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

Measuring the ROI of a Product Image Refresh

A method for proving whether new product photos paid for themselves, built around a baseline, a holdout group and the metrics images actually move.

|conversion optimization catalog management photography costs e-commerce

Most product image refreshes are judged on a feeling. The new photos look better, the team is pleased, and if sales are up the next month the photos get the credit. If sales are down, the season does. Neither verdict tells you whether the refresh paid for itself or whether the next one is worth funding.

Measuring the ROI of a product image refresh is not hard, but it has to be planned before the first new image goes live. Once the old photos are gone, the baseline is gone with them. What you need is a recorded "before", a group of products that keeps its old images for a few weeks as a comparison, a short list of metrics that images genuinely influence, and an honest accounting of what the refresh cost.

This guide walks through that method step by step, with a worked example you can adapt to your own catalog. It applies whether the new images came from a studio reshoot, an in-house retoucher or an AI tool.

What ROI means for an image refresh

ROI is the gain from the refresh minus its cost, divided by its cost. The formula is simple; the work is in being honest about both sides.

The cost side is more than the production invoice. Include:

  • Production: the reshoot, retouching or generation cost for every image in scope.
  • Staff time: briefing, reviewing, rejecting and approving images, at a realistic hourly rate.
  • Rollout: uploading, writing alt text, reordering galleries and checking every product page on mobile.
  • Samples and logistics: pulling stock, shipping to a studio, steaming garments.

The gain side is the extra gross profit (not revenue) the new images produce over a defined window. Using revenue overstates ROI by the full cost of goods. If a refresh lifts revenue by $10,000 on a 40% gross margin, the gain is $4,000.

The formula

ROI = (incremental gross profit over the window − total refresh cost) ÷ total refresh cost. Payback period = total refresh cost ÷ incremental gross profit per month.

Pick the measurement window before you start. Ninety days is a sensible default: long enough to smooth weekly noise and capture return-rate effects, short enough that seasonality does not swamp the signal.

Record the baseline before a single image changes

The baseline is the most commonly skipped step, and without it there is nothing to measure against. Export at least four weeks of data, ideally eight, for every product in scope, at the product level rather than store-wide:

  • Product page views (sessions that reached the product)
  • Click-through from collection and search pages to the product
  • Add-to-cart rate (sessions with an add-to-cart ÷ product page sessions)
  • Conversion rate and units sold
  • Return rate, and the share of returns tagged "not as described" or "looks different"
  • For paid channels: ad click-through rate and cost per click on the products' shopping ads

Shopify's analytics reports give views, cart additions and conversion per product, and GA4's ecommerce events (view_item_list, select_item, add_to_cart) add list click-through if your theme sends them. Save the export as a file with a date on it. Dashboards recalculate, attribution models change, and you want the numbers as they stood.

Also write down what else is planned for the window: promotions, price changes, theme updates, new ad campaigns. You cannot always avoid them, but you need to know when they happened.

Design the rollout so the result can be measured

Swapping every image on the same day is the fastest rollout and the least measurable one. Any change in sales afterwards is mixed with whatever else changed that week. Two rollout designs give you a usable comparison.

Holdout collections

Split the products in scope into two matched groups: similar categories, price points and traffic. Refresh one group now and leave the other on its old images for four to six weeks. Compare how each group's metrics moved from baseline. If both groups rose 5% because of the season but the refreshed group rose 14%, the refresh is worth roughly the 9-point difference. This is a difference-in-differences comparison, and it is the most reliable method available to a typical store.

Staggered rollout

If holding back products feels too costly, refresh in waves a few weeks apart, collection by collection. Each wave's "after" can be compared with the not-yet-refreshed waves over the same weeks. It is less clean than a true holdout, but far better than a single switch-over.

Everything at once

  • Fastest to ship
  • No comparison group
  • Season and promotions mixed into the result
  • ROI becomes a guess

Holdout or staggered

  • A few weeks slower for some products
  • Built-in comparison over the same dates
  • Seasonality cancels out
  • ROI you can defend in a budget meeting

Change the featured image first. On Shopify it is the image in position 1, the one shown on collection grids and in search, so it drives click-through, the earliest metric to move. If you push images with Retouchable's Shopify integration, you can set position, filename and alt text in the same step, which keeps the rollout to one action per product. For a full walk-through of rolling out a catalog by collection, see refreshing a Shopify catalog image set in bulk.

The metrics that move, and when they move

Images influence some metrics quickly and others slowly. Knowing the order stops you from calling a result too early or looking in the wrong place.

MetricWhat the image changesWhere to read itTypical lag
Collection and search click-throughThe featured image competing in a gridGA4 list events, marketplace search reportsDays
Shopping ad click-throughThe image shown in the adGoogle Merchant Center, Meta AdsDays
Add-to-cart rateConfidence from the full galleryShopify product analytics1–3 weeks
Conversion rateConfidence plus checkout behaviourShopify, GA42–4 weeks
Return rateWhether the photo matched the productReturns platform, order tags4–10 weeks

Click-through tells you the featured image is working. Add-to-cart tells you the gallery is persuading. Return rate tells you whether it persuaded honestly. A refresh that lifts conversion but also lifts "not as described" returns has not earned what the conversion number suggests, which is why returns belong in the gain calculation. Our piece on product returns linked to image quality covers how to tag returns so this is measurable.

