How Harlo Cut Catalog Production Time by 70% With AI

Priya Mehta had worked in e-commerce long enough to know that catalog images decide more than aesthetics. They decide whether a product gets clicked, whether it converts, and whether it passes the image quality gates that Amazon and major marketplace partners enforce before listing.

Harlo, the direct-to-consumer pet accessories brand she managed, was restocking a full seasonal collection — 90 SKUs across collars, leads, beds, and travel accessories. Every SKU needed a clean white-background main image for marketplaces and a lifestyle scene for the Shopify PDP. The previous approach was to send supplier photos to an agency, wait two weeks, and pay per image.

With a restock timeline that had compressed to six weeks due to a shipping delay, two weeks of production time was no longer available.

For e-commerce catalog work at scale, the variable that matters most is throughput — how many finished images per day. AI background removal changes that variable enough to restructure the entire production timeline.

The Catalog Bottleneck Isn't Photography — It's Post-Processing

Harlo's supplier photos were usable as source material. They were sharp, well-lit, and showed the product clearly. What they weren't was marketplace-ready: mixed backgrounds, inconsistent shadows, varying color temperatures.

The agency workflow's bottleneck wasn't skill — it was the per-image manual selection work that even a skilled editor has to complete one product at a time. For 90 SKUs each needing two output formats (white background + lifestyle scene), that's 180 individual edits before any creative work begins.

According to a 2025 Marketplace Seller Report by Jungle Scout, listings with professionally edited white-background main images convert at 23% higher rates than those using unedited supplier photos, even when the product is identical. For Harlo, running at the volume of a seasonal restock, that conversion difference compounds across every SKU.

The question Priya needed to answer was whether AI background removal could handle pet product edges — collars with textured weave, mesh leads, soft fabric beds — well enough to produce marketplace-quality main images without per-image correction.

Four Days From Supplier Photos to Marketplace-Ready

The first test was the category most likely to cause edge problems: a padded fabric dog bed with irregular texture and a fringe trim along the edge.

Background Remover isolated the bed cleanly. The fringe edges retained individual strand detail. The soft fabric boundary showed no halo or clipping artifacts. Priya ran the full category — twelve bed SKUs — the same afternoon. Every image passed her own quality check without manual correction.

In our experience with soft-goods e-commerce, fabric edges are frequently cited as the category where AI background removal breaks down. What we've found is that the failure mode is almost always at the input stage rather than the processing stage: source images with low contrast between the product edge and a similarly-toned background will produce ambiguous edges regardless of the tool. Harlo's supplier images had consistent lighting with clear product-to-background contrast, which is the actual predictor of processing quality.

By the end of day two, 60 of the 90 SKUs had clean isolated-subject files. Visual Enhancer ran as a final sharpening pass on each image, correcting any slight softness from the removal step. The isolated subjects were placed on a standard white background at 2048 × 2048px for Shopify and exported as 2000px JPEGs for Amazon.

For the lifestyle scenes, Image Expander took the isolated product and extended it into a contextual environment — warm home setting, outdoor path, travel scenario — appropriate to the product category. The lifestyle images didn't need to match the precision of the white-background main images; they needed to communicate use context and brand aesthetic. This step added a half-day across the full catalog.

What 70% Reduction in Production Time Actually Means

Harlo's previous catalog production took twelve to fourteen days with the agency: one week for editing, three to four days for round-trip revisions, final delivery at the end of week two. Cost: a per-image rate that across 180 final deliverables was the largest single line item in the restock launch budget.

The AI workflow took four days, with Priya managing the process directly. The final images passed Amazon's technical requirements on first submission. The Shopify PDP update went live before the inventory arrived — which meant the listing was indexed and indexed correctly before the first units were available to ship.

The downstream effect of being early rather than concurrent with inventory arrival is measurable: new listing indexing takes time, and starting the clock before restock rather than after it means the listings have ranking history before peak demand begins.

For the e-commerce catalog use case, the value of AI background removal isn't primarily the cost savings — though those are significant. It's the compression of the production timeline that makes earlier listing possible, and earlier listing that compounds into better organic placement.

For the full workflow behind this approach, see AI Product Photography Without a Studio and E-commerce Catalogs.