Local businesses appear to be adopting generative AI for responding to online reviews faster than customers are adopting it to write them. A new study by AI software company Revfer, based on approximately 88,000 Google reviews from 400 businesses in Greater Boston and MetroWest, found a growing presence of writing patterns associated with AI in owner responses, while customer reviews showed comparatively limited changes. The research also identified a rise in five-star ratings and more positive language, raising questions about how AI-assisted communication could affect trust in online reviews.
Generative AI writing tools have become increasingly accessible since late 2022, giving businesses new ways to draft customer communications, marketing content and responses to online feedback. But adoption may not be happening at the same pace on both sides of the customer relationship.
Research from Revfer suggests that local businesses have changed how they respond to Google reviews more noticeably than customers have changed how they write them. The company’s analysis examined public review histories from 400 businesses across Greater Boston and MetroWest, covering approximately 88,000 reviews published between 2019 and 2026.
The sample included 100 businesses in each of four categories: hair salons, restaurants, dental practices and auto repair shops. Revfer analyzed patterns in customer reviews and business-owner replies to identify changes that emerged around the wider availability of generative AI tools.
The findings suggest that AI-assisted writing has become more visible in business responses, while customer review language has remained comparatively stable.
AI Writing Patterns Appear More Frequently in Business Replies
One of the study’s indicators was the use of em dashes, a punctuation mark frequently associated with AI-generated writing. In customer reviews, the punctuation appeared in 0.2% of cases in 2019, increasing to 2.9% in 2026.
The increase was more pronounced in owner replies, where em dash usage rose from 0.1% to 4.2% over the same period.
Revfer also examined words and phrases commonly associated with chatbot-generated text, including “delve” and “testament to.” These expressions remained almost absent from customer reviews, according to the research.
The results point to a difference in how businesses and consumers are using generative AI. Businesses have an operational reason to adopt writing assistance: responding to reviews takes time, particularly for organizations managing multiple locations or receiving frequent customer feedback.
AI tools can help draft responses, maintain a consistent tone and reduce the effort involved in acknowledging customer comments. For small businesses with limited administrative resources, that convenience can make automated or AI-assisted writing attractive.
Customers, by contrast, generally write reviews to describe individual experiences. The study suggests that most continue to use relatively familiar writing patterns rather than adopting the more recognizable language associated with chatbots.
Revfer’s earlier research found that owner replies had roughly doubled since 2019 and that responses published from 2023 onward increasingly displayed patterns consistent with AI writing tools.
Customer Reviews Are Becoming More Positive
The study identified another significant development: customer ratings have become more favorable.
The share of five-star reviews in the sample increased from 69% in 2019 to 86% in 2026. The average rating rose from 4.40 to 4.60.
The change was visible across several business categories. Hair salons increased from an average rating of 4.58 to 4.79, while restaurants moved from 4.29 to 4.48.
Review language also became more positive. Words such as “amazing,” “incredibly” and “genuinely” appeared more frequently, while “good” and “nice” declined.
The proportion of reviews containing a star rating but no written text also fell, from 47% to 24%.
These changes could influence how consumers interpret online ratings. As five-star reviews become more prevalent, written descriptions may carry greater weight in helping prospective customers distinguish between businesses. Negative reviews could also attract more attention when the overall rating distribution is heavily concentrated at the top.
However, the research does not establish why customer ratings increased. Changes in customer expectations, review solicitation practices, business quality, platform behavior or the composition of reviewers could contribute to the trend. The findings should not be interpreted as evidence that generative AI caused the rise in ratings.
Frequent Reviewers Show a Different Writing Pattern
Although the typical customer review has remained approximately 140 characters long since 2021, Revfer identified a difference among people who contribute reviews frequently.
Customers who had written at least 100 Google reviews were producing reviews approximately 60% longer than before ChatGPT’s release, according to the analysis. People who had written only one review, meanwhile, were writing shorter ones.
This divergence suggests that review-writing behavior is not uniform across the platform. Frequent contributors may have different habits, levels of familiarity with online reviews or motivations for sharing detailed feedback.
The study does not establish whether the longer reviews from frequent contributors were generated or assisted by AI. Length alone is not a reliable indicator of AI authorship.
AI-Assisted Responses Create a New Trust Challenge
Revfer’s findings highlight a growing challenge for businesses using generative AI in customer communication: efficiency does not necessarily translate into authenticity.
An AI-generated reply can help a business acknowledge feedback quickly, but generic language may make the response feel disconnected from the customer’s actual experience. This is particularly important when responding to detailed complaints or emotionally charged reviews.
The company’s founder, Pankaj Aggarwal, argues that templated replies can be noticeable beneath personal customer feedback. As highly positive reviews become more common, businesses may also face greater scrutiny when responding to negative experiences.
There is an important limitation to the research. Google reviews do not carry a label identifying whether a customer or business used AI to write a particular contribution. Revfer therefore treats punctuation and language patterns as indicators consistent with AI writing tools, not proof of AI authorship. Smartphone keyboards and prompts built into Google’s services could also influence writing behavior.
The analysis is based on a regional sample of 400 businesses, not a representative study of all Google reviews or businesses globally. Its findings offer a useful snapshot of changing communication patterns, but broader research would be needed to determine whether the same trends hold across other regions and industries.
For businesses, the practical implication is that AI can help manage the volume of review responses, but the quality of those interactions still depends on relevance and human oversight. As generative AI becomes more embedded in customer-facing workflows, maintaining a recognizable business voice may become as important as responding quickly.
Market Landscape
Generative AI is increasingly being incorporated into customer experience and reputation-management workflows. Businesses can use AI writing assistants to draft review responses, summarize feedback and support customer communication at scale.
The Revfer study highlights an emerging distinction between business adoption and consumer behavior. While owner replies show a greater increase in language patterns associated with AI tools, customer reviews remain comparatively stable in length and vocabulary.
For the broader AI market, the findings underline a challenge for customer-facing automation: businesses must balance response speed and consistency with authenticity. AI-generated communication that appears generic or disconnected from individual feedback could undermine the trust these tools are intended to support.
Top Insights
- Business adoption is more visible: Revfer found a larger increase in AI-associated writing patterns in owner replies than in customer reviews.
- Em dash usage increased: Its presence rose to 4.2% in owner responses and 2.9% in customer reviews by 2026.
- Ratings became more positive: Five-star reviews increased from 69% in 2019 to 86% in 2026.
- Frequent reviewers differ: People with at least 100 reviews now write reviews approximately 60% longer than before ChatGPT’s release.
- AI authorship remains unconfirmed: The study identifies linguistic indicators, not verified AI-generated content.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
