Case Study: How a Skincare Brand Handled 20,000 Product Images in One Pipeline

A skincare brand selling serums, moisturizers, cleansers, and sunscreen products operated across four sales channels: their own Shopify store, Amazon EU, eMag, and a B2B wholesale site. With 3,200 SKUs and an average of 6 images per product, their catalog held approximately 20,000 product images. Each channel had different size, format, and naming requirements. Product image handling in this environment was consuming 22 hours of team time per week.

The Challenge

Product photography was handled by an external studio that delivered high-resolution TIFF and JPEG files at 5,000 x 5,000 pixels. The raw files arrived naming-free — the studio used sequential shot numbers (DSC_0001.jpg through DSC_8400.jpg for the most recent shoot). The brand’s catalog team was responsible for everything that happened between receiving raw files and images appearing correctly on each channel.

The four channels had requirements that were incompatible without channel-specific processing:

• Shopify: Square images at 2048 x 2048 px, JPEG with white background, filename format: {SKU}-{sequence}.jpg (e.g., SERUM-HA01-1.jpg).
• Amazon EU: Square images at 2000 x 2000 px minimum (2500 preferred), white background, JPEG, filename format: {ASIN}_{sequence}.jpg, with the main image file named MAIN. Technical specifications required no additional text, graphics, or watermarks in the main image.
• eMag: Maximum 1200 x 1200 px, JPEG or PNG accepted, filename must match the eMag product ID (not the brand’s SKU), maximum file size 500KB per image.
• B2B wholesale site: 800 x 800 px, JPEG, watermarked with a low-opacity logo in the bottom-right corner, filename using the brand’s internal product code.

A single product image from the studio had to be processed four times — once per channel — resulting in four files with different dimensions, different compression levels, different names, and in one case a watermark applied. For a shoot delivering 400 new product images, that meant 1,600 processed files to generate before any image was published.

The team’s workflow was built around Photoshop batch actions and Bridge for bulk operations. The process:

1. Receive RAW files from studio, rename manually to match SKU.
2. Run Photoshop batch action to resize and convert for Shopify.
3. Run a second batch action for Amazon (different size, check compliance).
4. Run a third batch action for eMag (resize, compress to file size limit).
5. Run a fourth batch action for wholesale (resize, apply watermark).
6. Manually rename each file set per channel naming convention.
7. Upload each set to the correct channel.

The Photoshop actions handled steps 2 through 5 reasonably well. Steps 1, 6, and 7 were fully manual. Step 1 alone — renaming 400 RAW files by matching them visually to product records — took one person approximately 9 hours per shoot.

Three categories of errors were recurring. First, rename errors: the wrong RAW file renamed to the wrong SKU, causing the wrong product image to appear on one or more channels. In the 12 months prior to implementing a new solution, this happened 23 times across channels, with 7 instances where a customer on Shopify was shown the wrong product image and placed an order. Second, compression errors: eMag rejected approximately 6% of images per upload batch for exceeding the 500KB file size limit, requiring manual re-export at higher JPEG compression. Third, Amazon compliance failures: the Amazon main image check rejected images where the background was not a pure white (RGB 255,255,255) — off-white backgrounds were flagged. This happened for approximately 4% of uploads and required re-shooting or retouching.

[SCREENSHOT: MicroPIM image pipeline overview showing a single product with 6 source images linked, and 4 channel output columns — Shopify, Amazon, eMag, B2B Wholesale — each showing the processed image thumbnail, file size, and dimensions confirmation]

What They Tried First

The team’s first attempt at reducing the manual workload was building a more structured Photoshop action set combined with a naming convention spreadsheet. The spreadsheet listed SKUs with their corresponding RAW file numbers, allowing the renaming step to be done by referencing the sheet rather than visual matching.

This reduced renaming errors significantly but did not eliminate them — the RAW file numbers still needed to be entered into the spreadsheet manually, which introduced a data entry error rate. It also did nothing to address the four-pass processing requirement or the file size and compliance issues.

The team also investigated Adobe Lightroom’s export profile system as an alternative to Photoshop batch actions. Lightroom’s export presets could theoretically handle multiple channel outputs in a single session. However, Lightroom did not support conditional filename generation based on external data (the channel-specific naming conventions required pulling data from the catalog), and the watermark configuration was less flexible than what the wholesale channel required.

