How We’re Using LLMLens to Grow Ideal Heating’s AI Visibility

So, Where Was Ideal Heating Actually Showing Up?

Before we could decide where to focus, we needed to understand what AI platforms already seemed to know about Ideal Heating, and perhaps more importantly, where competitors were being chosen instead.

The initial LLMLens analysis gave us a much more detailed picture than a single share of voice figure. We could see the prompts Ideal was appearing for, which brands were being recommended alongside them, and the topics where the brand already had a stronger foothold. Several clear themes emerged that weren’t just areas where the brand happened to get a mention; they indicated the expertise and credentials LLMs were already connecting with Ideal.

By cross-referencing what LLMLens was showing us with Ideal’s existing website, we identified several areas where people were actively asking questions, Ideal was relevant to the answer, but competitors such as Worcester Bosch and Vaillant were more likely to be recommended.

That included direct brand comparisons, boiler warranties, reliability, UK manufacturing, the best combi boilers for three-bedroom homes, and customer aftercare. Heat pumps stood out too, with Ideal’s Logic Air significantly less visible than established names such as Daikin, Mitsubishi, NIBE and Samsung.

This gave us something much more useful than “we need to improve AI visibility.” It showed us where Ideal already had authority to build from, where that authority wasn’t translating into recommendations and which gaps were worth investigating further.

The next job was working out which of those opportunities actually deserved a place in the strategy.

Has It Actually Worked?

The short answer is yes, and we can see that growth in both AI visibility and traditional search performance.

When we began tracking Ideal Heating through LLMLens in May 2026, the brand held an 11.5% AI share of voice, with 2,737 mentions. By September, that had increased to 11.9% and 6,421 mentions, with Ideal’s trajectory beginning to close the gap on Baxi in mention volume. Our tracking had expanded too, from 1,050 to 1,634 prompts across eight LLMs, giving us a much broader view of where and how the brand was appearing.

And there were some particularly strong signals within that wider growth. On GPT-5.1, Ideal reached 6.57 mentions per prompt, the highest citation rate of any brand on any model within our dataset. In other words, we’re not just seeing Ideal mentioned more often, we’re also seeing areas where the brand is becoming a particularly strong source within AI-generated answers.

But remember that bit about not optimising for AI in isolation? This is where the Google results become particularly interesting.

The content strategy informed by that wider analysis hasn’t been about creating individual pages for individual prompts. We’ve been strengthening existing content, filling genuine gaps and building out wider topic clusters, including the combi boiler cluster we touched on earlier.

Since July, 30 pieces of optimised content have gone live, with the combi boiler content becoming one of the biggest ranking drivers. Across the combi keyword set, average ranking position improved by around six places in four weeks, while the number of keywords sitting in positions 1–3 increased 12.5% month on month.

And some of those movements are substantial. “Which combi boiler” moved from position 15 to 2, while “which combi boiler is best” moved from 100 to 12. The Boiler Costs guide also reached page one for four high-intent cost terms representing a combined 7.2K searches per month.

We’re seeing that improvement translate further down the journey, too. In August, organic sessions increased 10.3% month on month, while organic conversions increased 20.3% to 15,733.

That’s why we’re not looking at the LLMLens numbers in isolation. The AI visibility is improving, but so is the wider search performance of the content and topic areas we’re investing in. Growing Ideal’s presence in AI hasn’t meant choosing AI over Google; we’re using the additional intelligence to make better decisions across both.

What Have We Learned?

LLMLens gives us access to a completely new layer of search intelligence. But the data alone isn’t the strategy.

It still takes people to understand what the data is telling us, decide what matters and turn those insights into meaningful action.

That’s why we see LLMLens as an intelligence layer, not an instruction manual.

As AI becomes a bigger part of how people discover, research and compare brands, simply tracking how many times you appear in ChatGPT isn’t enough.

The real value comes from understanding why you’re appearing, where the opportunities are and using that intelligence to make more effective marketing decisions across the whole search and discovery journey.