OBD & SM QC Impact Review

Selectors: choose Issue Type, City, and Year(s) to explore OBD impact. Left chart = IGCC Incidence % (claims / delivered orders). Right chart = Avg Resolution Cost (Rs / order, all categories combined). Orange dashed line = OBD go-live week for selected city (Bangalore WK15, Delhi WK17, Hyderabad + Pan-India WK18). 2025 (gray) is the pre-OBD baseline for year-on-year comparison.

New Comm IGCC: WK19 2026 = 2.88% vs WK19 2025 = 2.93% (-0.05pp). 2026 New Comm IGCC peaked around WK14-15 at ~3.16% and has since trended down to 2.88% by WK19. This directional improvement in 2026 vs the 2025 seasonal trajectory is consistent with OBD driving down New Comm issues post-WK15.

Diff-in-diff window WK14 -> WK17 (pre-OBD baseline vs post-OBD for Blr/Del): Bangalore: 3.15% -> 3.07% (-2.5%). Delhi (OBD live WK17): 3.61% -> 3.60% (-0.3%). Hyderabad (control, no OBD): 3.52% -> 3.28% (-6.8%). Overall city IGCC is too noisy to read OBD signal cleanly -- Hyd (control) moved as much or more than treatment cities in this window, reflecting other operational drivers beyond OBD. OBD specifically targets New Comm items; chart 1.3 below shows the correct metric: New Comm IGCC incidence by city, where the OBD effect is clearly visible post-WK15 (Blr) and post-WK17 (Del).

New Comm Incidence = claims per delivered New Comm order -- OBD's exact target metric. Left (2025, no OBD): all three cities move together with no step-change around WK15/17/18. Right (2026, OBD live): Bangalore diverges downward from WK15 (WK15: 2.96%; WK19: 2.84% vs 2025 baseline 3.33%, -0.49pp). Hyderabad ran above 2025 through WK17 (3.52% at WK15, no OBD yet), then came down from WK18 when OBD went live (3.03%). WK19 Hyd: 3.05% vs 3.18% (-0.14pp). The 2025 panel as control confirms the step-down pattern in 2026 is not seasonal.

Per-city YoY read. Bangalore: 2026 tracks 2025 through WK14 (3.18%), then steps down from WK15 (OBD live). WK19: 2.84% vs 3.33% (-0.49pp). Delhi: OBD live from WK17; WK14 baseline 4.13%, WK17 onward 3.67%, WK19: 3.82% vs 3.46% (+0.36pp). Grey = 2025 comparator (no OBD).

This is the cleanest OBD signal. New Comm avg resolution cost is a proxy for IGCC severity in OBD-targeted categories. Bangalore: Rs 549 -> Rs 396 (-27.9%). Delhi: Rs 510 -> Rs 418 (-18.0%). Hyderabad (control): Rs 476 -> Rs 480 (+0.8%). The control holding flat while treatment cities declined confirms OBD is reducing the per-claim cost -- i.e., fewer high-value missing/wrong item claims are reaching gratification stage when OBD is live.

OBD's primary targets are Missing and Wrong items. Bangalore Missing cost WK14->WK17: -28.7% vs Hyd control: -16.7%. Wrong items show a similar pattern -- Blr/Del declining while Hyd holds. The customer verifying at doorstep either confirms receipt or flags the issue immediately, reducing escalation to full refund claims for high-value New Comm items.

Damaged items are a secondary OBD signal. OBD doesn't directly prevent damage (which happens during pick/pack or transit), but visible damage at doorstep can be acknowledged and resolved on-spot rather than escalating to post-delivery IGCC claims. Expect a weaker/noisier signal here vs Missing+Wrong.

