Day91 Labs Brief — Issue 017: Special edition · Diwali and AI

01. Platforms sold Diwali on AI assistants. None has published how many people used them.

A · read — Amazon's 3 Nov festive wrap-up opened. Meesho and Eternal figures are B, from coverage of their releases and shareholder letter.

FACT. Amazon India's 3 November update on its Great Indian Festival reports more than 393 crore customer visits, shopping volume up over 50% year on year, EMI spending up 40%, and Amazon Now daily orders up 18 times. It gives no usage figure for Rufus, its shopping assistant, which Amazon promoted throughout the sale. Flipkart launched Big Billion Days on 9 October with a conversational AI assistant and search by voice and image; we found no post-sale figure on its use. The one platform that quantified AI was Meesho: personalised recommendations powered about 85% of Day 1 orders, with Day 1 orders up about 75%. On Eternal's 21 October call, its CFO put agent-led commerce adoption in India at close to zero, and the company said agent traffic to Zomato and Blinkit is negligible (B).

READ. "AI drove festive shopping" covers two different things. Recommendation feeds, which Meesho measured, are mature machine learning and clearly move orders. Conversational assistants and agents, which every platform advertised, have no published usage numbers this season, and the operator closest to the customer's doorstep says agents barely show up. If chat assistants had moved the needle, the companies that built them would have said so in their wrap-ups.

SO WHAT. When a prospect, vendor or analyst says AI drove growth, ask which AI: a ranking feed or a chat assistant. Don't size an Indian agentic-commerce opportunity on festive press releases; size it on usage numbers, and none were published.

→ Amazon India, Great Indian Festival 2026 wrap-up (3 Nov) → Meesho Day 1 release, coverage (12 Oct) → Flipkart Big Billion Days 2026 launch, coverage → Eternal Q2 FY27 call and letter, MediaNama (22 Oct)

02. The AI statistic everyone is quoting this Diwali comes from last Diwali.

A · read — Google India's 8 Oct post and its footnotes opened.

FACT. Google's festive post (8 October 2026) says 51% of festive shoppers credit AI with helping them decide, and that shoppers consulted chatbots 7.5 times on average. The footnotes show both figures come from a Google-commissioned Ipsos online survey run September to November 2025, covering last year's season. The 51% is among 2,275 respondents who had used AI; the 7.5 is an average across 2,450 shoppers in seven categories. The same survey found that among shoppers who used AI chatbots, 94% also used Google Search and 87% used YouTube.

READ. The numbers are real, but they describe 2025, they were commissioned by a company that sells AI search and ads, and the 51% is a share of people who already used AI, not of all shoppers. The most telling figure is the 94%: AI is an extra step in the journey, not a replacement for search.

SO WHAT. If you cite a festive AI statistic in a deck or a pitch, give its date, its sponsor and its base. "51% of AI users, 2025, Google-commissioned" is defensible. "Half of Indian shoppers use AI" is not.

→ Google India, two-speed festive shopper (8 Oct 2026)

03. How big was Diwali? A traders' body says record. Urban households said 6% less.

B · reported — CAIT statement and LocalCircles survey, both via press coverage. Neither is audited sales data.

FACT. The Confederation of All India Traders said on 7 November that Diwali-related trade had passed ₹6 lakh crore nationwide and would exceed ₹7 lakh crore, with Delhi alone above ₹1 lakh crore. CAIT's figures are compiled from member trade bodies and are not independently audited. LocalCircles' Mood of the Consumer survey (more than 67,000 urban responses, 307 districts, published mid-October) projected urban festive spending of ₹1.97 lakh crore, down 6.2% from ₹2.10 lakh crore in 2025 (calculated), with 83% of households planning to draw on savings. It cited September retail inflation of 5.54% and noted that 2025's season was lifted by GST cuts that took effect on 22 September 2025.

READ. The two figures measure different things: all festive trade reported by sellers, against intended discretionary spending by urban households. CAIT's number is about 3.6 times LocalCircles' (calculated). Both can be true at once: traders can book record turnover while urban households feel stretched and fund it from savings and EMIs. Neither is a measure of what consumers actually spent.

SO WHAT. Don't put "Diwali spending hit ₹7 lakh crore" in a forecast as if it were a sales figure. If you sell into Indian consumer brands, expect their Q3 commentary to sound strong on volume and cautious on the urban middle.

→ CAIT figures, Hans India / IANS (7 Nov) → LocalCircles survey, Business Today (16 Oct)

04. Online grew far faster than the mood, financed by EMIs.

A · read — Amazon wrap-up opened. Redseer estimates are B, via PTI and trade coverage.

