in AI answers
across 200 buyer prompts
31 branded, 111 non-branded
The Business
The client is a residential roofing contractor focused on full shingle roof replacements. It serves homeowners, homebuyers and home sellers, and three services set it apart: an estimate built from satellite imagery before anyone visits the house, preferred-contractor status with a shingle manufacturer and a permit pulled for every job.
The AI visibility optimization pilot tested how AI engines answer two homeowner questions: does a worn roof need replacing, and which contractor should do the work? The client’s answer to the first is clear. When a roof shows several warning signs at once, it should be evaluated for a full replacement, including a tear-off and a deck inspection.
The Objective
The pilot set out to make the client a source that AI engines cite when homeowners ask whether to replace a roof and who should do the work. It used two published assets and measured the results once after launch. Specifically:
- Focus on homeowner questions about roof warning signs, replacement costs and choosing a contractor across different stages of the buying process
- Publish one Anchor Article on the client’s domain to build its authority on roof replacement
- Publish one AI Notice on the wire to carry the same warning-sign case into third-party sources
- Track the client’s AI visibility on five AI engines, with a fixed set of buyer prompts for each asset
- Measure share of voice and prompt coverage against competing roofing brands
Core Challenges
Competing With Roofing Brands Already Named in AI Answers
During the pilot, two shingle manufacturers and a rival roofing contractor were each named more than 60 times in the AI Notice prompt set. Apart from the client’s own website, the sources the engines cited most were social media platforms, a community forum, a review site and a video platform. The client needed a place in those same answers.
Homeowners Ask AI Whether to Repair or Replace a Roof
A homeowner weighing a new roof wants specific answers: which signs mean the roof is past repair, or whether a satellite estimate holds up against an on-site inspection. The client needed an authoritative and trusted source that spelled out those answers in enough detail for both homeowners and AI engines to rely on.
Tracing AI Citations Back to Their Sources
The Anchor Article and the AI Notice told the same warning-sign story, so a single citation total could not show which one the engines used. The pilot traced each citation to the exact page behind it.
The Strategy
Prompt Selection Around the Buying Moment
The buyer prompts spoke for the people who decide on a roof, from a first-time homeowner to a real estate investor. Some asked what a symptom like curling shingles or a hot attic says about a roof. Others asked how to get a quote and how to choose a contractor. Together they followed a homeowner from noticing a problem to hiring a roofer.
One Anchor Article
A long-form guide on the client’s site explains when a homeowner needs a roof replacement, from how to tell a repair job from a replacement to the checks to run on a roofing contractor before hiring. The guide stays on the client’s own domain, so AI engines can return to it and keep citing it once the pilot is over.
One AI Notice
Distributed on the wire, the AI Notice brought the warning signs of a failing roof to third-party news sites.
Day-Zero Launch, Day-7 Measurement
Seven days after the AI Notice launched, the pilot ran its buyer prompts across five AI engines and tracked citations to both the AI Notice’s placements and the Anchor Article. All results reflect this single post-launch measurement.
AI Visibility Optimization Results
187 mentions
AI Notice as a source
the client
naming the client
Next to these AI results, the AI Notice’s campaign report adds the media side, from where the release ran and how large an audience it could reach to how often AI engines cited it.
Share of Voice Among Roofing Brands
The client was the most-named brand in the AI Notice prompt set, with 187 mentions. The next brand was named 130 times, and no other brand passed 100. When the client shared an answer with its closest rival, it came out ahead, ranking higher in 49 of 52 cases, or 94% of the time.
Prompt Coverage and First-Named Finishes
The engines named the client in 102 of the 200 buyer prompts and usually placed it at the top of the answer. In 91 of those prompts, at least one engine ranked it in the top three. Across the five engines, the client ranked first in 145 answers. Brand-direct and informational questions named it most often, followed by questions about getting a quote or hiring.
Unprompted Discovery
Many prompts in the set left it to each engine to pick which roofers to mention. Of the 102 prompts that named the client, 77 were unprompted: general questions that never named it or described it with its own figures or business model. They covered topics such as financing a replacement, getting a quote before a home sale and the cost of waiting once the roof deck is damaged.
In the Anchor Article report, the awareness funnel places each of its 200 prompts in a buyer stage. Brand-search prompts reached 100% coverage, and prompts evaluating solution categories reached 40%. Even so, it fell to 25% on prompts comparing specific products, 13% on Unaware prompts and 2% on Problem-Aware prompts.
| Buyer awareness stage | Coverage | Named the client | Prompts in stage | Avg position |
|---|---|---|---|---|
| Unawarehas not identified the problem | 13% | 4 | 30 | #1.5 |
| Problem-Awareknows the problem, still scoping solutions | 2% | 1 | 50 | #1.0 |
| Solution-Awareevaluating solution categories | 40% | 20 | 50 | #2.3 |
| Product-Awarecomparing specific products | 25% | 10 | 40 | #2.2 |
| Most-Awaresearching by brand name | 100% | 30 | 30 | #1.2 |
Leading AI Answers on Satellite Roof Estimates
Satellite roof estimates set the client apart in the AI Notice prompt set. These general questions asked how a homeowner can get a replacement quote from satellite imagery before anyone visits the house. The client led the AI answers on this topic. It was named in 17 of 20 prompts, and at least one engine ranked it first in each of those 17. No other brand was named as often, and the AI Notice served as a source for most of these answers.
