India Promo ROI Data Scraping Case Study — AI Identified Killer Discounts, +18% Contribution

The client is a national India restaurant chain operating 180+ outlets across 24 Indian cities on Zomato and Swiggy, with a marketing team running 12 concurrent promotional programs — everything from platform-featured discounts, first-time-user offers, weekend flat-percentage promos, category-specific bundles, day-of-week specials, festival promotions, and coupon-code campaigns. The marketing director's promo budget had grown 40% over 18 months while contribution margin had drifted downward — a pattern that suggested some meaningful share of the promo spend was funding orders that would have happened anyway, or worse, funding orders that lost money on a per-transaction basis. The problem was that traditional promo ROI measurement was structurally hard in the multi-promo India delivery environment: promos overlapped, incremental attribution was fuzzy, review velocity confounded promo effect with underlying restaurant momentum, and the platforms' own attribution reports were optimized for platform interests rather than chain economics. The chain needed reliable India promo ROI intelligence that used external Zomato and Swiggy signals — competitor promo cadence, discount depth patterns, review-velocity dynamics before and after each promo, ranking evolution — to identify which of the chain's own 12 promotional programs were structurally destroying contribution versus which were actually earning incremental orders worth the discount cost. Names are anonymized for confidentiality; metrics are shown exactly as delivered.

India restaurant promo economics on Zomato and Swiggy have become structurally complex. The chain's marketing team ran 12 concurrent promotional programs with overlapping targeting, overlapping timing, and overlapping customer segments — meaning the same order could be triggered by, or credited to, multiple different promo mechanics. The marketing director's growing suspicion was that some of these 12 promos were structurally destroying contribution — either because the discount depth exceeded the incremental-order lift, or because the promo was cannibalizing orders that would have happened at full price, or because the promo was training customers to wait for discounts rather than buy at margin. But proving it required data the chain did not have: what was the actual incremental-order lift attributable to each specific promo, controlling for baseline demand, competitor promo overlap, and underlying restaurant momentum? A specific example: the chain's flat-25%-off weekend promo across all 180+ outlets had been running for 14 months. Marketing believed it drove meaningful weekend order lift. But post-hoc analysis using external Zomato and Swiggy signals showed that during the same weekends the chain ran the promo, 6 major competitors were running similar promos — meaning the chain's promo was partially matching competitor discounts rather than driving incremental capture. On weekends where competitor promo activity was thin, the chain's promo drove strong incremental lift; on weekends where competitors matched, the chain's promo was essentially subsidizing existing demand at negative contribution. Without the external signal analysis, the marketing team was flying blind on which of the 12 promos to keep, which to modify, and which to kill.

FoodDataScrape built an India promo ROI data scraping pipeline across Zomato and Swiggy covering the chain's 12 promotional programs plus the promo activity of 25 direct competitors — with 18 months of historical promo cadence, discount depth patterns, review-velocity dynamics before and after each promo window, and AI promo-effectiveness scoring. The AI layer combined external competitor-promo signals with the chain's internal order and margin data to produce per-promo incremental-contribution estimates controlling for competitor overlap, baseline demand, and underlying restaurant momentum. The build went live in seven weeks; the first month's analysis identified the 3 structurally negative-ROI promos, the marketing team killed them in the following quarter, and contribution lift of 18% was measured 90 days after the promo portfolio rationalization.

On top of the raw feed, an AI promo-effectiveness layer turned promo-cadence and review-velocity data into India promo ROI intelligence: it identified competitor promo overlap for each of the chain's 12 promotional windows, measured actual incremental-order lift controlling for baseline demand and competitor matching, scored each promo on structural contribution ROI, and produced kill-modify-keep recommendations. The 3 promos flagged as structural negative-ROI were killed by the marketing team over the following quarter. Two of those had been running for over a year and had large budgets — killing them freed marketing spend that was reallocated to the promos flagged as high-incremental-ROI, and total contribution lifted 18% within 90 days of the rationalization.

The pipeline captured a full India promo ROI intelligence view. Every data point below feeds the promo-effectiveness layer — chain-promo cadence reveals timing, competitor-promo overlap reveals matching pressure, discount depth reveals customer-value proposition, review velocity reveals sustained capture versus one-off order, and ranking evolution reveals platform algorithm response: