Install Retouchable in ChatGPT Retouch your product photos from a ChatGPT chat. Install in ChatGPT
  1. Turn on Developer mode

    In ChatGPT, open Settings → Security and login and switch on Developer mode.

  2. Copy the address, open ChatGPT

    https://api.retouchable.ai/api/mcp

    Copy & open ChatGPT ↗
  3. Create the app

    On ChatGPT's Plugins page, select + and choose Create MCP App. Name it Retouchable, paste the address, set authentication to OAuth and create it.

  4. Sign in to Retouchable

    ChatGPT opens our sign-in. New to Retouchable? That starts your 14-day free trial, no card.

Then, in a chat, choose Retouchable under Developer mode and ask: “Grade the photos on my store” or “Put my product photo on white.”

Needs ChatGPT Plus, Pro, Business, Enterprise or Edu, on the web. Menus look different? OpenAI's steps

Hover Images on Collection Grids: Do They Convert?

A second image on hover is a free extra look at every product in the grid, but only if it shows something the first image could not.

|conversion optimization e-commerce Shopify

A hover image is the second product photo a collection grid shows when a shopper points at a card. Whether hover images convert depends almost entirely on what that second image shows. A hover that reveals the back of a jacket, the product in a hand, or the inside of a bag answers a question before the click. A hover that shows the same product from a slightly different angle on the same white background adds a second image download per card and answers nothing.

Most merchants turn the setting on because the theme offers it, and never look at it again. That leaves the decision to whatever happens to sit in position 2 of each product's media list, which is usually whichever photo was uploaded second. On a catalog of a few hundred products, the hover layer of the collection page is effectively random.

This guide covers when secondary images help, when they hurt, what a good one shows by category, how to keep them consistent across a grid, and how to tell whether yours are working.

What a Hover Image Actually Does on a Collection Grid

A collection grid has one job: get a shopper to open the right product page. Every card competes on a single thumbnail, and the shopper decides in a glance whether a product is worth a click. The first image decides most of that click-through. The hover image is a second chance inside the same glance.

On desktop, a secondary image on hover does three useful things when it is chosen well:

  • It answers the next question without a page load. What does the back look like? How big is it? Is it shiny or matte? A shopper who gets that answer on the grid clicks with more intent, or moves on without wasting a visit.
  • It signals that the card is interactive. A change on hover tells the shopper the grid responds, which encourages scanning more products.
  • It adds context the clean first image deliberately leaves out. The featured image is usually a plain packshot built to keep the grid consistent. The hover can carry the lifestyle or on-model view that would make the grid look messy if it were the first image.

That last point is the real argument for hover images. They let you keep a disciplined, uniform first image across the grid while still giving shoppers the richer view on demand.

2Images per card once hover is on
0Hover events on a touch screen
#2Media position most themes use for hover

When Secondary Images Help Conversion, and When They Hurt

Whether a hover image helps conversion is not a yes or no question about the feature. It is a question about each product's second image. The same setting can help one collection and hurt the next.

Hover images tend to help

  • The second image shows something new: scale, fit, back, interior, texture
  • The first image is a strict packshot and the hover adds context
  • Every product in the grid has a deliberate position 2 image
  • Most traffic browses the collection on desktop
  • Products are bought on visual detail (apparel, bags, decor)

Hover images tend to hurt

  • The second image is a near-duplicate of the first
  • Some products have no second image, so cards behave differently
  • Hover images differ in crop, background, or scale from card to card
  • Pages are already heavy and the grid loads slowly
  • Products are bought on specs, not looks (parts, consumables)

The failure modes are quieter than the wins. A near-duplicate hover does not lose a sale on its own. It just spends bandwidth and attention for nothing. An inconsistent hover layer is worse: when one card flips to a model shot, the next to a detail crop, and the third does nothing, the grid feels unfinished, and the uniformity you built into the first images is undone the moment a shopper moves the mouse.

There is also the cost of the extra image itself. Turning hover on adds a second image request per card, which theme settings and lazy loading can defer but not remove. On a grid of 24 or 48 products, that is real weight. If the second image is not earning its place, it is slowing the page that does.