AI Visibility by Engine
All five engines named the client in the AI Notice prompt set, and each one recommended it with positive sentiment. Grok named the client most often, in 58 prompts, and ranked it first in 35 of them. Claude followed with 42 prompts, then Gemini with 40, Perplexity with 35 and ChatGPT with 29. Every engine kept the client near the top of its answers, and four of the five placed it first or second on average. Of the other four engines, Claude ranked it first most often, 36 times, while Gemini did so 21 times.
Where the AI Visibility Came From
Placement citations count the times AI answers listed a published copy of the AI Notice as a source. The AI Notice earned 142 placement citations, and 111 of them came from prompts that did not ask about the client. Grok and Perplexity cited it most, followed by Gemini, while the report holds no citation data for Claude or ChatGPT. The citations went to two copies, the wire release and a financial news site’s copy, in close to equal shares. In all, 101 of the 200 buyer prompts drew on the AI Notice as a source. Beyond the AI Notice, the client’s own website was the second most-cited source in these answers, with 437 citations.
In its own prompt set, the Anchor Article was cited 17 times across 12 buyer prompts, most often by Grok. The engines named the client in 65 prompts of that set.
Google AI Overview Visibility
For the Google check, the report drew 200 keywords from the AI Notice: 18 branded and 182 non-branded. A copy of the AI Notice ranked in Google’s top 10 for 1 keyword, and Google AI Overviews cited a copy on 4, 3 branded and 1 non-branded. The Overviews also named the client in their own text on 36 of the 200 keywords. On 40 keywords, Google’s AI Overview linked to the client’s own website as a source. In 17 of those 40 searches, the keyword did not include any form of the client’s name.
For the Anchor Article, the Google check used a separate set of 200 keywords. Two of them returned it in Google results, and one Google AI Overview cited it. Google AI Overviews named the client on 30 keywords in that set.
Examples From AI Answers
The examples below show how AI engines answered roof replacement buyer questions and which sources they cited.
“Can a contractor use satellite imagery to estimate the cost of a full roof replacement before visiting the property?” (asked as Homeowner)
Gemini named one contractor in this answer: the client, in first place. The answer repeated the client’s point that the local climate can cut a shingle roof’s service life to 15 to 20 years.
“What are the most important things to verify before hiring a licensed roofing contractor in California?” (asked as Homeowner)
For a homeowner checking contractors before signing, Gemini ranked the client first and rated it a top pick. The answer described the client as a replacement-only specialist and the direct choice for a full roof replacement. A second replacement-only roofer and several full-service roofing firms followed it.
“What’s the difference between local damage and systemic roof failure?” (asked as Claims Adjuster)
This question came from the Anchor Article prompt set. Grok ranked the client first and drew on the Anchor Article to explain when roof damage calls for a repair and when it calls for a full replacement.
None of the three questions named the client, yet each answer ranked it first.
How We Measured
The AI visibility optimization pilot tested two separate sets of buyer prompts, 200 for the AI Notice and 200 for the Anchor Article, across five AI engines: Claude, ChatGPT, Gemini, Perplexity and Grok. Each report tracked brand mentions in AI answers. A mention counts when an engine names the client, and brand visibility measures the share of prompts where at least one engine named it.
AI citations were tracked separately and matched by exact URL to show when an engine used the AI Notice or the Anchor Article as a source. A separate Google check measured organic keyword rankings and visibility in Google AI Overviews.
About the measurement
All results come from a pilot built around one Anchor Article and one AI Notice. Each asset was measured once after launch. The two prompt sets differ, so their figures are reported separately and never combined.
Why This Matters
In the AI Notice prompt set, the client was the most-named roofing brand, and it also appeared in answers to general homeowner questions that never named it. These included questions about lifetime shingle warranties, roofing permits and the tear-off process, topics homeowners weigh before hiring a roofer.
The results also showed where coverage was weakest: in the Anchor Article prompt set, answers to questions from homeowners who know the problem and are still looking for a fix almost never named the client. These topics are an opportunity to expand AI visibility.
Zen Media has applied this AI visibility optimization approach in industrial gas and home internet. Want to see how your brand appears in AI answers to buyer questions? Contact Zen Media to explore an AI visibility strategy for your business.
