Google Lens: your customers are searching with a photo. Since 24 September, you can finally see it.

Since 24 September 2026, Google Search Console counts the visits that reach your site from a photo: Google Lens, Circle to Search, an uploaded image, a right-click in Chrome. It is the first time a local business can measure a habit that is already huge — nearly 20 billion visual searches a month — and that follows a rule almost no website respects: a photo types no keyword. What gets you found is the image itself.

What Google added on 24 September

The announcement went almost unnoticed in a busy month. It is signed by two product leads, one from Google Lens and one from Search Console, and it fits in one line: performance reports now have a 'multimodal' option in the 'Search type' menu, alongside Web, Image, Video and News.

That filter groups four entry points with one thing in common: the search starts from an image, not a sentence.

Two details matter. The filter exists in the Search results report and in the Generative AI features report, the one I took apart in the article on measuring AI visibility: a photo can therefore also trigger an AI Overview or AI Mode answer that cites you. And the rollout is worldwide from day one.

Until now, a visit from Lens was blended into the rest of your traffic. Nobody knew it existed, so nobody worked on it.

A habit that is already huge, and that nobody measured

The latest official figure dates from November 2024: Lens was then handling nearly 20 billion visual searches a month, 20% of them shopping-related. Two things have sped the trend up since.

First, the camera joined the conversation. Since 26 March 2026, Search Live lets people film what is in front of them and ask their question out loud, in every country and language where AI Mode is available. In France that has been the case since 22 July, when Google switched its AI features on there. Google says Search can 'see what your camera sees' and offer links to learn more: those links are web pages. Possibly yours.

Second, AI Mode itself changed scale: more than one billion monthly users, Sundar Pichai announced on 22 July.

For a local business, a photo search looks like three scenes you have already seen from the other side of the counter:

• The dish seen elsewhere. A photo in a story, the plate on the next table, a cake on a friend's feed. The customer doesn't know its name, so they photograph it or circle it: 'what is this, and where can I eat it nearby?'
• The object spotted at someone's place. A lamp, a pair of shoes, a bouquet, a vase. They can't describe it in words — which is exactly why they take a photo.
• The shopfront passed on the street. It's 9 pm, you're closed, they like the window. They photograph it to remember, then search from the photo to find your hours or your site.

None of these three people will type your business name. They don't know it yet.

A photo types no keyword

This is where visual search departs from everything you know about SEO. Google says it plainly in its image documentation, updated in March 2026: it uses alt text 'along with computer vision algorithms and the contents of the page' to understand what an image is about.

In a photo search the order flips. The image makes the match first, and the text confirms it. Three direct consequences for a business website:

1. If your real photos are not on your site, you cannot be the result. Lens will send your future customer to whoever published a similar image: a delivery platform, a marketplace, a Pinterest board, a competitor.
2. A stock photo points to nobody. The perfect croissant bought on a photo library is published on thousands of pages. It cannot lead to your bakery rather than another one. Only a photo that you alone own can identify you.
3. A photo Google cannot see does not exist. Images set as CSS banner backgrounds, carousels loaded later by JavaScript, menus as PDFs: Google explicitly recommends the standard image element so its crawlers can find and process your visuals.

The last row echoes a principle I repeat from article to article because it holds for all these systems: say what you are, where, and at what price, where the machine is looking. For ChatGPT, that is the first two hundred characters under your title. For a photo search, it is right next to the image.

What it does not mean

I would rather be clear before someone sells you a 'Google Lens optimisation' package at a premium.

1. There is no separate Lens SEO. Everything that matters is in Google's public image documentation, and none of it is exotic. What is new is the measurement, not the method.
2. Your numbers will be small, at least at first. The filter only appears if your site already gets that traffic. For a neighbourhood shop, it may read zero for weeks. A zero is a baseline, not a verdict.
3. You will not see your customers' photos. Search Console tells you a visual search led to a given page; it obviously does not show the image the user took.
4. For the shopfront, your Google Business Profile does most of the work. When someone photographs a place, what lets Google recognise it is first and foremost your listing's photos and information in Maps. I explained how Gemini now reads those photos in the article on Ask Maps. For the dish and the product, though, your website is what makes the difference.

That last point is the real lesson. Your Google listing makes you findable when people look for you. Your site, with your real photos properly described, makes you findable when people look for what you sell without knowing you. It is exactly what the audit of 4,776 venues I covered this summer measured: having a website of your own is the first factor for getting into AI recommendations, ahead of your star rating.

The checklist, in order

1. Run the photo test on your three best-selling products or dishes, and on your shopfront. Note who comes up: that's your starting point.
2. Replace stock photos with your own, starting with the pages that matter. A recent phone, daylight near a window and a neutral background are enough.
3. One photo, one piece of information. Next to every important image: what it is, the price, the neighbourhood. A restaurant menu as text with photos of the dishes, not a photo of the menu.
4. Describe every image. Alt text that says what you see — 'Lemon meringue tart, pâtisserie on rue de Charonne, Paris 11th' rather than 'dessert' — and a meaningful file name: lemon-meringue-tart.jpg rather than IMG_4032.jpg.
5. Move important photos out of CSS backgrounds and JavaScript carousels. A standard image element, on a page about the same subject.
6. Add the image to your structured data. The image field of your business or product markup tells Google, unambiguously, which photo matches what.
7. Update your Google Business Profile photos: shopfront by day and by night, interior, products. That's what a passer-by photographs. If you are starting from scratch, your listing and your site work as a pair.
8. Keep images light. Modern formats (AVIF, WebP), several sizes for different screens. A real photo that takes four seconds to load on mobile loses the customer Lens just sent you; I put numbers on what a slow site costs a local business.
9. Record your baseline this week. Search Console, Performance, 'Search type', multimodal. Note the figure, even if it is zero, and compare in three months.

Frequently asked questions

What is the 'multimodal' filter in Search Console?

It is a new value of the 'Search type' filter in the performance reports, added by Google on 24 September 2026 and rolled out everywhere. It isolates clicks and impressions coming from a search started with an image: Google Lens, Circle to Search on Android, an image uploaded to Google Search, or the 'Search image with Google' right-click in Chrome. The filter exists in the Search results report and in the Generative AI features report. It only shows up if your site already receives that kind of traffic.

How do I get my business to show up in Google Lens?

By publishing your real photos on your own site, where Google can read them: in a standard image element, with alt text describing what you see, a meaningful file name, and next to text that says what it is, what it costs and where it is. A photo search contains no keyword: the image makes the match, the text confirms it. For the shopfront, your Google Business Profile and its photos do most of the work.

Do stock photos hurt my website?

They are not penalised, but they do not identify you. A stock photo lives on hundreds or thousands of sites: when a customer photographs a croissant, nothing lets Google connect it to your bakery rather than another. Only a photo that you alone own can point to you. On the pages that matter, replace them with your own, even taken on a phone.

Is posting photos on Instagram enough?

No. Instagram decides what it lets search engines index, and a photo found there sends the customer to Instagram, not to your site, your hours or your booking button. Instagram is still very useful for being seen. But to be found through a photo search, the same images also need to live on your own site, on pages that say what they show.

I have no multimodal data. Is that a problem?

Not necessarily. The filter only shows figures once image searches have led to your site, and for a small business the volumes start very low. A zero is a baseline. It becomes a signal if it stays at zero after you have published your real photos on well-described pages. Note the figure now so you can compare in three months.

If you want to know where you stand, send me your site address: I'll run the photo test on your key products, spot the images Google cannot read, and tell you in one page what to change. It is one of the workstreams of my work on AI visibility, and it is built into every business site I build: real photos, properly described, that machines can connect to your address.