Appearance
Product scope: baseline, differentiators, modules, and what to refuse
⚠ SUPERSEDED 2026-08-22 by product decisions
This page was written against a generic dealership operating layer and it reasoned about the market rather than about QRSETU's own product. Product decisions replaces it and is organised around the Setu Card as the foundation, which is the actual product thesis.
Kept, not deleted, for two things it still holds that the newer page does not repeat: the copyability-and-stickiness assessment of each candidate differentiator (§2), and the AI sequencing (§6). Its refusal list (§5) is correct and is carried forward in full.
⚠ Where the two disagree, product decisions wins. Two specific disagreements. Pricing: this page's per-site tiering is superseded by the annual per-outlet package book in commercial model - which returned to the physical site as the unit on 2026-08-23 after a per-card interlude, renamed it from the undefined jargon rooftop to outlet (why), and raised the floor from Rs 30,000 to Rs 75,000/yr (QRS-844). Sequencing: this page ranks service annuity “Build first”, which is no longer the build order — the question was reopened and settled on behaviour change required rather than margin, in value proposition §6 and go to market §3.
Part of
car_sales— the dealership operating layer. Evidence-class legend there.
1 · Baseline — expected, not differentiating
🔎 Table stakes. Ship them competently, spend no strategy on them, and never let a demo turn into a feature comparison against a DMS.
| Capability | Note |
|---|---|
| Lead inbox | Source attribution, assignment, aging, duplicate detection across channels |
| Follow-up engine | Scheduled tasks, escalation, reassignment when a rep is on leave — a named pain |
| Customer registry | One record per person, phone-primary. 📘 This is the party primitive, not a bespoke contacts table — see QRS-823 |
| Mobile-first field app | Non-negotiable. The named gap in every OEM DMS critique |
| Dashboards | Outlet, team and individual rollups |
| WhatsApp send and receive | Transport, not a marketing product |
2 · The differentiators, assessed
| Capability | The prize | Who does it in India | Copyable | Sticky | Verdict |
|---|---|---|---|---|---|
| D1 · Service annuity engine vehicle asset + odometer/time recurrence + WhatsApp | 🌐 20:1 serviced-to-sold at 41.2% gross margin | Nobody, dealer-side | Medium | High — owns the vehicle history | Build first |
| D2 · Cross-brand group consolidation | The group's whole P&L view | 🌐 Nobody. Excellon's DMS page never mentions it | Low — needs the workspace tree | Very high | The moat |
| D3 · Discount governance | 🌐 3.02x PAT | Nobody | Medium | High | Build 4th, and read the CCI caution |
| D4 · F&I attach management workflow only, never the commission | 🌐 2.76x the ad budget | Nobody dealer-side | Medium | High | Build 3rd |
| D5 · Missed-enquiry capture | 🌐 A US vendor measured 158 missed service calls per dealership per month | Nobody | High | Medium | Fast follower |
| D6 · Review / GBP funnel | 🌐 Podium: $100M revenue by 2019, $3B valuation; ❓ ~$399-599/location/mo | Generic tools only | High | Medium | Bundle, do not lead — but ⚠ the TRIGGER is not copyable: ours fires on an attributed interaction (teardown) |
| D7 · Targets and incentive transparency | 🌐 29.53% frontline attrition | Nobody | High | Medium | Cheap add. 📘 ADR-0024 already scoped targets |
Design constraint on D3 that is legal, not cosmetic
🔎 Discount governance must be dealer-owned and invisible to the OEM by design. 🌐 The CCI penalised Maruti Suzuki over a Discount Control Policy enforced by mystery shoppers, with dealer penalties escalating to ₹2,00,000 and threats to stop model supply; the matter is on appeal at NCLAT.
A tool helping a dealer principal govern his own team's discounting is squarely on the right side of that order. A tool exposing dealer discount data to a manufacturer is on the wrong side of it.
Never build an OEM-facing view of discount or margin data, and say so in the sales conversation: "your numbers never go to the manufacturer" is both a legal safeguard and the strongest trust position available against a competitor whose customer is the manufacturer.
3 · What makes it hard to leave
🔎 Stickiness here comes from three things, in ascending order of strength:
- Vehicle service history per asset. Two years in, the vehicle records are the dealership's memory of its own customers. Leaving means losing them.
- The WhatsApp relationship. If the dealer's own WABA, template library and opt-in ledger live in the platform, migration is a re-consent exercise rather than an export.
- Consolidated cross-brand reporting. Once a dealer principal runs his Monday review off one screen covering six brands, no single-brand DMS replaces it. This is the only one of the three that cannot be copied cheaply, because it needs the tenancy tree.
