The purchase-confidence layer AJIO Business didn't have
A PDP redesign for AJIO Business, India's largest B2B fashion wholesale platform. It introduced pack-type nomenclature, a customisable MOQ flow and an e-Sampling mechanism to build retailer purchase confidence across three product categories.
- Role
- Product Designer + content design
- Scope
- PDP · MOQ flow · e-Sample flow
- Team
- PD + UX Research
- Platform
- Mobile (Android)
- Problem
- Retailers couldn't tell what was inside a packuntil after buying, couldn't customise a pack for their region and couldn't test a new brand below the minimum order. All three gaps pushed purchase intent offline.
- My move
- Interviews across footwear, apparel and sarees in three cities, then a PDP rebuilt as a confidence ladder: a named pack-type taxonomy explained on the card, a customisable MOQ flow and e-Sampling for first buys from unknown brands. I owned the content design as well as the product design.
- Outcome
- Shipped on a platform serving 8.8 lakh monthly active merchants at 53.2% WAU/MAU stickiness (platform-level figures, honestly attributed). User testing validated that pack composition was clear without expanding a single card.
- Learning
- Settle the pack terminology with retailers via a card sort before locking the component architecture, so the UI stops explaining what the data structure should have named.
Retailers had different ideas of a good pack. The underlying job was always the same: certainty before commitment.
The job under every category's mental model
The PDP is the moment of decision
Everything upstream leads to the PDP: it's where a retailer decides whether a product fits their shop, their region and their customers.
AJIO Business connects retailers across India with brands and sellers in apparel, footwear and sarees. Retailers browse products, evaluate packs and place bulk orders to stock their shops.
I worked on the PDP as the product designer alongside UX Research, and owned the content design: how each new feature was named, labelled and explained at the point of decision.
- “Will this sell?”
- “Do sizes fit my region?”
- “Can I trust this brand?”
- “Can I test before buying?”
Three broken stages of the same journey
Ambiguous packs, no customisation and no way to sample: three gaps in the same buying journey, all ending in dead inventory or offline workarounds.
Confusing pack system
An “assorted pack” could mean anything: 2 pieces per size, or 4 in one size and nothing in the rest. Retailers couldn't tell until after they bought. Wrong size distribution meant dead inventory.
No customisation path
If a retailer needed a pack weighted toward a fast-selling size, the platform had no path for it. They went offline to get it: a whole category of purchase intent that couldn't happen on AJIO Business.
No way to sample
For a retailer meeting a new brand, the minimum order was the only option. No low-commitment way to test whether something would sell before committing to a full pack.
Three categories. Three mental models. One broken page.
Field interviews in three cities revealed that each category defines a “good pack” differently, but every retailer wants the same thing: certainty before commitment.
We ran in-depth interviews across retailers in footwear, apparel and sarees: offline store visits and online sessions across three cities. Category heads and cluster teams gave the seller-side view.
“Smaller sizes sell more: 6, 7, 8, especially 7 and 8. Stock of 9, 10 always stays with me.”
Size needs vary sharply by region. A national size pack creates dead inventory for anyone who doesn't match the average.
“I don't want to re-buy a full pack when some 8s and 9s are left. I wait for it to sell out first.”
The platform had no path for cut-set orders. Retailers covered this gap offline: purchase intent AJIO Business was losing entirely.
“Offer samples or small-quantity packs to try and buy, especially for new brands and high price points.”
First orders from new brands carry real financial risk. Sampling was the missing trust mechanism across every category.
Five decisions that built the confidence ladder
Every decision removes one reason to hesitate: scan packs without expanding, understand a pack from its name, customise when the catalogue falls short, sample when trust is missing.
123- 1Collapsed pack cards: scan to compare, expand to commit.
- 2MOQ customise: real-time size availability, minimums shown up front.
- 3e-Sample: order below MOQ to test a new brand before a full pack.
Collapsed pack cards, expandable on demand
Retailers need to scan all packs quickly to compare, and to evaluate one pack in detail before buying. Both jobs happen on the same screen. Collapsed by default, a card shows what you compare on: pack type, sizes, piece count, price. Expanded, it reveals the full size-to-quantity table.
A pack-type taxonomy, named and explained on the card
The work wasn't just visual, it was semantic. We defined six pack types, each with a name, an icon and inline copy that lived directly on the card: not in a help article, not behind an info icon.
An MOQ flow for the pack the catalogue didn't have
When existing packs didn't match a shop, retailers had no option on the platform. The MOQ flow gave them one, for two entry points: a retailer who already knows their size requirement and one who discovers customisation while browsing. The customise screen shows size availability in real time with clear minimums, so there are no surprises at checkout.
e-Sampling for the first buy from an unknown brand
The highest-friction moment isn't the buy decision, it's the first buy from a brand a retailer has never stocked. Research said it consistently: try-and-buy was the only way they'd take a chance. e-Sampling let a retailer order below the standard MOQ to test a product before committing to a full pack.
Scope discipline: what we scrapped and deferred
A compact pack component read cleanly for footwear but broke for sarees, so it wasn't scalable. Pack-type filtering on the PLP depends on pack types being tagged upstream in the catalogue layer, so we designed the PDP ready for it and let the filter wait on another team's timeline.
Shipped to 8.8 lakh merchants
The attributable proof is the usability testing on the redesigned PDP: merchants could read and compare pack composition from the collapsed cards alone, without expanding a single one. The two jobs the card had to do, scan to shortlist and expand to commit, both held up.
Honest attribution on the big numbers: these are platform-level figures for AJIO Business. No isolated attribution exists for the PDP specifically. What is true is that each feature directly addressed a high-friction gap that generative research named.
What I'd do differently
Settle the pack language with retailers before locking the components: the UI ended up explaining what the data structure should have named.
The pack-type taxonomy should have been settled with retailers before the component architecture was locked. Defining the icons, labels and inline copy was necessary work, but it was carrying a load a cleaner catalogue layer should have resolved upstream. With the runway again, I'd run a structured card sort on pack terminology before drawing a single wireframe: shared language would have made every downstream decision faster and the components more honest.