AI Shopping Assistants Driving Incremental Sales Growth

A new report by Bart Turczynski, Head of Growth at Tidio, titled "AI Shopping Assistants for E-Commerce," examines the impact of AI shopping assistants on e-commerce. The report, which draws on 39 published sources, including retailer earnings disclosures and market data, reveals that these assistants are increasingly becoming a significant part of the retail landscape.

Amazon, one of the largest players in the e-commerce market, reported that more than 300 million customers used its shopping assistant in 2025. The assistant helped deliver nearly $12 billion in incremental annualized sales, according to the company. However, it's important to note that this figure is attributed to incremental sales, while the company separately reported that shoppers using the assistant were 60% more likely to complete a purchase. This distinction highlights the difference between attributed sales and actual incremental value.

Etsy's A/B test provides a clear example of the challenges in attributing sales to AI shopping assistants. A recommendation change increased recommendation clicks by 28.3%, but conversion rose by only 0.22% and GMV fell by 0.25%. At the same time, clicks on organic search results fell significantly by 1.4%. Etsy's causal mediation analysis found that the recommendation change itself increased conversion by 0.50%, while the resulting decline in search activity reduced conversion by 0.28%. In other words, roughly half of the genuine conversion gain was consumed by displaced search activity.

The report argues that AI shopping assistants should be evaluated based on their incremental commercial contribution, not just engagement or attributed sales. Walmart's higher average order value among Sparky users and Magalu's report of conversion three times higher in its WhatsApp shopping experience than in its app are both commercially meaningful but are observational comparisons, not randomized estimates of causal lift. Salesforce's $262 billion is an attribution estimate of sales influenced by AI and agents, drawn from data on more than 1.5 billion shoppers.

Tytsus Gołas, Founder and CEO of Tidio, wrote in the report's foreword that retailers like Amazon, Walmart, Home Depot, and Magalu are deploying these systems at scale and already reporting results that reach the P&L. However, most retailers do not have the engineering budgets to match. For smaller retailers, the first measurable value may appear elsewhere. MattressNextDay, a Tidio customer, reports that its assistant resolves 73% of customer conversations without a human, returning more than 400 hours per month to its support team. UK flooring retailer Renovate Direct, also a Tidio customer, reports that its resolution rate increased to 69% from 42% after the assistant began routing tickets by intent.

The report also suggests that proactive AI assistance makes incrementality easier to test. Because a retailer controls whether a proactive prompt is shown, it can expose one comparable group of shoppers to the intervention while withholding it from another. This creates a practical holdout test without requiring the retailer to replace search, navigation, or its existing recommendation systems.

Retailers should evaluate such pilots using eligible sessions, exposure rate, incremental order-rate change, and average order value, before accounting for margin, returns, and operating costs. Once a retailer has a real holdout result, the report identifies incremental contribution per eligible session as one of the most useful summary metrics. The correct forecast comes from the retailer's own experiment, the report concludes.

In summary, the report argues that the question for retailers is no longer whether AI shopping assistants can matter, but where they create measurable incremental value. The full report is available at www.getlyro.ai/reports/ai-shopping-assistants.