FACT. Amazon reports shopping volume up more than 50%, over 70% of visits from beyond the metros, EMI spending up 40%, and 9 in 10 rupees of EMI spending on no-cost EMI; EMIs made up half of spending on mobiles, TVs and appliances. Premium Android phones above ₹30,000 grew 50%. Redseer estimated ₹24,000 crore of e-commerce GMV in the first two days of the sales, about six times a normal day, with mobiles 38–40% of value and first-time shoppers doubling. Before the season it forecast festive e-commerce GMV of $15–16 billion, up about 25%.

READ. The caution in the household survey and the records on the platforms describe different people and different payment methods. Interest-free instalments let buyers upgrade without feeling the cost now, which pulls spending forward and moves it into the next few quarters' household budgets.

SO WHAT. For anyone selling to Indian consumer finance, lenders or retailers: the festive story is credit at the checkout, not just discounts. Expect EMI-funded demand to show up as softer discretionary spending in early 2027.

→ Amazon India, Great Indian Festival 2026 wrap-up (3 Nov) → Redseer first-two-days estimate, Business Standard (11 Oct) → Redseer festive forecast, BestMediaInfo

05. The company that owns the delivery van says AI agents can't take its customers.

B · reported — Eternal's Q2 FY27 shareholder letter and call, via MediaNama. Results figures via coverage of the filing.

FACT. Eternal (Zomato, Blinkit) reported Q2 FY27 on 21 October. Blinkit's net order value rose 79.3% to ₹20,940 crore across 2,751 stores, and Eternal's cash fell ₹1,813 crore in the quarter to ₹16,475 crore, mainly because Blinkit stocked up ahead of the festive season. In the letter, Deepinder Goyal wrote that "an agent can place an order, but it cannot own the outcome", placing Zomato and Blinkit in a low-risk zone for agentic commerce because they run fulfilment themselves. He put District, its ticketing and table-booking business, in a medium-to-high risk zone, since agents can book the same seats through other channels; the board moved District into a subsidiary the same day. Redseer forecast quick commerce's share of festive e-commerce rising from about 11% to 18% this year (B).

READ. This is the clearest public map yet of which businesses an Indian operator thinks AI agents can disintermediate: those that only list someone else's supply. Businesses that hold stock and deliver in minutes see agents as just another way to receive an order. The fastest-growing festive channel is also the one least exposed to AI shopping agents.

SO WHAT. Use Eternal's test on your own accounts and competitors: does the company own the outcome, or only the listing? Listing-only businesses (ticketing, travel, comparison) are the ones that need to watch what AI assistants say about their prices and availability.

→ Eternal Q2 FY27 shareholder letter (PDF) → MediaNama, five takeaways from Eternal's Q2 FY27 call → Google–Redseer quick-commerce festive forecast, BestMediaInfo

06. A 19-day calendar shift is distorting every Indian comparison this quarter.

B · reported — FADA's October release, via its published text and coverage. Titan management commentary via coverage of its results.

FACT. FADA reported 31,72,466 vehicle registrations in October 2026, down 23.80% year on year but up 25.05% on September; passenger vehicles were down 14.23%. FADA's president called the annual fall "almost entirely optical" because Dhanteras and Diwali fell in November this year and October last year. Over the 10-day Navratri window (11–20 October), registrations were 11.23 lakh against 12.01 lakh, down 6.5%, while EV share rose to about 10.5% from 7.7%. Titan said demand shifted from its July–September quarter into October–December, and its jewellery business still grew about 22–24% in Q2 FY27.

READ. Year-on-year numbers for October, and for the July–September quarter, are flattered or punished by the calendar, not by demand. Last year's season also had a one-off GST boost. The honest comparison is the whole festive window across two years, which most companies won't report.

SO WHAT. If you track Indian customers or competitors, compare September to November combined, not October alone. Expect "record Q3" headlines in January that are partly timing, and discount them before you change a forecast or a territory plan.

→ FADA October 2026 retail data (release text) → FADA Navratri window, PSU Watch (21 Oct) → Titan Q2 FY27, Business Standard (6 Nov)

07. UPI set a record in October, and merchant fees were pushed past Diwali.

B · reported — NPCI's monthly statistics, via coverage. NPCI's own page was not opened.

