Google Multisearch: Searching With Images and Text Together

Google multisearch is a new way to search with a picture and words at the same time. You point Google Lens at something, then add a few words to tell Google what you actually want, like a different colour or a matching item.

Google announced it on 7 April 2022. Right now it is a beta feature, in English, in the US, inside the Google app on Android and iOS. Below I explain how it works, how to try it, and what it means if you run a website or an online store.

What Is Google Multisearch?

Until now, you had two separate choices. You could type a query, or you could search with an image using Google Lens. Multisearch joins the two into one search.

The idea is simple. Some things are hard to describe in words, and a photo alone does not say what you want. A photo plus a short phrase covers both gaps.

Google gave three examples in its announcement of multisearch:

• Same item, different colour: take a screenshot of an orange dress and add the word “green” to find a similar dress in green.
• A matching item: take a photo of your dining set and add “coffee table” to find a table that goes with it.
• Information about a thing: take a photo of a rosemary plant and add “care instructions”.

Notice the pattern. The image says “this thing”, and the text says “but tell me this about it” or “but change this part”.

Where It Is Available Right Now

• Country and language: the United States, in English only.
• Status: beta, so Google is still testing it.
• Where: the Google app on Android and iOS, through the Lens camera icon. You need the latest version of the app.
• Not available: desktop search, and countries outside the US for now.

Google has not given dates for other countries or languages. I would not plan any work around a wider rollout until Google confirms one.

What Makes It Work

Google says multisearch is built on recent progress in AI that helps it understand images and language together. It also says it is looking at how MUM could make results better for more kinds of questions.

MUM stands for Multitask Unified Model. It is an AI model Google introduced in 2021 that can understand information across text and images. Google showed an early idea of this kind of Lens search at its Search On event in September 2021, so the beta is that idea reaching real users.

How to Use Multisearch in the Google App

If you are in the US and want to see it for yourself, it takes less than a minute. I always suggest trying a new search feature with your own products before forming an opinion about it.

1. Update the Google app. Multisearch needs the latest version on Android or iOS.
2. Tap the Lens icon. It is the camera icon on the right side of the search bar.
3. Pick an image. Take a new photo of something near you, or choose a screenshot or photo from your gallery.
4. Let Lens find matches. Lens shows visual matches first. You can adjust the box around the object you care about.
5. Swipe up and tap “+ Add to your search”. This opens a text box under the image.
6. Type a short phrase. Add a colour, a brand, an item type or a question, such as “green”, “coffee table” or “care instructions”.
7. Review the results. Google combines the image and your words and shows results that fit both.

Keep the text short. The image already carries most of the meaning, so you only need to add what is missing.

What Multisearch Is Good At (and Where It Struggles)

Google says multisearch works best for shopping searches right now. That makes sense. Products have clear visual traits like colour, shape and pattern, and shoppers often want “this, but slightly different”.

Here are the kinds of searches where I expect it to be most useful:

• Changing one attribute: “this shirt, but in blue” or “these shoes, in leather”.
• Finding matching items: a lamp that suits a sofa, or a cushion that matches a rug.
• Adding a brand: “this style of backpack” plus a brand name.
• Asking about an object: how to care for a plant, how to clean a fabric, or what a part is called.

Searches that are not about shopping do work, but Google itself says the results are strongest for shopping. Since it is a beta, I would expect some odd results while Google tunes it.

Why Multisearch Matters for SEO

Most searchers will not change their habits overnight. Typing is still the default, and multisearch is limited to one country and one app. So this is not a reason to panic or rebuild your site.

But it is a clear signal. Google keeps adding ways to search that do not start with typed words: voice, Lens, and now images plus text. In my experience, the sites that benefit from new search features are the ones whose basics are already in good shape when the feature grows.

There are three things I take from this launch:

• Images are now a starting point for queries. A shopper can begin with a photo of your product, or a competitor’s product, and end up on a results page.
• Visual traits matter more. If Google needs to match “this dress, but green”, it helps when your pages clearly say the item comes in green.
• Product data matters more. Colour, material, brand and size need to be easy for Google to read, not hidden in images or scripts.

Who Should Pay Most Attention

If you sell physical products with a strong visual side, this is most relevant to you. Think fashion, furniture, home decor, beauty, plants and garden, and accessories.

If you run a service business or a B2B software site, multisearch will probably not change much for you this year. I’d spend that time on your core pages instead.

How to Prepare Your Site for Visual Search

Google has not published any special guidance for multisearch. There is no tag to add and no setting to switch on. What you can do is make your images and product pages as easy as possible for Google to understand.

These are the steps I would take, in this order.

1. Use Original, Clear Product Photos

If ten stores use the same photo from the manufacturer, Google has little reason to pick yours. Your own photos give you a better chance of being the page that matches.

• Show the product clearly, on a clean background, with good lighting.
• Add photos from more than one angle.
• Show real use, such as a chair in a room or a dress being worn.
• Give each colour option its own photo, so Google can see that a green version exists.

When I review a store, the missing colour photo is one of the most common gaps I see. The product page lists five colours but only shows one.

2. Write Helpful Alt Text

Alt text is a short written description of an image in your HTML. It helps people who use screen readers, and Google uses it to understand what the image shows.

