It’s not rocket science, but it does expose a frustrating reality: even tiny differences in how customers ask questions can make or break your visibility.

3. AI Model Updates

AI systems evolve constantly. New updates, feature integrations, and parameter tweaks can suddenly shift what these models prioritise. In our dataset, we’ve seen instances where a brand’s mention rate drops overnight, purely because of a model update. If you’re not keeping track of these changes, you’re essentially flying blind.

4. Provider-Specific Biases

This one’s a bit controversial, but it shouldn’t be ignored. Each AI provider has its own agenda. Google, for instance, has a vested interest in promoting advertising-driven content or products that align with its search ecosystem. OpenAI’s ChatGPT might lean towards sources that align with broader conversational patterns. These biases aren’t malicious, but they shape outcomes in ways you can’t afford to ignore.

The Impact on Businesses

Now let’s get practical. What does all this mean for real businesses competing for attention in an AI-driven world?

First, if your brand isn’t mentioned consistently across multiple AI systems, you’re losing out on potential customers. Of the 35 brands we track, only 18 achieved a high visibility rate (above 70%). Meanwhile, six brands, including Castle Pumps and TOCA Social, had visibility rates below 30%. Imagine being invisible to 70% of queries in critical AI systems. You wouldn’t accept that on Google search, so why tolerate it here?

Second, even within high-performing brands, the disparity can hurt. A client I worked with recently was thrilled to see their 90% mention rate on ChatGPT, until we pointed out they barely hit 50% on Google AI. The result? They were missing key customer segments who used different systems to search for solutions.

And lastly, let’s talk competition. In the same space, we’ve seen brands using targeted optimisation to displace rivals entirely. If your competitor scores 100% visibility across systems while you’re stuck in single digits, the game’s already over. They’ve won.

How to Bridge the Gap

So how do you ensure your brand shows up consistently, no matter which AI system a customer uses? It’s not easy, but it’s doable if you’re strategic. Here’s what I’ve seen work best for clients tackling this challenge.

1. Run Cross-Provider Audits

You can’t fix what you’re not measuring. Start by auditing your brand’s visibility across systems like ChatGPT, Google AI Overview, Perplexity, and Claude. Tools like Contxt make this easier by centralising data from multiple platforms. I've written about this separately in this piece about open-source vs commercial llm visibility tools: lessons l.... In our dataset, this kind of cross-provider analysis revealed critical gaps for over 70% of tracked brands.

Focus on questions that matter most to your industry: buying-intent queries, competitor comparisons, and niche-specific searches. Be thorough here, the devil’s in the details.

2. Optimise Content for AI Systems

Businesses spend years refining SEO strategies but often ignore how AI systems interact with their content. It’s not just about keywords anymore; it’s about creating materials that AI understands, trusts, and prioritises.

• Use natural language in FAQs and product descriptions to align with conversational AI patterns.
• Publish authoritative content that attracts citations from other sources (AI loves reliable references).
• Regularly test and tweak your prompts to see what works best for each system.

In our experiments, brands that actively adjusted their content saw visibility gains in under six months. It’s a strategy that pays off.

3. Stay Ahead of Model Updates

This one’s non-negotiable. AI systems change fast, and if you’re not monitoring updates, you’ll fall behind. You might also want to check out our article on emerging trends in ai visibility: what to expect in 2026. I've written about this separately in our article on ai visibility deadlocks: why competing engines could be s.... Incorporate regular testing runs into your marketing operations to spot trends and recalibrate as necessary.

For instance, our model-update-monitor experiments flagged a shift in how recommendations were being weighted by ChatGPT earlier this year. Brands that adjusted quickly reaped the benefits, while others struggled to recover lost ground.

4. Test Competitor Displacement

Make no mistake: your competitors are after the same visibility you are. One effective approach is to actively test competitor-displacement strategies. Can you create better content? Optimise for queries where they’re dominant? Outperform them on customer reviews or product attributes?

Our data shows that displacement works. In one case, a mid-sized brand in residential technology increased its visibility by targeting specific competitor-focused prompts. The result? A 300% jump in overall impressions across systems within three months.

The Future of AI Visibility

Looking forward, it’s clear that AI systems will only grow in influence. What we’re seeing now with ChatGPT, Google AI Overview, and others is just the start. As these platforms evolve, they’ll become even more embedded in consumer decision-making. That’s both exciting and terrifying.

For businesses, the challenge will be keeping up. You’ll need to stay agile, track performance obsessively, and adopt new tools as they emerge. AI visibility isn’t static, it’s a moving target. But the brands that invest in it now will reap the rewards later. No question about that.

My Take

Honestly, I think many businesses still underestimate the importance of AI brand recommendations. It’s not just a niche concern anymore; it’s core to how customers find and choose products. If you’re not testing, optimising, and experimenting across systems, you’re basically leaving your visibility to chance. And in 2023, that’s commercial suicide.

The good news? Tools like Contxt are making it easier to get a handle on this. You don’t need a team of data scientists to figure out where your brand stands, you just need to start looking. The sooner, the better.

What’s Next?

First step: audit your visibility. Run some basic tests across different AI systems and see where you stand. If you’re not sure where to begin, we can help. From there, it’s about creating an action plan, a practical, data-driven roadmap to bridge the gaps and ensure consistent brand visibility wherever your customers are searching.

But don’t wait. AI systems are evolving, and every day you delay is a day someone else gets discovered by the customers you’re trying to reach.

Are your competitors already ahead of you? Time to find out.