Check before you switch it on

Filter a collection to the products with only one image. If more than a handful of cards would have no hover, either fill those gaps first or leave the setting off. A grid where some cards respond and others do not reads as broken.

The Touch-Screen Problem

Hover does not exist on a phone. A finger cannot point at a card without tapping it, and a tap opens the product page. On a touch screen, the secondary image is never seen on the grid at all.

For most stores, mobile is the larger share of collection page traffic, so the honest framing is this: a hover image is a desktop feature that costs every visitor something and benefits only some of them. That does not make it a bad idea. It means two things for how you use it:

  • Never put essential information only in the hover. If shoppers need to see that a product comes in a set of three, or that it is much smaller than it looks, that belongs in the first image, the title, or a badge. Our guide to product images on mobile covers what the mobile grid has to carry on its own.
  • Measure desktop and mobile separately. A blended conversion rate will dilute any effect the hover image has, because most of the sessions in it never saw the hover.

Some themes offer swipeable image carousels on collection cards as the mobile counterpart. Those are a different design decision with their own trade-offs (smaller tap targets, accidental swipes while scrolling) and should be tested on their own, not assumed to inherit whatever result hover produced on desktop.

What the Second Image Should Show, by Category

The rule is simple: the hover image should answer the most common question the first image leaves open. What that question is depends on the category.

CategoryFirst image (grid)Best hover imageAvoid as hover
ApparelFlat lay or ghost mannequinOn-model, full lengthFabric close-up
Bags and luggageFront, on whiteWorn or carried, for scaleSame bag at a 3/4 angle
FootwearSide profileOn foot, or pair from aboveSole only
JewelryPackshot on a plateWorn, for sizeExtreme macro
Homeware and decorPackshot on a plateStyled in a roomBack or base view
FurnitureFront 3/4 on whiteIn a room with a person or known objectDetail of a joint
Beauty and skincareBottle frontTexture swatch or in handIngredient label
Electronics and toolsProduct frontOften better left offPorts and specs

A pattern runs through the table. The best hover images supply scale or context: the product on a body, in a hand, in a room. Those are exactly the views a shopper cannot get from a clean packshot, and exactly the views that would make a grid inconsistent if they were the first image. Detail shots, labels and spec views are valuable on the product page, where the shopper has already chosen to look closely, but they are poor hover images because they are hard to read at thumbnail size.

Electronics, parts and consumables are the exception worth taking seriously. When shoppers compare on specification and the products look alike, a hover image rarely changes a decision. Leaving it off keeps the grid lighter.

Making Position 2 Deliberate Across a Whole Catalog

On Shopify, a product's images are an ordered list. Position 1 is the featured image that shows on the collection grid and in search. Most themes that offer a second image on hover, including Shopify's own free themes, take it from position 2. So the hover layer of your entire store is decided by one question: what is in position 2 of each product?

For most catalogs, the answer was never decided. Position 2 is whatever was uploaded second, which might be a back view on one product, a size chart on another, and a variant colour on a third. Fixing the hover layer means making position 2 a standard, the same way a collection image standard fixes position 1:

  1. Pick one hover role per collection. On-model for apparel, in-room for decor, in-hand for small goods. One role per collection, not per product.
  2. Match the hover image's framing to the grid. Same aspect ratio as the first image, the product at a similar size in the frame. A hover that zooms in or crops differently makes the card jump.
  3. Keep variant images out of position 2. If a theme switches images when a colour swatch is chosen, a variant photo in position 2 can make the hover show a different colour from the card's default. Put variant images later in the list.
  4. Fill the gaps. List the products with fewer than two images and decide, product by product, whether to add a hover image or exclude that collection from the setting.

Reordering by hand in the Shopify admin works for a few products and becomes the bottleneck at a few hundred. Setting the position when the image goes up avoids the second pass. When Retouchable pushes a finished image to a Shopify product, it can place it in a specific position, not only as the featured image, along with its filename and alt text, so a new in-context shot can go straight into position 2 as the hover image. Our guide to Shopify image position and gallery order covers how positions behave across the rest of the gallery.

