Explore

Trending search recos for sellers

Trending Searches

Flow

Search Tower Generation -> Search Tower Prioritization -> Supplier resume buildup -> Supplier to reco mapping

Search Tower Prioritization

Step I - Tower creation

Aggregation based on taxonomy X category details aggregated from the search query

Input feed - Search broken down by taxonomy (Owner - Sujith)

Step II - Tower cutoff

All towers with 15 days search frequency >=50 (Customizable) to be the ones to be actioned upon

Cutoff value would remain dynamic and can be played around based on the net count of recommendation generated per iteration

Only search tower with at least one taxonomy attribute and mapped to any SSCAT to be filtered

Step III - Key prioritization levers per tower

Following metrics to be computed per tower for global rank allotment

Conversion (O/Se) - Sum(Orders) / Sum(Searches) per tower

Engagement gap - Delta between Avg catalogs viewed per session for the mapped categories (Business SSCATs) and avg catalogs viewed per session for the search tower

To be floored to 0 if engagement for search tower > ssact search engagement

Taxonomy Weight - Count of taxonomy details present in the tower weighted by category taxonomy relevance. Different taxonomies can also be given different weights to alter their priority. Example - Fabric > Colour

Demand Score - #Searches for the tower / #Total Search volume

Current Supply Estimate - #Live products with the search tower taxonomy / #Total live products in the tagged search tower SSCAT

Current Supplier Estimate (Competition Score) - #Suppliers catering to search tower taxonomy / #Suppliers catering to search tower SSCAT

Step IV - Normalizing metrics between 0-1

O/Se normalization post outlier removal

O/Se - Outlier removal - Capping O/Se values higher than upper bound by the upper bound

Upper bound - Q3(75th percentile O/Se) + (1.5 * IQR)

IQR - Difference between 75th and 25th percentile O/Se

Normalize the values as { 1 - (O/Se for tower - Min (O/Se) / (Max(O/Se) - Min(O/se)) } - O/Se would be negatively correlated with the rank

Rest metrics would be normalized between 0-1 via min-max normalization

O/Se , Current Supply Estimate and Current Supplier Estimate would have -ve correl with the final global rank

Step V - Final Score Computation -

Final Score - Sumproduct of lever weights and normalized score

Lever weights would be configurable at SSCAT Level, with the following default values

O/Se (Conversion) - 40%

Demand Score - 20%

Current Supply Estimate - 10%

Taxonomy Score - 15%

Engagement Gap - 10%

Competition Gap - 5%

Final Output -

Search Tower | SSCAT mapping | Taxonomy Details | Prioritization metrics | Global Rank

Supplier Resume Buildup

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

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

CtrlP

) instead.