Explainer
What is an MMV code, and why does every insurer have a different one?
MMV means make, model and variant. What an insurer's MMV code is, why one car has a different code at every insurer, and what a wrong one costs.
Every motor insurer in India keeps its own list of the vehicles it covers. People in the trade call it the insurer's master. Each row is one make, model and variant (MMV for short) with a code, and that code is what the insurer's quote and policy systems understand. Send the wrong code and you get the wrong premium. Send none and you get no quote.
So far, so simple. The trouble starts when you try to get from a registration certificate to that code.
One car, five codes
Here is a real RC record, spelled exactly the way the registration data spells it:
HYUNDAI MOTOR INDIA LTD / CRETA 1.6 VTVT SX / petrol, 1,591 cc, 5 seats
And here is the listing we pick for it in five insurer masters:
| Insurer | Make | Model | Variant | Code |
|---|---|---|---|---|
| A | Hyundai | Creta | VTVT SX | 12934 |
| B | HYUNDAI | CRETA | 1.6 SX VTVT | 1111810275 |
| C | HYUNDAI | CRETA | 1.6 SX Petrol | 38783 |
| D | HYUNDAI | CRETA 1.6 VTVT SX | (left blank) | 21686 |
| E | Hyundai Motors | CRETA 1.6 VTVT | SX BS IV | 4569 / 457589 |
It's one car. Five insurers write its model and variant five different ways, their codes don't share a format, and insurer C lists what looks like this exact car twice, under two codes. None of these masters is wrong as such. Different teams built them at different times for different systems, and each one makes sense on its own.
The RC side isn't tidy either. The registration record above calls this Creta a station wagon. Two-wheelers are worse. We see the same Hero Super Splendor on real RC records as SUPER SPLENDOR(DRUM-SELF-CAST), SUPER SPLENDOR-DRUM-SELF-CAST and SUPER SPLENDOR(SV)DRUMSELFCAST, and that's before anyone abbreviates anything.
Why close enough isn't good enough
It would be convenient if any Creta listing would do. It won't, because the variant and the engine size decide money.
The insured declared value (IDV), the amount the policy covers, is fixed on the manufacturer's listed selling price of the brand and model, then reduced for the vehicle's age. That wording comes from the India Motor Tariff (General Regulation 8), and insurers still print it in their policy documents. The base and top variants of one model can be a few lakh rupees apart in price, so the wrong variant moves the IDV, and the own damage premium moves with it.
Engine size matters on the third party side. The government sets the third party base premium by engine capacity band (G.S.R. 394(E), May 2022). For private cars the bands are up to 1,000 cc, 1,000 to 1,500 cc, and above 1,500 cc, and two-wheelers have bands of their own. Map a 998 cc hatchback to a 1,197 cc listing and you've crossed one.
What a wrong code costs
Most of the cost is quiet, which is why it's easy to underestimate.
- A quote fails, or comes back at a price the customer didn't expect, and they leave. Nobody files a ticket. You just see a lower conversion rate.
- Someone on your team looks the vehicle up by hand. This is a real job: search any Indian job site for "MMV mapping executive".
- A wrong variant slips through, the policy is issued, and someone has to fix it later with an endorsement. If nobody notices, it can come up at claim time, which is the worst possible moment to find out.
- Insurers change their masters as models launch and variants are dropped. A mapping table built by hand starts going stale the week it's finished.
Why text search isn't enough
The obvious fix is to search the master for the closest string to whatever the RC says. That works on easy cars. It fails where it hurts: the variant, the engine size, and models whose names overlap.
We measured this. On a held-out test of 150 real RC records across five insurer masters, the best standard matcher we tried found the exact listing 66.2% of the time. Dorvy Tech found it 77.0% of the time, and made less than half the wrong picks (about 13%, against 32%). The test set is deliberately hard, and how we measured explains it in full.
What a good answer looks like
A code on its own isn't much use if you can't tell how far to trust it. For every vehicle and every insurer, Dorvy Tech returns the listing (make, model, variant and code), a confidence level of HIGH, MEDIUM or LOW, and the closest alternatives. When nothing in a master fits, it says "no listing" and gives the reason.
That last part matters more than it sounds. A system that always returns something will sometimes return the wrong thing with a straight face, and those are the errors nobody looks at twice.
If you'd like to know how your current mapping does, our post on running a 300-vehicle audit shows how to measure it yourself. Or send us the vehicles and we'll run the comparison for you, free.