FACT. UPI processed 25.47 billion transactions worth ₹31.53 lakh crore in October 2026, up 23% in volume and about 16% in value year on year, and up 5.8% in volume on September. Average daily value passed ₹1 lakh crore for the first time. Merchant payments were 63.2% of transactions, but person-to-person transfers carried most of the value. NPCI had planned to introduce a merchant discount rate on some UPI payments from 15 October, reported as 0.4% on payments above ₹2,000; on 13 October it deferred this until after Diwali, with a new date to be announced.

READ. Value growing more slowly than volume means the average payment is getting smaller, consistent with UPI spreading into everyday small purchases. The fee deferral protected the festive peak, but it means merchants will meet a new cost on larger payments just as the season ends.

SO WHAT. If you sell checkout, payments or pricing tools to Indian merchants, the period right after Diwali is when they will be re-pricing payment options and steering customers between UPI, cards and EMI. Have a cost comparison ready before NPCI names the date.

→ UPI October 2026, Entrackr → UPI P2M and P2P split, MediaNama → MDR deferral, Entrepreneur India

08. Businesses shopped Diwali too, and marketing budgets followed the platforms.

A · read — Amazon Business figure from Amazon's wrap-up. Ad-spend figures are B: pre-season industry forecasts, not actuals.

FACT. Amazon says Amazon Business sales rose 40% year on year during the festive season and new business sign-ups grew nearly 20%. Before the season, Datum Intelligence forecast festive advertising of ₹60,000–62,000 crore, up 13–17%, with digital at 67.8% of it. Retail media and e-commerce advertising were forecast to take about ₹11,500 crore, up 27%. Google's festive post pitched its AI Max search campaigns to the same advertisers.

READ. Diwali is a procurement season for Indian companies (gifting, festive stock, office supplies), and that buying is moving online. On the marketing side, the fastest-growing slice of festive budgets is retail media, where the platform owns the shopper data, the ad and, increasingly, the AI that ranks the products.

SO WHAT. If you sell marketing or revenue tools into India, your buyers' festive money is flowing into platforms they can't measure independently. Lead with cross-platform measurement and a verified record of what each platform showed shoppers, not another channel to spend in.

→ Amazon India, Great Indian Festival 2026 wrap-up (3 Nov) → Datum festive ad-spend forecast, Exchange4media → Retail media forecast, coverage

The running record

• `+` Diwali date — 8 Nov 2026 (Dhanteras 6 Nov) · 20 Oct in 2025
• `+` CAIT — festive trade >₹6 lakh crore, projected >₹7 lakh crore · Delhi >₹1 lakh crore · trade-body estimate, unaudited
• `+` LocalCircles — urban festive spend ₹1.97 lakh crore, −6.2% (calc) · 83% using savings · 67,000+ responses
• `+` Amazon GIF 2026 — 393 crore+ visits · volume +50% · EMI spend +40% · Amazon Now daily orders 18x · Amazon Business sales +40% · no Rufus usage figure published
• `+` Redseer — first two days ~₹24,000 crore GMV, ~6x a normal day · season forecast $15–16bn (+~25%) · quick commerce share 11% → 18% (forecast)
• `+` Meesho — Day 1 orders +~75% · personalised recommendations powered ~85% of Day 1 orders
• `+` Eternal — Blinkit NOV ₹20,940 crore (+79.3%) · cash −₹1,813 crore on festive inventory · Zomato and Blinkit rated low risk from AI agents, District medium to high
• `+` Google / Ipsos (2025 survey) — 51% of AI-using festive shoppers say AI helped decide · 7.5 chatbot consultations · 94% of chatbot users also used Search
• `+` FADA October — 31.72 lakh registrations, −23.80% YoY, +25.05% MoM · Navratri window −6.5% · EV share ~10.5%
• `+` UPI October — 25.47bn transactions · ₹31.53 lakh crore · +23% volume, +16% value YoY · MDR deferred until after Diwali
• `+` Festive ad spend (forecast) — ₹60,000–62,000 crore (+13–17%) · digital 67.8% · retail media ~₹11,500 crore (+27%)
• `-` Not carried — Rufus "leaked" sale dates (the assistant's own guesses, not Amazon's) · an unsourced claim that AI lifted festive transaction values 46% · Meta-attributed "80% used AI for festive ideas" (second-hand) · Salesforce's global holiday AI forecasts (not India) · CAIT category shares (aggregator only) · Amazon's sale end date (its release says 10 Nov in one place, coverage says 11 Nov)
• `-` Not yet available — Flipkart's post-sale figures · Walmart's quarter covering Big Billion Days · Redseer's full-season actuals · FADA's November data, which will show whether the shift washes out