Optimove Says 49% of Gift Shoppers Reach for AI Before a Product Page
Optimove Insights surveyed 648 US gift buyers in April 2026 and found 49% now regularly use ChatGPT, Claude, or Gemini for gift recommendations, with another 23% using them occasionally. Mother's Day falls on May 10, which gives retail brands roughly 17 days to make their product copy citable by a model. The underlying trend is that 72% of gift shoppers have at least touched an AI tool inside their purchase journey, so the fight for attention moved one step upstream from search.
The PPC Land writeup of the Optimove report is the cleanest summary I have seen. The sample is 648 US consumers aged 18 and up with household incomes of $75,000 or more, which matters for the conclusion because that is the segment most likely to actually convert on jewelry, flowers, and experiences, the three top Mother's Day categories in the study.
The 49% is the headline. 72% is the actual problem.
Forty-nine percent is the quotable stat. The 72% combined (regular plus occasional) is what should actually worry a retail marketer. If close to three-quarters of your target has used an AI tool at some point during a gift search, you cannot treat this as a fringe channel anymore. The shopper either asks ChatGPT first and lands on the three brands it mentioned, or they don't. There is no middle option where you get "considered" because your organic rank is solid.
The wealthier skew of the Optimove sample is worth sitting with for a second. $75K+ households buy higher-ticket gifts and are also the segment most fluent with new interfaces. So the behavior is most concentrated in the buyers you care about most, which is a different story than "47% of the broad population uses AI for shopping sometimes." It is closer to "the buyers that matter are already there."
17 days out is not a coincidence, it is the new shopping window
Optimove also reports that 26% of respondents buy at least a month before Mother's Day, 35% buy about two weeks ahead, and 24% buy about a week out. Stacked against today's date, the bulk of Mother's Day spend is going to get committed inside the next ten days. That lines up with Adobe's finding that AI-driven traffic to US retail sites was up 393% year over year in Q1 2026 and converting 42% better than non-AI traffic in March, per the TechCrunch breakdown of the Adobe numbers.
I think the most overlooked piece here is the conversion flip. One year ago, Adobe reported AI referral traffic converting 38% worse than standard channels. Between March 2025 and March 2026 that number didn't just close, it crossed. AI referrals now spend 48% more time on product pages and deliver 37% more revenue per visit than non-AI channels, on Adobe's numbers. Something real changed about who is arriving on your PDP via an AI tool, and my read is that it is the higher-intent, higher-ticket buyer arriving with a pre-vetted shortlist.
And to be fair, none of this is entirely new. LLM-to-retail funnels have been climbing since last holiday, when Adobe logged a 693% year-over-year AI traffic jump during the 2025 season. What is new is the behavior is now seeping out of Cyber Week and into a mid-spring gift holiday. Mother's Day was an early leading indicator in 2024 too, where Optimove's 2024 survey found 70% of respondents purchased at least two weeks before the holiday. The early-purchase pattern is repeating with a different discovery surface sitting on top.
The uncomfortable fix: your product copy, not your ad creative
Most of the conversations I see about AI shopping jump to agents, checkout, Operator, and the broader agent-commerce storyline. That is the right long view and the wrong two-week fix. For Mother's Day 2026, the thing an LLM actually looks at when someone prompts it with "what should I get my mom who likes gardening" is your product description, your category page copy, and the review density on your product detail page. If those three surfaces do not read as a specific answer to a specific gift-giving scenario, you are not in the response.
From what I have seen, the product copy most retailers have is written for a human scroller, not a model. Benefit-forward, aesthetic, vague on the use case. A model needs the use case spelled out. "Lightweight gardening gloves with a reinforced palm for a mom who spends her weekends in the yard" is different from "Premium gardening gloves, engineered for performance." The first wins a prompt. The second loses it.
Adobe's own analysis of generative AI shopping behavior found that brands cited in an AI Overview see roughly a 35% CTR lift from the citation. The LLM answer block is the new rich result. And the block is pulling from PDPs that have cleanly written, scenario-specific copy, not from whichever brand spent the most on Demand Gen last quarter.
The 72-hour audit that actually matters
Here is the order I would personally run it in, before the weekend:
- Query ChatGPT, Claude, and Gemini with 10 buyer-intent prompts for Mother's Day. "Gift for a mom who loves cooking." "Gift for a mom who works from home." "Gift for a mom under $75." "Last-minute Mother's Day gift for a mom who has everything." Log which brands come up. If you are not in at least five of the ten answers for your category, you have a visibility problem you can measure now, not after May 10.
- Pull your top 20 Mother's Day SKUs and test the copy against one question. Can a model answer "who is this for" from the description alone, without the product image? If not, rewrite the first two sentences of the description to name the recipient scenario, not the feature set.
- Check that your PDPs expose structured data cleanly. Adobe's own team has publicly flagged that a lot of retail sites are not machine-readable for AI tools right now, which is a solvable problem if someone on the engineering side gets thirty minutes.
- Look at your reviews. Product review text is disproportionately cited in AI shopping responses because it contains the kind of scenario language marketing copy usually skips. If your best-sellers have low review counts because you rotated off your old review widget last year, this is the week to import the backlog.
- Don't touch your paid campaigns yet. Adobe's data says AI referrals are converting 42% better than paid channels in March 2026, so reallocating budget to a new Demand Gen push to "compete with AI" is probably the wrong instinct. Fix the discovery surface first. Chase the budget line second.
The 15-minute version for smaller brands is simpler: rewrite the first two sentences of your five top Mother's Day SKUs to name a specific recipient scenario. That alone often flips whether a model includes your product in a response. It is also, honestly, the kind of thing that should have been done eight months ago and probably wasn't.
The sequencing nobody is pricing in yet
One detail in the Optimove report has sat with me all morning. 80% of respondents said they would buy earlier if prompted by a sale. That is normal retail demand elasticity. But put it next to the AI discovery number and the sequencing starts to matter. The old model was: Google search, PLA, discount nudge, conversion. The new model looks more like: ChatGPT prompt, shortlist of two or three brands, brand search on Google, discount nudge, conversion. The discount is still doing the work of closing. It just is not starting the journey anymore, which means earlier promo emails will not save you if the model cut you from the shortlist three steps back.
Pivoting to be "visible in LLMs" is less of a channel add and more like a second-layer SEO that sits on top of the one your team already runs. On paper that sounds like a burden. In practice it is probably the most ROI-positive content work a retail brand can do in Q2, because the surface is still uncrowded and the conversion rate data says the traffic is high intent. We covered a related thread when Google's Liz Reid conceded AI Mode queries run 2-3x longer than standard keyword tools track. The AI shopping funnel and the AI search funnel are behaving more and more like the same workflow, and the practitioners who internalize that first are going to have an unusually easy Q2.
If you can only do one thing this week
Rewrite the first two sentences of your top five Mother's Day SKUs to name a specific gift scenario, out loud, for a human reader. Then paste each one into a model and ask it to recommend gifts for that scenario. If your product comes up, your copy works for both the shopper and the tool reading on their behalf. If it doesn't, you have a starting point that took a coffee's worth of time. The seventeen days are real, but so is the fix, and the retailers who do this work quietly in the last two weeks of April are going to be the ones whose May revenue makes the rest of the team nervous.
Notice Me Senpai Editorial
