LEOPRD: AI creates new reputation blind spot for PR & communications teams

LEOPRD has released its latest study, Reputation to Revenue 2026, which found that most of the information AI uses to evaluate brands sits outside their own websites, creating a wider reputation environment for communications teams to monitor.

The report analysed 31,200 AI responses from a wider dataset of more than 214,000 monitored-brand responses. This spanned 27 brands, seven categories, eight AI platforms, and two markets (Australia and Great Britain) during March 2026.

The study found 86 per cent of citations came from outside the monitored brands' own properties, including commercial and comparison sites, editorial coverage, reviews, and community sources.

Celia Harding, founder of LEOPRD, said AI had created a new reputation blind spot for communications teams.

"AI does not care which department owns an issue. A customer-service problem can become a reputation problem the moment somebody asks AI whether a company is worth choosing. The same is true of pricing confusion, product limitations, delivery complaints, regulatory concerns, or competitor comparisons."

Being visible does not mean being recommended

The study found that when users asked AI for recommendations without naming a brand, the monitored brand was absent from 36 per cent of answers. It was mentioned but not selected as the primary recommendation in 35 per cent, while it was the primary recommendation in 29 per cent.

The report also found that greater citation volume did not correspond with stronger recommendations. Answers where a monitored brand appeared but another option was selected contained an average of 9.2 citations, compared with 8.4 when the monitored brand was the primary recommendation.

According to LEOPRD, what sources said about a brand and whether they provided a clear reason to recommend it mattered more than source volume alone.

Drawing on a wider reputation ecosystem

The research found AI uses different types of public information depending on what it is evaluating.

It found that commercial and comparison sources help define options and trade-offs, editorial coverage provides independent context and validation, reviews and communities provide evidence of customer experience, while owned sources help answer questions about areas such as pricing, availability, features and suitability.

Owned information became more prominent when brands were selected. Monitored-brand domains accounted for 0.5 per cent of citations when the brand was not named in the answer, rising to 4 per cent when it was mentioned, and 7.9 per cent when it was the primary recommendation.

The evidence mix also differed by category. Commercial sources accounted for 63.9 per cent of primary-recommendation citations in B2B software, while editorial sources accounted for 41 per cent in travel.

Reviews are key when users ask about risk

The source mix changed when prompts shifted from choosing a product or service to asking what could go wrong.

In shopping prompts, reviews represented 11 per cent of cited sources and monitored-brand-owned sources represented 2 per cent . For red-flag prompts, reviews rose to 33 per cent and owned sources to 9 per cent.

The report identified safety and regulation, customer service, and innovation / technology among the strongest recurring risk associations with answers where brands were not recommended. It cautions that these are associations rather than evidence that the factors caused the recommendation outcome.

"When people ask AI what to buy, it leans heavily on commercial and comparison content. When they ask what could go wrong, it turns much more heavily to reviews, peer experience, and the company’s own public explanation," Celia said.

Competitor content can shape a brand's AI reputation

The report shared a case study of SafetyCulture (now Mitti), where 55 per cent of citations came from commercial content, including comparison pages and competitor sites, compared with 10 per cent from SafetyCulture's own domains.

Across six frequently surfaced competitor sources, SafetyCulture was repeatedly associated with key words and phrases such as being easy to use, mobile-first, suited to frontline teams, strong for inspections and templates, and require minimal training.

The report says repeated descriptions across comparison pages and reviews can reinforce how AI understands a brand, even when the content originates from competitors trying to position their own products as alternatives.

AI platforms use different evidence

The study tested eight AI models, although only six exposed citation URLs and could therefore be included in its source-mix comparison.

Among those platforms, ChatGPT had the highest share of editorial citations at 25.3 per cent, while Copilot had the highest share of commercial citations at 35.8 per cent. Perplexity had the highest proportion of reviews at 26.6 per cent, while owned sources accounted for 19.1 per cent of Gemini's citations.

LEOPRD says the differences mean there is no single source strategy that guarantees the same performance across AI platforms.

One AI answer is not a reputation measure

Only 40 per cent of repeated test groups produced the same recommendation outcome throughout. The report found that sources could change even when the recommendation remained the same, while recommendations could also change despite similar sources.

“Typing your company into ChatGPT and screenshotting the answer is not reputation monitoring,” Celia said.

"One answer is an observation. The real task is to understand which issues, sources, and narratives recur across platforms and over time - and whether they affect leads, sales, complaints, churn or customer-service demand."

What communications professionals need to know

The research suggests AI reputation extends beyond what a brand publishes on its own channels. Reviews, earned media, communities, and competitor comparison pages can all contribute to the public information AI uses to understand and evaluate an organisation.

Communications professionals should:

• Monitor the wider information environment around the brand, including reviews, comparison content, and recurring third-party narratives.

• Make important claims, product information, and limitations clear on owned channels, while building credible independent evidence that can support them.

• Test AI responses repeatedly and across platforms rather than treating a single answer or visibility score as representative of the brand's AI reputation.