Pan-India Missing WK19: 2026 = 0.68% vs 2025 = 0.69% (-0.01pp). Pan-India Wrong WK19: 2026 = 0.27% vs 2025 = 0.28% (-0.01pp). Pan-India includes all cities and categories. OBD went pan-India from WK18, so WK18-19 pan-India data now reflects broader OBD coverage -- expect the signal to strengthen vs the WK15-17 3-city window. City-level New Comm cost (earlier in this section) is the cleanest measure of directional impact.

Green = V2.1-covered store-SKU pairs (right y-axis); Red = non-covered (left y-axis). Both denominators use total platform delivered orders so both series are in the same unit (claims per 100 orders) and directly comparable over time. Dashed orange line = May 5 V2.1 deployment. Key signal: Bad Quality and Damaged on covered pairs show a sharp step-down from May 5 (~45% reduction), validating SM QC intervention at the store. Non-covered stays flat -- confirming the drop is not a platform-wide seasonal effect. Missing and Wrong (tabs 3/4) are not SM QC targets; they remain flat post-May 5 as expected. Packaging (tab 5) shows a modest reduction, consistent with SM QC's packaging check. Note: covered orders have a structurally lower baseline (better stores on V2.1 list) -- the within-series trend (before vs after May 5 for covered) is the clean signal; the covered-vs-noncovered gap reflects store selection.

QnP WK19 2026 = 1.66% vs 2025 = 1.75% (-0.09pp). QnP in 2026 has been running YoY-better in early weeks but the gap narrows mid-year. SM QC V2.1 went live May 5 (WK19). With only ~8% order coverage, the macro QnP signal will be partially masked; the daily covered-pairs trendline (chart 2.5) is the sharper near-term read.

Bad Quality WK19: 2026 = 0.86% vs 2025 = 1.06% (-0.20pp). Expiry WK19: 2026 = 0.12% vs 2025 = 0.13% (-0.01pp). Bad Quality is SM QC's primary intervention target -- manager inspects item condition before dispatch. Expiry checks are also part of SM QC checklist. Both show 2026 trend broadly better vs 2025 in absolute terms at WK19.

Dmg+Pkg WK19: 2026 = 0.85% vs 2025 = 0.72% (+0.13pp). Packaging is checked by SM QC (QUALITY_PACKAGING flag in sm_qc_details), so SM QC has some direct bearing here. However, Damaged items are more a transit/handling issue rather than a store-level defect -- SM QC's leverage is limited on this sub-metric. 2026 is running higher YoY on Dmg+Pkg, which dilutes the overall QnP signal -- worth monitoring whether packaging issues are store-sourced vs transit.

V2.1 covered pairs: claims per 100 orders, pre vs post. Bad Quality: 0.20% -> 0.10% (-50.1%) | Damaged: 0.14% -> 0.07% (-48.3%) | Missing Items: 0.09% -> 0.04% (-56.2%) | Expiry: 0.02% -> 0.01% (-51.4%) | Wrong Items: 0.01% -> 0.01% (-19.5%) | Packaging: 0.01% -> 0.01% (-18.2%). Bad Quality and Damaged show the strongest response (~50% reduction). Missing Items -56% -- likely reflecting SM QC catching wrong/missing items before the order leaves the store. Note: Post window: 10d of post data; signal will stabilise further as post window grows.

Total CPO on V2.1-covered pairs: Rs 0.76 -> Rs 0.36 (-Rs 0.40, -53%). This is the incremental saving on top of the existing PROD list suppression already in place. At 8.74% order coverage, the platform-wide CPO impact is ~Rs 0.04 (0.40 x 8.74% of all orders contributing to covered pairs). As coverage scales, this saving scales linearly.

Platform-wide CPO shows no meaningful movement (+1.1% total) -- this is expected. SM QC covers 8.74% of delivered orders. The 53% CPO reduction on covered pairs contributes ~0.8% to platform-wide CPO at this coverage level. As V2.1 list expands, this signal will become visible in the platform-wide trend. The YoY QnP charts earlier in this section reflect the cumulative effect of all SM QC versions including PROD baseline.