68% of Local Searches Are AI Overviews. Your Local Pack Isn’t Enough Anymore.

A Client Calls Me on a Tuesday Morning. He’s Lost 43% of His Local Clicks.

A client calls me on a Tuesday morning. He runs a clothing retailer with 12 physical stores. 11,000 organic sessions per month a year ago. Today, 6,200. His pages — « boutique Paris 11 », « boutique Lyon Part-Dieu » — still rank first or second on Google Maps. His local pack is green. Traffic is collapsing. I hear this: « Stéphane, I don’t get it. I’m ranked well everywhere. »

The first reflex is to check the Google Business Profile listings. They’re clean. Recent reviews, current hours, photos. No manual penalties. So I open 15 high-intent queries, the ones that used to drive the most traffic: « evening dress Paris 11 », « men’s suit Lyon », « alterations in-store near Lille ». Result: on 10 searches, 6 display a massive AI Overview at the top on mobile, answering directly. No map pack. No listing. No website. Just Google’s AI aggregating content. My client doesn’t appear anywhere in these answers. He’s invisible to 6,000 monthly searches. Ranked well, but invisible. That’s the new paradox of local SEO.

29-Point Gap. The Local Pack Protects Nothing Anymore.

A Whitespark study, covered by Search Engine Journal (source), shows AI Overviews appear in 68% of local searches. The local pack shows up in just 39% of the same queries. A 29-point hole. For informational queries like « how long does an in-store fitting take », the AI generates an answer 92% of the time. For hybrid intent — like « average price of a custom suit in Lille » — that jumps to 97%. These aren’t edge-case queries. These are questions real buyers ask every day, mid-decision.

Yet businesses that stop at the local pack miss 6 to 7 buyers out of 10. I see with my e-commerce clients who also have physical stores that these AI pages generate up to 3 times more impressions than the map pack for mid-funnel queries. The user doesn’t click anymore. They read. They get their answer without ever seeing your listing, let alone your site. Google controls the entire expérience end-to-end. If your content isn’t structured to be the source of that answer, you don’t exist.

Your Google Business Profile Is Clean. The AI Ignores It Anyway.

At first, I thought the problem was the listings. Reviews, catégories, hours. All wrong. The website needs to serve as a trusted source for the synthesis algorithm. AI Overviews don’t read hours. They aggregate passages from pages, FAQs, reviews, blog articles, structured data. If your « boutique Lyon » page contains only a Google Map and a phone number, it’s silent to the AI.

With the same client, I analyzed 12 « store » pages. On 12 pages, 11 had zero structured LocalBusiness markup. No FAQ schema. No answer to the question « what’s the turnaround time for alterations in-store? » — yet that query consistently triggers an AI Overview. The website had a blog with 47 articles, but none referenced a physical store. Worse: title tags were optimized for Google Maps (« Buy dress Paris 11 »), not for real intent. The AI needs informative content, not just transactional. That disconnect renders the site invisible.

Voici comment adapter le framework DOSE (Define, Observe, Structure, Evaluate) à vos pages locales, comme je l’ai fait pour ce client retail.

I Applied DOSE to Local Pages. Here’s What Changes.

Guillaume Attias at BMO Academy teaches the DOSE framework: Define, Observe, Structure, Evaluate. I adapted it for local SEO in the AI Overviews era.

1. Define: map all hybrid and informational intents around each location. Not keywords, but the answer objects Google synthesizes. Examples: « average alteration price », « how many fittings possible », « fabrics available in-store ».
2. Observe: for each query, watch which sources Google cites in the AI Overview. Often product pages, competitor FAQs, forums. Spot the patterns.
3. Structure: rebuild each localized page with enriched LocalBusiness schema (openingHours, priceRange, department, makesOffer), FAQ tags via Question/Answer type, and a synthetic content block of 80–120 words per intent, paraphrasable by the AI.
4. Evaluate: measure appearance rate in AI Overviews via geolocated queries, click-through rate on synthesis links, and call volume from these pages.

The classic mistake: stuff the page with 800 words to « feed the algo ». No. The AI favors concise answers extracted from an authoritative page. The schema is the trigger, the content is the answer.

12 Stores, 90 Days. +320% Impressions, +47 Calls Per Month.

We rebuilt 12 store pages, one by one. Three actions:

• LocalBusiness markup with 11 properties filled in (including geo, openingHoursSpecification, areaServed, hasMap).
• FAQ of 4 questions on each page (turnaround, average prices, services available, parking).
• 18 local blog articles linked to each store (« How to Choose a Suit When You Live in Lyon Part-Dieu »), with anchored links to the store page.

With another client, a 9-store sports chain, we also tagged inventory with Product schema including availability and pickupLocation. Result in 90 days: +320% impressions in AI Overviews (measured manually across 25 key queries), +180% clicks to the site via AI links, and +47 qualified calls per month. No added ad spend. Just a redistribution of existing content. The Montpellier store went from invisible to cited in 82% of hybrid queries tested. It doesn’t dominate top rank — it’s the source.

The number that matters: before, 4% of calls came from AI. After, 34%.

Automating Listings Won’t Break Your Silence.

I see offers: « We broadcast you to 50 directories automatically. » Tempting. But AI Overviews don’t pull from directories. They hunt original content, answers — not copy-pasted NAP (Name, Address, Phone). Last time I tested an auto-sync solution on 3 stores, AI appearance didn’t budge a pixel. Zero improvement. The problem isn’t citation consistency. It’s lack of source material.

A « store » page must tell a story. What you sell in-store that isn’t online. Services you offer in-person. Average alteration time. Questions your sales staff hears 20 times a day. You can’t outsource this. You collect it in-store, with a voice recorder and a notebook. I do this with clients: one afternoon observing behind the counter yields more indexable content than 6 months of standard SEO production. That’s the content the AI cites.

Check Your Invisibility Yourself in 12 Minutes.

Test 5 hybrid queries per location: « how much does [your product] cost in [city] », « where can I try [product] near [neighborhood] », « what’s the turnaround for [service] in [city] ». Run them from mobile in private browsing, with geolocation on. Count how many times your site appears as a source in the AI Overview box. If it’s zero, you know what to do.

Next, verify your store page markup: Google’s rich results testing tool should confirm a valid LocalBusiness type, with priceRange, areaServed, and openingHours. Without valid schema, even world-class content is orphaned to the AI. Finally, search « [your brand] [city] reviews » and watch if review snippets are pulled. Google cites them frequently. The more your reviews contain useful vocabulary (price, service, selection), the more the AI feeds on them. Ask customers to mention the specific service that convinced them, not just « great store ». A long-tail keyword in a review beats a locked-down listing. This diagnosis takes 12 minutes per store. Time enough to realize you’re losing 6 customers out of 10 without knowing it. Is it worth waiting for the local pack to come back?

Stéphane Jambu

SEO & AI Engineer

I build growth systems / AI / Neuroscience | 650+ clients · 80 LinkedIn testimonials · 30 years of expertise · 15 years of systems running without me.

Follow on LinkedIn