What did DoubleVerify's new shopping study find?
DoubleVerify's new study found that only 17% of consumers use AI chat to research products before they buy. Online reviews (40%) and search engines (39%) are still the top two research sources, and 52% of consumers bought something in a physical store in the past 12 months. The report came out November 19, 2026.
DoubleVerify (DV) checks where ads run and whether they work for big advertisers. The press release says the report draws on surveys of 22,000 consumers and 2,020 marketers across 25 countries, plus DV's own Rockerbox ad spend data. According to PPC Land, which read the full report, Sapio Research ran the fieldwork in February 2026, and the consumer half covers 21 countries, including the U.S. and Canada. The same figures appear in the GlobeNewswire version of the release.
Every week someone tells a business owner that ChatGPT is the new Google and their old marketing is dead. This is a 22,000-person survey saying that's not where buyers are yet. People still check reviews and search before they spend money.
Where do shoppers research products before they buy?
Shoppers research mostly on online reviews and search engines, then in physical stores, on comparison sites, with friends and family, and on social media. AI chat ranks eighth of nine sources at 17%, ahead of only influencers at 11%. Respondents could pick more than one source.
| Research source | Share of consumers |
|---|---|
| Online reviews | 40% |
| Search engines | 39% |
| Physical stores | 30% |
| Comparison sites and apps | 26% |
| Friends and family | 26% |
| Social media | 26% |
| Video reviews or tutorials | 24% |
| AI chat | 17% |
| Influencers | 11% |
Source: DoubleVerify 2026 Global Insights, as reported in the release and by PPC Land. Where people bought was similar: 52% in a physical store, 47% on an online marketplace and 25% through social media.
What does this mean for a business that runs ads?
If you run ads, your reviews are doing part of the selling. Most buyers see the ad, then search your name and read what other customers said before they call or book. A strong ad sends people to a weak review profile, and the click is wasted.
I've been running ads since 2013, and I've watched this happen in hundreds of accounts. Somebody sees a Facebook ad for a roofer, a med spa or a lawyer. They don't fill out the form right away. They open Google, type the business name, and look at the stars and the last few reviews. Three reviews from 2022 lose to a competitor with 140 recent ones, even if your ad was better.
So the review profile is part of the funnel. Ask every customer, every time, by text, right after the job. Our free Google review link generator gives you a direct link and QR code that drops people straight onto your review form, which removes most of the friction. If you want it fully hands off, here's how to get Google reviews on autopilot.
How are marketers measuring whether their ads work?
Most marketers in the survey lean on the ad platforms to grade themselves. 55% use platform-native reporting, 48% use marketing mix modeling, 40% use view-through attribution, 38% use multi-touch attribution and 35% use last-click. PPC Land noted that incrementality testing wasn't even one of the options.
DV's CEO Mark Zagorski put it this way in the release: "The greater challenge is determining whether advertising is generating additional purchases." DV sells measurement tools, so of course they'd say that. He's still right.
For a local business the fix is simple. Ads Manager counts a lead the second a form is submitted. Your bank account counts a sale when someone shows up and pays. Track every lead in a CRM from first contact to booked to showed to paid, and judge the ads on the last number. That's how we price our own ad management: a monthly base plus a share of new revenue, so the scoreboard is money in, not clicks.
Should you ignore AI search then?
No. 17% is nearly one in five buyers, and the survey was fielded in February 2026, nine months before the report came out. AI assistants also build their answers from the same reviews, listings and websites people read directly, so the same work covers both.
Keep the caveats in mind. This is a global survey of self-reported habits, it covers products more than local services, and DV has a business interest in advertisers worrying about measurement. The sample is still large: 22,000 consumers across 21 countries.
Where KenjiAI fits
KenjiAI sends review requests by text and email after every job, drafts replies to new reviews with AI, and answers every call and text with an AI agent so the person who just read your reviews gets a booking instead of voicemail. Everything lands in one CRM, so you can see which ads turned into paying customers.
What should a business owner do this week?
Spend an hour on the two research sources that lead the list, reviews and search, before spending another dollar on AI search tools. Then make sure the leads those pages send you get answered fast.
- Google your own business name on your phone and read it like a stranger would. Count reviews from the last 90 days.
- Turn on an automatic review request that goes out by text within a few hours of each finished job.
- Reply to every review from the last month, good and bad.
- Ask ChatGPT or Perplexity for "the best [your service] in [your city]" and note who it names and which review sites it cites.
- Check how fast you answer a new lead. If it's longer than five minutes, read why leads go cold and fix that next.