A freelance developer was consulted about building a custom Python script that would automate the renaming and batch processing pipeline. The developer produced a working prototype that handled Shopify and Amazon. When they scoped the eMag file size constraint — which required iterating JPEG quality settings until the output fell below 500KB while staying as high quality as possible — the complexity increased significantly. The prototype was never completed to production quality, and the developer moved on to another project after two months.

The Solution

The brand implemented MicroPIM as the central hub for product image handling, configuring per-channel image processing profiles that define exactly how each image should be transformed for each destination.

Step 1: Configuring Channel Image Profiles

Each sales channel was configured with a specific image profile in MicroPIM. An image profile defines:

• Output dimensions (width x height in pixels).
• Format (JPEG or PNG) and JPEG quality target.
• File size constraint (maximum KB, with automatic quality adjustment to meet the limit — used for eMag).
• Background handling (white fill for images with transparent layers or off-white backgrounds — used for Shopify and Amazon).
• Watermark configuration (position, opacity, image file — used for B2B wholesale).
• Filename template (using catalog field variables: {SKU}, {ASIN}, {eMagProductID}, {InternalCode}, plus sequence numbers).

The filename template for eMag was the most complex configuration, because eMag product IDs are external identifiers managed in MicroPIM as a custom product attribute. The template pulls {eMagProductID} from the product record at export time, so each eMag image file is automatically named correctly without manual intervention.

Configuring the four channel profiles took approximately 6 hours, including testing with a sample product set.

Step 2: Linking Source Images to Product Records

The existing 20,000 images in the Photoshop-processed format were imported into MicroPIM and linked to product records by matching filenames (which already used SKUs after the rename step) to catalog SKUs. The bulk import matched 19,340 images automatically. 660 images required manual SKU assignment due to naming inconsistencies — a one-time task that took 3 hours.

For new photography shoots, the studio was given a naming convention for RAW file delivery: instead of sequential shot numbers, the studio now delivers files named {SKU}_{sequence}_RAW.jpg. This eliminated the manual renaming step entirely. Files arrive correctly named and are imported directly into MicroPIM linked to the correct product record.

[SCREENSHOT: Product image library view for a single serum SKU showing 6 linked source images with drag-to-reorder sequence controls, and the “Generate Channel Outputs” button that triggers per-profile processing for all 4 channels simultaneously]

Step 3: Automated Multi-Channel Processing

With source images linked to products and channel profiles configured, MicroPIM generates channel-specific outputs on demand. When a new shoot is imported, the team triggers a bulk processing job for the affected products. MicroPIM processes each source image against each channel profile in parallel and stores the four output files per image — 24 processed files for a product with 6 images.

The eMag file size constraint — which previously required manual re-export for 6% of images — is handled by the profile’s automatic quality iteration: MicroPIM reduces JPEG quality in 2-point increments from 90 until the file falls below 500KB, then stops. The team has not received an eMag file size rejection since implementing this approach.

Amazon background compliance — previously causing 4% rejection rates — is addressed by the white fill setting in the Amazon profile, which replaces any pixel above RGB 240 in all channels with pure white. Images that previously failed Amazon’s background check now pass.

Step 4: CDN Delivery and Channel Sync

Processed images are delivered via MicroPIM’s CDN, with URLs per image per channel profile. For Shopify, images are pushed directly via API. For Amazon, the processed JPEG files with ASIN-based names are exported as a bulk upload package. For eMag and the wholesale site, file packages are generated on a scheduled basis and delivered via FTP.