Watch sample sizes

A product with 40 sessions a week cannot show a reliable conversion change in a month. Measure low-traffic products as a group, and judge individual products only when they have a few thousand sessions in each period.

A worked example

Here is the calculation for an illustrative mid-sized store. The numbers are an example, not a benchmark; substitute your own.

The store refreshed 400 images across 100 products using a studio reshoot. Production came to $14,000 (about $35 an image including retouching, within the commonly quoted range for professional work). Staff time for briefing, review and rollout was 40 hours at $40, or $1,600. Total cost: $15,600.

The 100 products were split into two matched groups of 50. Group A was refreshed; group B kept its old images for six weeks, then was refreshed too.

+6%Holdout group conversion change (season)
+15%Refreshed group conversion change
+9 ptsLift attributable to the images
−1.2 ptsChange in return rate

Before the refresh, all 100 products together earned $60,000 a month in revenue at a 45% gross margin, so $27,000 in gross profit. A 9% lift attributable to the images adds $5,400 a month in revenue and $2,430 in gross profit. The lower return rate saves roughly $300 a month in refunds, return shipping and handling, bringing the monthly gain to about $2,730.

Cumulative gross profit gain (refresh cost: $15,600)
Month 2
$5,460
Month 4
$10,920
Month 6
$16,380
Month 12
$32,760

The payback period is $15,600 ÷ $2,730, or about 5.7 months. Over twelve months the ROI is ($32,760 − $15,600) ÷ $15,600, roughly 110%. Note how sensitive this is to cost: the same lift on a refresh that cost a fraction as much pays back in weeks, which is why production method matters as much as image quality. Our breakdown of the ROI of AI product photography for small brands looks at that cost side in more detail.

Confounders that fake a win (or hide one)

Even with a holdout group, a few things can distort the result. Check each one before you report a number.

  • Promotions applied unevenly. A sale on one collection but not the other breaks the comparison. Either include both groups in every promotion or exclude promotion days from the analysis.
  • Price changes. A price cut during the window will lift conversion regardless of the images. Note every change at the product level.
  • Traffic mix. A new ad campaign sending cold traffic to refreshed products can push their conversion rate down even while the images help. Compare by channel where volume allows.
  • Stock-outs. A bestseller going out of stock in one group drags its numbers. Remove products that were unavailable for a meaningful part of either period.
  • Novelty. Returning customers sometimes click more on anything new. If the first two weeks look far better than the following four, report the later, steadier figure.
  • Theme or app changes. A new gallery app or page speed change affects every product page. Freeze them during the window if you can.
Report a range, not a point

If the lift is 9% with one cleaning rule and 6% with a stricter one, report "6–9%". A range survives scrutiny better than a single number and makes the next budget conversation easier.

Using the result to plan the next refresh

The real value of measuring one refresh is that it tells you where to spend on the next one. Three uses stand out.

Prioritise by gap times traffic. The products with the most sessions and the weakest current images offer the largest absolute gain. A product with heavy traffic and a dated featured image is worth more attention than a low-traffic product with a merely average one.

Find the cheapest lever. If most of the lift came from click-through, the featured image did the work, and the next refresh might only need new hero shots rather than full galleries. If add-to-cart moved most, the supporting gallery images (detail, scale, in-use) earned their cost. Testing individual images is covered in our guide to A/B testing product images.

Set a standard so the next refresh is smaller. Much of the cost of a refresh is fixing inconsistency: different backgrounds, crops and lighting accumulated over years. A written image standard (plate colour, product scale, crop, gallery order) applied to every new product keeps the catalog from drifting, so the next refresh is a top-up rather than a rebuild.

Keep the measurement files with the refresh brief. The next time someone asks whether new photos are worth it, you will have your own answer, from your own catalog, rather than an industry average.

Frequently Asked Questions

How long should I wait before measuring the ROI of new product photos?

Click-through changes show within days, add-to-cart and conversion within two to four weeks, and return-rate effects take four to ten weeks. A 90-day window captures all of them; report early click-through results as a signal, not a verdict.

Can I measure an image refresh without an A/B testing tool?

Yes. A holdout group of matched products that keeps its old images for four to six weeks gives you a comparison over the same dates, so seasonality and store-wide changes cancel out. It needs only your normal analytics exports.

Should I use revenue or profit to calculate image refresh ROI?

Use gross profit. Revenue overstates the gain by the cost of goods sold. Multiply the incremental revenue by your gross margin, then add any savings from fewer returns.

What if my products have too little traffic to measure?

Measure them as a group rather than individually. Combine low-traffic products into a refreshed group and a holdout group of similar size and compare the totals. Individual product results need a few thousand sessions per period to be reliable.

Make your next image refresh cheaper to pay back

Retouchable brings a whole catalog to one consistent standard and pushes the featured image straight to Shopify, so the refresh costs less and the payback comes sooner.

Try Retouchable Free No credit card required