4 · Module architecture, in three commercial tiers
🧮 The grey substrate is done and tested. The missing block is the real cost of this plan — see the index for the measured detail.
5 · What to explicitly avoid
🔎 Refusing well is worth more than any feature here. Each row carries its reason, so no stays cheap when it is proposed again in six months.
| Do not build | Reason |
|---|---|
| DMS replacement — enquiry-to-delivery, invoicing, allotment, warranty, parts | OEM-mandated. Competing with a system the dealer cannot switch off, against a ₹74 Cr bootstrapped incumbent |
| Vahan market-intelligence dashboards | 🌐 Dead as of 15 Aug 2026. The notice is verbatim: "Vahan4Dashboard will be discontinued after 15 August 2026", and the 2019 Bulk Data Sharing Policy was scrapped in June 2020 on privacy grounds. The Aug 2025 National Transport Repository policy names insurers, law enforcement and intelligence as users, not commercial analytics firms, and publishes no fee schedule. ❓ The replacement's exact public scope is unconfirmed |
| Insurance or finance commission | 🌐 Licence-gated. An MISP is appointed by an insurer and may not solicit motor insurance from anyone who did not buy the vehicle from them; OEM-captive brokers already hold that pool and have been fined for MISP violations. RBI's Digital Lending Directions 2025 add India-resident storage, CIMS registration and an unbiased all-offers view. Sell the attach workflow; never touch the commission |
| Agency retainers, creative, 360° imaging, lead-gen resale | Labour businesses that do not scale past two people, and 🌐 the marketplace lead products (CarDekho, CarWale, Cars24, Droom, Spinny) do not publish prices at all |
| Recruitment and training delivery | Real pain at 29.53% attrition, and entirely a services business. Refer it out |
| Accounting, GST, parts inventory | Tally and Marg territory, and the OEM owns parts |
| Owning telephony | 🌐 Integrate instead: Exotel, MyOperator, Acefone, CallHippo at ₹1,300-1,999/user/month or ₹5,000-15,000/month for ten users |
| Any in-app purchase CTA on native | 📘 ADR-0002 / Apple 3.1.3(d). A store-review risk, not a preference |
| Advertising or profiled placement | Needs a consumer base that will not exist, and brings DPDP profiling duties. 📘 Same conclusion as strategy change C7 |
| A second card-product abstraction | 📘 The standing naming rule already forbids a shared contract across card products that do not exist |
The used-car segment looks open and is not, for a reason worth recording
🌐 Independent used-car dealers are reachable and numerous, and have no OEM system at all. But Cars24, CarTrade (AutoBiz, Lead2Retail) and Mahindra First Choice give dealer tooling away as customer-acquisition cost for transaction revenue. 🔎 You cannot out-price free. Do not lead with used-car despite its apparent openness — and note that the "used-car dealers have fatter margins so they can pay more" argument rests on a margin conflation a fact-checker struck (operational-gaps).
6 · Where AI genuinely helps, later
🔎 Sequenced behind a paying customer, because each needs data the product does not yet collect.
| Opportunity | Why it works here | When |
|---|---|---|
| Call-transcript QA in Indic languages | 🌐 Tele-calling is done by 262 people against 11 IT staff at one group. Scoring calls is the highest-leverage automation in the building | After Wave 2 |
| Service-due propensity | Ranks which of thousands of due vehicles to contact first, directly on the 41.2%-margin line | After Wave 1 has 12 months of data |
| Discount-anomaly detection | Flags per-unit give-away outliers by executive. The largest rupee prize, and pattern detection is the right tool | Wave 3 |
| Lost-lead reactivation scoring | Aged leads are starved by fresh-enquiry bias, which vendors describe in their own customers | After Wave 2 |
| Used-car photo processing | 🌐 Spyne built a company on it. ⚠ Which also means the category is taken | Only if used-car becomes the focus |
| AI replies inside WhatsApp | ⚠ 🌐 Meta's Business-Agent interactions bill at roughly ₹3.5-4.5 each. ✅ CORRECTED 2026-08-23: that is the cost of buying META's AI, not of AI. 🧮 Our own Claude Haiku inference is ₹1.24 for a FIVE-TURN conversation — so running our own model and sending through the ordinary messaging path is 14-18x cheaper per interaction | Add-on, from Wave 2 — see AI strategy |
Related
- Operational gaps — the pains these capabilities answer
- Commercial model — which of these is in which SKU
- Go to market — the build waves and their sequencing