Pawan Mishra

Dubai's q-commerce market is fiercely competitive — Talabat Mart, Noon Minutes, and Careem NOW all promise 20-minute grocery delivery. But the real battleground isn't average delivery time. It's peak-hour reliability.

During high-demand windows (7–9 PM weekdays, Friday evenings), on-time delivery rates drop significantly across players. Customers who experience a failed or delayed promise churn at 3–4x the rate of satisfied customers. The product failure isn't at the rider level — it starts upstream, at the dark store.

• Demand spikes during peak hours overload picker queues at dark stores — orders sit unassigned for 4–6 minutes before a picker starts
• Static inventory placement means high-velocity items (dairy, beverages) require more travel time per pick during peak hours
• Slot availability shown to customers doesn't account for real-time dark store capacity — leading to overpromising and underdelivering
• Rider supply doesn't match demand spikes — incentive triggers are static, not dynamic

Data signals I'd use: order-to-dispatch timestamps by hour, picker utilisation rate, per-SKU pick frequency, rider idle time vs. demand queue depth

• Dynamic slot throttling: Integrate real-time dark store capacity into slot availability shown to customers. If picker queue depth exceeds threshold → show 25-minute slot instead of 20 minutes. Honest > fast. Reduces post-order NPS damage from missed promises.
• Predictive inventory placement: Use order history by hour and day to pre-position high-velocity SKUs at the front of picking zones during predicted peaks. Reduces average pick time per order by ~30 seconds — multiplied across 200+ orders/hour = significant aggregate impact.
• Real-time rider incentive triggers: Build a dynamic incentive layer that fires surge bonuses to idle riders when demand queue exceeds supply threshold — not on a fixed schedule. Reduces the supply gap at peak without over-spending on incentives during off-peak hours.

• Dynamic throttling may reduce order volume — showing longer slots could deter some orders. Accepted trade-off: fewer overpromised orders > more failed deliveries.
• Predictive placement requires ops buy-in — dark store teams must re-rack SKUs. Mitigated by starting with 2 pilot stores before scaling.
• Incentive budget risk — dynamic surge bonuses must have a cap to prevent runaway costs. Maximum incentive pool per hour defined in the model.

Meesho is India's largest social commerce marketplace with 100M+ monthly active users. I joined as Associate PM for Vendor Monetization — responsible for growing the platform's revenue from its multi-million seller base without compromising the seller experience that drove supply growth.

• Vendors were finding workarounds to avoid the platform's take-rate — revenue leakage was silently suppressing net monetization
• Vendor drop-off during the onboarding funnel was high — new sellers weren't activating, which reduced supply quality
• The existing pricing tier structure was opaque — sellers didn't understand what they were paying for or what they got in return, leading to resentment and churn

• Diagnosed the leakage: Used SQL funnel analysis to map exactly where revenue was being lost — identified 3 specific seller behaviour patterns causing systematic underpayment
• Rebuilt the pricing tier logic: Launched automated pricing tiers based on seller segment analysis — making tier assignment dynamic and harder to game, while making the value exchange explicit to sellers
• Redesigned the onboarding experience: Ran usability testing with 20+ vendors, identified the 3 highest-friction points, and rebuilt those flows — reducing drop-off by 20%
• Changed the framing: The key insight — sellers weren't resisting the price, they were resisting feeling extracted from. Changing the communication from "here's what you pay" to "here's what you get at this tier" improved acceptance without changing the price

Monetization features fail when users feel extracted from, not served. The product fix and the communication fix are equally important. This insight applies directly to subscription products like Talabat Pro and Careem Plus — the value exchange must feel fair before users will pay.

Licious is India's first D2C meat and seafood unicorn ($1B+ valuation), operating a fully integrated supply chain from farm to doorstep. As PM for Last Mile, I owned the pricing and post-order experience roadmap for a high-volume daily-delivery platform.

• Users were abandoning at the delivery slot selection screen at disproportionately high rates — a bottleneck that wasn't being tracked as a distinct funnel step
• Pricing was uniform across user segments and time slots — not reflecting real demand, supply cost, or user willingness to pay
• Internal operations were heavily manual — cross-team workflows between delivery ops and product had no automation layer

• Identified the real drop-off point: Instrumented the slot selection screen as a distinct funnel step using Mixpanel — found that 31% of users who reached this screen didn't convert. This was the highest-drop funnel stage.
• Built dynamic pricing models: Segmented users by order frequency, location, and time-of-order into 4 distinct pricing cohorts. Ran A/B tests on price + slot presentation for each segment. Found that heavy users were willing to pay a premium for early morning slots; first-time users needed free delivery to convert.
• Redesigned slot UX: Changed slot presentation to lead with availability and convenience signals before showing price. Small UX change, significant conversion impact.
• Automated internal ops tooling: Built tooling that automated the daily delivery capacity planning workflow, reducing manual processing by 25%.

Conversion problems are rarely about the price. They're about whether the user believes the value exceeds the cost at the moment of decision. Fixing the UX context around the price moved the needle more than any price reduction we tested. This framework directly informs how I'd approach delivery fee and slot pricing for any MENA q-commerce platform.