Describe what is in the image in plain words. Include the traits a shopper might add in a multisearch, such as colour, material and type.

<!-- Weak -->
<img src="IMG_4821.jpg" alt="dress">

<!-- Better -->
<img src="green-linen-midi-dress.jpg" alt="Green linen midi dress with short sleeves and a tie waist">

Do not stuff keywords into alt text. One clear sentence is enough.

3. Use Descriptive File Names

A file name like IMG_4821.jpg tells Google nothing. A name like walnut-round-coffee-table.jpg gives a small but useful hint.

I don’t rename old images on a big site just for this, because it can break links. I make it the rule for every new upload instead.

4. Put Images Next to Relevant Text

Google looks at the text around an image to understand it. A product photo sitting beside a clear title, description and spec list is much easier to read than a photo in a gallery with no text.

Your product description should say the things a photo cannot fully show: material, size, care, and which colours are available.

5. Add Product Structured Data

Structured data is code that labels facts on your page in a format search engines read easily. For product pages, Product markup can label the name, image, brand, price, availability and reviews.

<script type="application/ld+json">
{
  "@context": "https://schema.org/",
  "@type": "Product",
  "name": "Round Walnut Coffee Table",
  "image": "https://www.example.com/images/walnut-round-coffee-table.jpg",
  "brand": { "@type": "Brand", "name": "Example Home" },
  "color": "Walnut",
  "offers": {
    "@type": "Offer",
    "price": "249.00",
    "priceCurrency": "USD",
    "availability": "https://schema.org/InStock"
  }
}
</script>

Make sure the markup matches what is visible on the page. If the price or stock status is wrong in the code, fix that first.

6. Keep Product Feeds Accurate

If you sell online in the US, a Google Merchant Center feed gives Google another clean source of product data. Fill in the colour, material, brand and size fields properly, not just the required basics.

In my experience, feeds are often set up once and then forgotten. A quick check every few months catches wrong prices and missing attributes.

7. Make Sure Google Can Crawl and Load Your Images

• Do not block image folders in robots.txt.
• Use real <img> elements with a src, not images loaded only through CSS backgrounds.
• Compress large images so pages still load fast on phones.
• Consider an image sitemap if Google may miss some of your images.

Google’s image SEO best practices cover all of this in more detail. Fast loading also ties into the page experience signals that now apply on desktop as well as mobile.

Content Ideas That Fit Multisearch

Look again at Google’s examples. “Care instructions” and “coffee table” are not just product searches. They are questions people have about things they already own or have seen.

That gives you content ideas:

• Care guides: how to clean a wool rug, how to care for a fiddle leaf fig, how to store leather boots.
• Matching guides: which coffee tables suit a mid-century sofa, which shoes go with a green dress.
• “What is this?” pages: clear pages that name and explain parts, styles or materials, with good photos.

Use your own photos on these pages too. A care guide with a clear image of the actual plant or fabric is more useful to readers and gives Google a better image to match.

If you run review content, the same rule applies: real photos and real detail. That lines up with what Google asked for in the March 2022 product reviews update, which rewards reviews that show first-hand use.

Can You Track Multisearch Traffic?

Not directly, as far as I can see. Google has not announced a separate multisearch report, and Search Console does not show a filter for it.

Here is what I would watch instead:

1. Check the Performance report in Search Console for mobile traffic to product and category pages.
2. Compare the “Image” search type with “Web” to see how much of your traffic comes from image results.
3. Look for new query patterns that mix product types with colours or attributes.
4. Note any change in mobile clicks to pages where you improved photos and alt text.

None of this isolates multisearch, so treat it as a rough signal. With the feature in beta and only in the US, I would not expect a visible change in most reports yet.

Common Mistakes to Avoid

• Only using supplier photos. They make your pages look like every other store selling the same item.
• Leaving alt text empty or stuffed. Both waste a simple chance to describe your image.
• Hiding colour options in a dropdown with no text. If the green version only exists inside a script, Google may not connect it to a “green” search.
• Heavy, slow images. Large uncompressed photos hurt mobile users, and multisearch is a mobile feature.
• Chasing the feature with gimmicks. There is no special trick for multisearch. Good images and clear product data are the whole job.

My Take on Google Multisearch

I think multisearch is one of the more useful search features Google has shipped for shoppers. It solves a real problem: you know what you want when you see it, but you cannot describe it in a search box.

For SEO, the lesson is not new, but it is now harder to ignore. Images are not decoration. They are content that Google reads, matches and ranks. Stores with original photos, clear alt text and complete product data are in the best position as visual search grows.

If you already have an image SEO plan, keep going. If you don’t, it is worth adding to the list of fixes in your SEO checklist for 2022.

Conclusion

Google multisearch lets people search with an image and text together through Google Lens. For now it is a US, English-only beta in the Google app, and Google says it works best for shopping.

You do not need to do anything special for it. Your next step is simple: pick your ten most important product pages and check each one for original photos, a photo for every colour, clear alt text, descriptive file names and accurate Product structured data. Fixing those helps you in regular search today and puts you in a good spot as multisearch grows.