Pro Tip

Give the hover image its own alt text. It is a different image and often shows something the featured image does not, such as the product worn or in a room. Describing that helps screen reader users and gives search engines one more accurate description of the product.

Producing Hover Images Without a Second Shoot

The usual reason position 2 is weak is that a good second image was never made. Product shoots are planned around the featured packshot, and the in-context or on-model view is either skipped or done for a few hero products only. Filling the gap for a whole collection used to mean booking another shoot.

It no longer has to. The photos a store already has can usually produce the hover image a collection needs:

  • Apparel: a flat lay or ghost mannequin photo can be put on a model, which gives you the on-model hover for every garment in a collection from the images you already sell with.
  • Homeware and decor: a packshot can be placed in a styled scene, giving the in-room context view without moving the product into a room.
  • Small goods: a second angle you already shot can be relit and matched to the featured image's background and framing, so at least the hover is consistent while you plan a proper context shot.

Whatever produces the image, judge it at the size the grid shows it, side by side with the first image. A hover image that is excellent at full size but unreadable at card size has not done its job. Also be honest with shoppers: an AI-generated on-model or in-room image shows the product in a setting it was not photographed in, so the product itself, its colour, shape and details, must stay true to what arrives in the box.

How to Test Whether Your Hover Images Convert

The setting is easy to switch on and off, which makes it tempting to compare two random weeks. That comparison will mostly measure seasonality. A cleaner approach treats the hover layer like any other image change. Our guide to A/B testing product images covers the method in general; for hover images specifically:

  • Test one collection, not the store. Pick a collection where position 2 is deliberate and consistent. Testing a collection with a random hover layer tells you only that random hover images do not help.
  • Segment by device. Look at desktop sessions only. Mobile visitors cannot see the change and will flatten the result.
  • Measure the grid, then the sale. The first metric is click-through from collection page to product page. The second is whether those clicks convert. A hover image that raises clicks but lowers the add-to-cart rate is attracting the wrong clicks.
  • Watch page speed. Compare collection page load times with the setting on and off. If the grid slows noticeably, compress the hover images or let the theme load them later before deciding the feature does not work.
  • Run it long enough. Collection pages get less traffic per product than a homepage. Wait for enough desktop sessions to reach a clear result rather than stopping at the first good-looking week.
ResultWhat it usually meansNext step
Clicks up, conversion steady or upHover answers a real questionRoll the same hover role out to similar collections
Clicks up, conversion downHover oversells or misleadsCheck the hover image matches what ships
No changeHover adds nothing newChange what position 2 shows, or switch it off
Grid slower, no gainWeight without benefitSwitch it off for this collection

The likeliest outcome for most stores is that hover images help in a few visual categories and do nothing in the rest. That is a useful result. It tells you where to spend effort on second images and where to keep the grid as light as possible.

Frequently Asked Questions

Do hover images on collection pages increase conversions?

They can, when the second image shows something the first one does not, such as the product on a model, in a hand, or in a room. A hover image that repeats the first image from a slightly different angle adds page weight without helping shoppers decide. Test it by collection and on desktop traffic only.

Which image does Shopify show on hover?

Themes that offer a second image on hover, including Shopify's free themes, generally use the image in position 2 of the product's media list. Changing which image sits in position 2 changes the hover image on every collection page that product appears on.

Do hover images work on mobile?

No. Touch screens have no hover state, so mobile shoppers never see the secondary image on the grid. Anything shoppers must know should be in the first image, the title, or a badge, and hover tests should be measured on desktop sessions.

Do hover images slow down a collection page?

They add a second image per product card. Theme settings and lazy loading can defer it, but it is still extra weight. Compress hover images to the same standard as the featured image and compare load times with the setting on and off.

What should the second product image show?

Whatever answers the most common question the first image leaves open. For apparel that is usually the garment on a model, for bags and jewelry the product worn for scale, and for homeware the product styled in a room.

Make every product's second image count

Turn the photos you already have into consistent on-model, in-room and in-hand hover images, and push them straight into position 2 on Shopify.

Try Retouchable Free No credit card required