[SCREENSHOT: Channel sync status panel showing last sync date, image count, and success rate for each of the 4 channels, with a log of the most recent 10 sync events including file count and any error items]

The Results

Five months after implementing the MicroPIM image pipeline:

• Manual image processing time reduced from 22 hours per week to 3.5 hours per week. The remaining 3.5 hours covers new shoot import, product linking review, and spot-check quality verification. The four-pass Photoshop workflow is gone.
• eMag file size rejection rate: 0%. Automatic quality iteration handles the constraint without manual re-export.
• Amazon background compliance rejections: 0%. The white-fill processing in the Amazon profile eliminates off-white background failures.
• Rename errors eliminated. New shoots arrive pre-named by the studio. Historical images are correctly linked in MicroPIM. The 23 cross-channel image errors recorded in the prior 12 months dropped to 1 in the first 5 months (a single case where a studio file was mislabeled at delivery).
• New product time-to-live for images reduced from 4 days to 6 hours. Previously, a new shoot required Photoshop batch processing over 1-2 days before images could be published to any channel. Now, the studio delivers pre-named files, MicroPIM processes all four channel outputs in approximately 40 minutes, and the team reviews and approves before publishing.
• Team capacity reclaimed. The two team members who spent the most time on image processing have been reassigned to product content quality and copy improvement work — tasks with direct impact on conversion rate.

The brand’s creative director noted that the studio naming convention change was the most impactful single change in the workflow: eliminating the 9-hour manual rename step per shoot was the clearest productivity gain.

Key Takeaways

• Product image handling at scale requires per-channel processing profiles, not separate Photoshop batch actions. The difference is that profiles are configuration-driven and source-image-agnostic — you define the rules once, and they apply to every image processed through the channel.
• Naming conventions enforced at the source — in this case, the photography studio — eliminate the most error-prone manual step in most image workflows. It is worth a brief conversation with your studio to align on naming format.
• Automatic file size iteration for channels with hard KB limits (like eMag’s 500KB cap) is more reliable than a fixed JPEG quality setting. A fixed quality produces varying file sizes depending on image complexity.
• Background compliance for Amazon’s white background requirement is best handled as a processing step rather than a retouching task. Most failure cases are minor tonal deviations, not fundamental photography problems.
• The real cost of a broken image workflow is not the labor time — it is the incorrect product images reaching customers and the conversion rate impact of compressed or improperly formatted channel images.

If your catalog has more than 1,000 SKUs and you are still running per-channel Photoshop batch workflows, the processing time and error rate described above will be familiar. MicroPIM’s image pipeline handles per-channel dimension specs, file size constraints, watermarking, naming templates, and CDN delivery in a single automated workflow. You can configure your first channel image profile and test it on a sample set of products the same day you sign up at app.micropim.net/register.

Related Reading

• CDN for Product Images
• Case Study: Jewelry Brand Marketplace Images
• Multi-Channel Ecommerce Strategy

Frequently Asked Questions

What source image formats does MicroPIM accept for product image handling?

MicroPIM accepts JPEG, PNG, TIFF, and WebP source images. TIFF files from photography studios are fully supported — they are converted and processed according to channel profile settings on output. For RAW camera files (CR2, NEF, ARW), the files need to be converted to TIFF or JPEG before import, which most studios handle as a standard step in their delivery process.

How does the automatic file size iteration work for channels with a hard KB limit?

When a channel profile specifies a maximum file size, MicroPIM processes the image at the starting JPEG quality level (typically 90), checks the output file size against the limit, and reduces quality in configurable increments until the output is within the limit. The minimum quality floor is configurable per channel profile — for eMag, the floor is set at JPEG quality 60, which is sufficient for the 1200 x 1200 px output size while maintaining acceptable visual quality. If an image cannot meet the file size limit at the minimum quality floor, it is flagged for review rather than delivered below the quality threshold.

Can MicroPIM handle product image sequences and ensure the correct image is set as the main image per channel?

Yes. Image sequence is managed at the product level with a drag-to-reorder interface. Sequence one is always exported as the main image for channels that distinguish main from additional images (Amazon’s MAIN naming requirement is handled by this). Channel profiles can also define which sequence positions to export — for example, the wholesale channel might export only sequences 1 and 2, while the Shopify channel exports all 6.

How are images updated when the studio delivers revised shots for existing products?

New images from the studio are imported and linked to the product record. If the new image replaces an existing sequence position, the team can swap the source image in MicroPIM and re-trigger channel processing for that product. The channel outputs are regenerated from the new source, and the updated images sync to connected channels. Previous image versions are retained in MicroPIM’s image history for the product and can be restored if needed.