Target, Lowe’s and Walmart have reported stronger shopping behavior around their AI assistants, including a Lowe’s claim that users converted at triple the rate of nonusers. The disclosures don’t establish how much additional revenue the assistants generated.
For small consumer-facing retailers, the figures make conversational product discovery a concrete testing opportunity, not a sales forecast. The companies measured different things, and shoppers who choose an assistant may already be more inclined to buy.
What the retailers reported
On the November 18, 2026, earnings call, Lowe’s CEO Marvin Ellison reported that online customers using Mylow converted at triple the rate of nonusers, according to the Lowe’s Q3 2026 earnings-call transcript. Ellison also reported that Mylow answered more than one million questions monthly. The disclosure provided neither a user count nor an estimate of incremental revenue.
Target CEO Michael Fiddelke reported on its November 18, 2026, call that early Ask Target beta testing showed guests moving to purchase at about twice the rate through the technology as elsewhere, per the Target Q3 2026 earnings-call transcript. Fiddelke also said that shopping volumes through external AI platforms remained low.
On November 19, 2026, Walmart reported U.S. Sparky usage growth at 129% year over year in its Walmart Q3 FY2027 earnings-call transcript and Walmart Q3 FY27 earnings release. Walmart CEO John Furner reported that baskets associated with Sparky users were expanding by 40% as customers found more items.
A November 20 closer look at Walmart’s Sparky shopping figures examines that disclosure.
Different measures, unresolved causation
The comparison warrants three distinctions:
- These disclosures do not establish a common AI shopping assistant conversion benchmark.
- Target described purchase behavior in beta testing; Lowe’s compared users with nonusers; Walmart described usage growth and basket expansion.
- None of those comparisons, as presented, establishes that the assistant caused additional purchases.
Those are limits of the disclosures, not evidence that the assistants don’t work. They leave the incremental-sales question unanswered.
There’s also a practical prerequisite beyond choosing a model. Gartner senior director analyst Sandeep Unni, in CIO Dive’s reporting, said that shopping agents depend on product content, inventory visibility, pricing accuracy and fulfillment promises, not simply the model.
For a small-team owner, that suggests starting with accurate, useful product information that customers and assistants can find. Any assistant test should compare comparable shopping sessions, track purchases and revenue, and use randomly assigned groups where feasible. The meaningful result isn’t how often shoppers open an assistant, but whether it produces purchases that wouldn’t otherwise have happened.