Skip to content
Platform filter usage has reached prior baseline at ~5% with restoration of legacy HVF models on 02/08
We observe a net ~2.74% jump in O/V offset by a ~2.83% drop in V/Vi leading to -0.17% negative impact on platform orders / visitor. Net T/Vi negative at -0.14% (Pre period was homogenous)Gold - ~25% jump in V/Vi offset by ~15% drop in conversion leading to ~6.13% improvement in platform order / visitor. Platform T/Vi better for test by ~5.43%PDP reco - While O/V has improved for test by ~9%, It’s been offset by a sharp V/Vi drop of ~42% leading to a net negative impact on O/Vi for the real estateUsage / CTR - Click through rates for non FTUX variant stands at ~5%, while FTUX variant stands at ~30%
Interaction profile - ~55% users are swiping to change the tabs while ~45% users are tapping on tabsReason for T/Vi drop - At a platform level, we see V/Vi dropping massively leading to O/Vi and T/Vi dropV/Vi at RE level signifies the major drop is stemming from engagement going down for PDP reco 
Super port level trends - Pending next steps - Map control’s V/Vi to ascertain drop is coming from the parallel feeds tab or native recoBuild the view for affected super portfolios at a SSCAT level to understand the exact cases which are worst affectedTo check - Drop is coming from people who switched the tabCheck V/Vi for normal reco between test and controlDoD trend - For plat metrics as well as PF level V/ViBounce rates - PLP → PDP click through rate between test and controlP1 - Price bucket wise trendsDetails at a SSCAT level for affected super portfoliosGender level cut (User level trends per gender) Design-tech handover for this piece is done and we have carefully de-scoped certain pieces to try and target mid Aug app release for sufficient experimentation and scale up duration pre code freezeIn tech solutioning, delivery timelines to be shared by early next weekMVP scope - v1 would only solve for integration of content recos to be showcased in bottom sheet Bottom sheet to be triggered via the following variants - 
Build would go out across the following REs - Wishlist / CLP / Collections / FY and Search FTUX version will be built in for one of the above 2 variants owing to low timelines for MVP app release (Targeted for 15th Aug)A/B configuration would allow choosing between variants and enabling FTUX. Real estate level configuration is not supported in MVP due to high timelinesFallback feed for cases where PDP recos don’t exist to be finalised between ranking and BEx devs by todayHVF - HVF v2 was launched on Aug 7th across REs, We are testing the following variants - IF - Model build for IF was picked on 08/08. Work in slow burner due to model prioritisation for hero product restorationModel setup TAT - 16th Aug, Experiment start TAT - 21st AugHero product model restoration - Current hero product logic works - At catalog level, we give priority to product with most number of transactors basis lifetimeIf there's a tie between products, then we take last uploaded pid in the catalogThis logic changes for sscats from number of transactors to first first uploaded pid in Education+Healthcare part of Digital SuperPorts which has negligible orders.We deboost PIDs which have been invalid in the past two hours, while ranking.Next steps alignment - Given A/B would need significant dev effort across taxonomy and feed aggregator, we can approach this in the following way - SSCAT / Portfolio level A/B - Update for certain categories and observe data against other similar categoriesStraight rollout across SSCATsLogic updates - Factoring in order recency and moving away from unique transactors to net orders - Building the model based on last 1/3 months orders(for recency) + adding 1 year orders fallback for catalog with no recent ordersAnalytics check - Team will close checking on the OC which will get impacted at a super portfolio level with the update, To be closed 09/08Success metric (TBC) - From a pre-post lens, team’s proposal is to compare conversion metric of the new model ( Pre vs Post) on RE X super portfolio level [ ObyV, CbyV , ObyC]Updates - Do a product level A/B (Check with Kunal / Rishita)DS filter label rationalisation - Post deployment, devs found an issue with new DS label values not being properly tagged to catalogs due to increase in config size and heavy compute. Dev is on-going here and In parallel, we’ve initiated QC of the new labels from Biz. Update - While dev would continue, deployment would be planned post MBS sale due to ramifications of the flow increasing latency of core services Roadmap / CPD - Given decent user comprehension of parallel feed MVP, team will start the conjoined product + design solutioning to -Design - Extend the build across REsProduct - Identify high conversion intent levers to be surfaced per parallel feed (Category etc)TBP - Last week of AugIC via search - Given search’s high conversion, We will pickup solutioning for integration of relevant high converting search queries as part of a feedPre requisite query classification to enable this has already been developed as part of IFsTBP - Mid SepUX for null filters showing on feedSelection outline of filters can be better
09/08/24
Metrics
Key discussion points -
Instrumentation alignment - We would be recording parallel_feed as part of primary real estate in base views and clicks events. Decision to be aligned cross org to align stakeholders on funnel attribution changes. TAT - 09/08
Resolution build for FTUX variant pushed in Aug 01st app release. Re launch planned for 13/08
IG on Wishlist
Variant
Dynamic hero enabled
Model type
Control_1
No
Old HVF
Control_2
No
No HVF
Test_1
Yes, Across REs
Meritocratic build live for all REs
Test_2
No
Meritocratic build live for all REs
Test_3
Yes, Across REs
Label wise slotting for all REs
Test_4
Yes, on CLP
Meritocratic build live for CLPs, Other REs on old HVFs
Test_5
Yes, on Search
Meritocratic build live for search, Other REs on old HVFs
Test_5
Yes, on Collections
Meritocratic build live for collections, Other REs on old HVFs
There are no rows in this table
Approach
Want to print your doc?
This is not the way.
This is not the way.
Try clicking the ··· in the right corner or using a keyboard shortcut (
CtrlP
) instead.