Explore

Problem sizing

What is the Problem/Opportunity?

Context

Strategic initiatives, IPO readiness and Mall, are heavily dependent on the platform’s capability to arrest Fakes (Unauthorised Brand products) and Other restricted product class (Sexual, Pills/ Drugs, etc.)

As of today, the methodology of identification of these non-compliant listings is heavily indexed on DS + Manual Ops processes post upload of catalogs. Also all DS interventions have been designed from a manual QC cost reduction and hence been used to primarily to replace manpower for cost reduction.

Most compliance actions (Deactivations, communication etc) has been post catalog upload, making it a highly reactive process with little to no obligation to change behaviour of source - i.e. Problematic Catalogs/Supplier who try to upload unauthorised catalogs on the platform.

Type of Unauthorised listing on the platform

Brand Fakes - Branded looking fake products uploaded by suppliers who do not have authorisation to sell that particular brand.

Restricted Classes

Sexual/Profane Images

Guns

Tobacco related products

Pills & Drugs

Alcohol

Problem Sizing

Sizing of the Problem is based on the following parameters

Leakage Volume (Probability of Discovery of Leaked Non-Compliant Listing) : Volume of different sub classes that gets leaked onto the platform through our existing QC process

Leakage Visibility Speed (Speed of Discovery): Speed at which these different classes gather views on platform, hence posing risk of discovery

High - Gets first 500 views in 3 days (Avg.)

Medium - Gets first 500 views in 6 days (Avg.)

Low - Gets first 500 views in more than 6 days (Avg.)

Real Estate Providing Visibility (Source of Discovery): Search, an Intent based real estate has easier way of being detected as compared to other exploratory real estates

Based on the 3 factors, Brand is the P0 problem, followed by Non-Brand classes of Sexual and Pills/ Drugs

Point of attack

Most of the restricted classes are discovered via search (primary real estate of view share). Search is indexes products based on - Product Name, Description & Attribute. The primary point of attack becomes reducing keyword based leakages that happen in Pre Live QC process.

Why restricted keyword based catalogs/products are leaked in the system through DS/Manual Intervention?

Bridging Gaps in QC

Bridging Gaps in QC

Problem Area

Decsription

Sizing/Urgency/Scope

SSCAT Filters & Keyword Leakage

Changes related to SSCAT - DS model have several SSCAT filters which filter out selected SSCATs for checking specific brands. Any new SSCAT/SSCAT/Products Uploaded in wrong SSCAT change will cause these checks to fail.

[Brand] - 30% of leakage happening due to this. Viewshare wastage - Keyword are indexed on searches, Keyword leakage leads to Fakes getting high visibility of platform.

[Restricted] - Legal Authorities tend to search on platform to find such illegal listing, keyword based leakages are prone to be indexed by search and has a chance of showing up high in platform ranking on platform.

TAT for Detection

Currently it takes 4-8hrs for the current Brand DS model infringement + Manual QC to get complete (Average Go Live time - 5hrs)

Brand QC delays can lead prolonged period of product being live on Platform

Name/Image Change throgh edit

Catalog Name/Description change through Catalog Edit.

<5% leakages attributed to this bucket

There are no rows in this table

⁠

Why is it important to solve?

Mall being a strategic initiative & reducing apprehension of brands to list on platform due to Fakes/ counterfeits

IPO readiness (Compliant E-commerce platforms strictly restrict upload and discovery of restricted category products)

Supplier Behaviour change - To instil a behaviour change with supplier mindset with clear guidelines and consequences of upload non compliant listing.

Long Term vision for the feature

End State of Compliance could potentially look to provide the following

Methodology to identify PIDs and flag at the time of upload to Supplier to take corrective action

Assigning Compliance score (TBD) for each PID to ascertain whether a given catalog should be held eligible for CT views or Ads

Run automated checks on platform as well as manual SME based checks to flag PIDs and have it added as a feedback loop to the models

Blocking of Catalog uploads for repeat offenders (thresholds on Listings & OC to be defined)

⁠

⁠

Want to print your doc?
This is not the way.

Try clicking the ··· in the right corner or using a keyboard shortcut (

CtrlP

) instead.