The challenge
The problem
Mole Valley Farmers' product catalogue spans tens of thousands of SKUs — from fencing and livestock equipment to power tools, clothing, and garden supplies. The data in their PIM was functional but thin: enough to list a product, but not enough to sell it well online.
For a business operating 50+ stores alongside a growing ecommerce channel, the gap between in-store and online was becoming a real problem. In-store, customers could pick up a product, read the packaging, and ask staff. Online, they were landing on pages with a product name, a price, and very little else.
The consequences were tangible: filters that couldn't work because the underlying attributes didn't exist, spec tables that were either empty or incomplete, and product descriptions that were too generic to rank or to help a customer make a decision. The team knew the data needed enriching — but with a catalogue this broad, doing it manually wasn't realistic.
The product data existed — scattered across supplier spec sheets, product images, packaging, and the web — but none of it was structured, connected, or sitting in the right fields.
The solution
The approach
Rather than hiring a team to manually research and re-key product data, Mole Valley Farmers used SKULaunch's enrichment agents to pull structured information from every available source — their existing PIM records, the open web, and product images — and map it back to the right fields automatically.
Web search agents were deployed across priority categories to find and fill gaps in product data. For each SKU, the agents searched manufacturer sites, datasheets, and product listings to pull in missing specifications, dimensions, materials, certifications, and technical details — then mapped them against the category schema.
Image extraction was used to pull structured data directly from product images. Where packaging, labels, or product shots contained specs, ratings, or feature callouts, SKULaunch's vision agents extracted the relevant information and added it to the product record — eliminating the need to manually transcribe what was already visible in the imagery.
Content generation turned enriched attribute data into publish-ready product descriptions and SEO copy. Rather than writing from scratch, the AI agents worked from the structured data that now existed — producing descriptions that were specific, accurate, and optimised for search, at a pace the team could never have matched manually.
All of this fed back into the catalogue to power the things customers actually care about: filters that work, spec tables that answer real questions, and product pages that give confidence to buy.
The results
Within the first phase of enrichment across priority categories, the impact on catalogue quality and site experience was significant. Based on benchmarks from comparable SKULaunch enrichment deployments, the following results are typical for retailers of this scale.
85%
of priority SKUs enriched to publish-ready
3x
increase in filterable product attributes
-70%
reduction in manual content enrichment time
We had the products. We had the images. We even had most of the data — it just wasn't in the right shape to do anything useful with online. SKULaunch pulled it all together and gave us a catalogue that actually works for our customers.
eCommerce Manager, Mole Valley
What’s next
With the core enrichment pipeline in place, the team is expanding coverage to further product categories and exploring more advanced use cases.
The team is enabling AI-powered agents to identify missing attributes across the catalogue, suggest values based on product context and category norms, and generate SEO-optimised descriptions — turning incomplete supplier submissions into publish-ready records with minimal manual effort.
Avg. supplier onboarding time
↓ 90%
Data completeness on first submission
↑ 35pp
Manual rework hours per week
↓ 80%
Active suppliers submitting data
↑ 3-4×
Time-to-publish (new product)
↓ 78%
Ops team time on admin tasks
↓ 40pp
• Extending enrichment agents to cover the full online catalogue across all departments
• Using image extraction to capture spec data from new supplier product imagery at the point of onboarding
• Channel-specific content variants — adapting descriptions and attributes for marketplace listings alongside the core site
• Automated completeness scoring to flag SKUs that drop below publish-ready thresholds as new products are added
.webp)
.jpg)
.jpg)
.jpg)
.png)