When you call a traditional insurer, the person on the other end of the line is not actually looking at you. They are looking at a category. Male, age 28–34, Lisbon postal code, hatchback under three years old. Your actual driving behaviour, your real annual kilometres, the fact that you commute by metro four days a week and only drive on weekends — none of that enters the calculation. You get the bracket price, which is averaged across everyone in your bracket, including the people who are worse drivers than you.
This is not because traditional insurers are lazy. It is because gathering individual data and pricing it individually was expensive and slow before the kind of computing infrastructure we have today. The bracket system was a reasonable approximation. The problem is that reasonable approximations are systematically unfair to the people who sit below the bracket average, and systematically generous to those above it. If you are a careful driver, you have been subsidising everyone else.
What changes when pricing happens in real time
At Indie, we built our pricing engine around a different question: what do we actually know about this specific person and vehicle, and what does that tell us about expected claims cost? Instead of looking up a rate table, we run a model. The model was trained on anonymised claims data — years of actual Portuguese auto insurance outcomes, covering the kinds of roads people drive on, the frequency of accidents by vehicle type, the correlation between annual kilometres and claims frequency, and several dozen other features.
When you start a quote on our platform, six questions appear. Vehicle registration number, annual kilometres, primary use (commuting, occasional, business), overnight parking (private garage, street, private car park), whether you have held a Portuguese driving licence for more than two years, and whether you have made any claims in the past three years. Those six inputs feed directly into the model. The model does not look up your demographic bracket. It estimates your expected claims cost as an individual, then adds a margin for operational costs and reserves. The number you see reflects you, not a generalisation about people vaguely like you.
Why individual scoring tends to produce lower prices for most drivers
This is the counterintuitive part that takes some explaining. If you are in a bracket and you are an average driver, the bracket price is fair. But if you are below average risk — which describes most careful, low-mileage drivers — the bracket price is too high. You are being charged for risk you do not carry. When pricing becomes individual, that overcharge disappears.
In practice, we see that drivers who park in a private garage or private car park, drive under 10,000 km per year, and have a clean claims record for the past three years typically receive quotes 15 to 25 percent lower than equivalent market rates. That gap represents the cross-subsidy they were previously paying.
We are not saying bracket pricing is fraudulent or malicious. It served the industry well for decades and it still produces adequate prices for people who genuinely sit at the average. What we are saying is that it is not precise, and imprecision has a cost that falls on safe drivers.
The role of the ASF framework
Every insurer operating in Portugal, including our insurance partners through whom Indie policies are issued, operates under supervision by the Autoridade de Supervisão de Seguros e Fundos de Pensões (ASF). The ASF sets rules on how insurers must calculate and maintain technical reserves — the money set aside to pay future claims. These rules exist to protect policyholders and they are non-negotiable.
What the ASF does not dictate is how you build the model that produces premium rates, provided the resulting rates are actuarially sound. Our model goes through independent actuarial review to confirm it meets ASF solvency requirements. The flexibility lies in the inputs and the methodology, not in whether reserves must be maintained. Any insurer in Portugal that offered prices it could not sustain would be in violation of ASF Norma 4/2007-R on technical provisions. We take that seriously.
Vehicle history and the role of the matrícula
One of the six questions we ask is your vehicle registration number — the matrícula. From the registration number we can retrieve the vehicle's technical data: make, model, year, engine displacement, and official category. This feeds into our model because vehicle type is a genuine predictor of repair costs. Repairing a newer premium SUV costs more than repairing a ten-year-old compact city car. That cost difference belongs in the price.
We do not use the matrícula to check your personal driving record against a public database — no such database is currently available for cross-insurer use in Portugal in the way it exists in some other EU markets. What we use is your self-reported claims history, which you confirm during the application. Under Portuguese insurance contract law (Decreto-Lei 72/2008), deliberate misrepresentation of claims history at application is grounds for policy voidance. Most people therefore report accurately, and we price accordingly.
What the model does not consider
Gender. Age. Nationality. Occupation beyond the primary vehicle use question. None of these are inputs to our pricing model. Some of these exclusions are legally mandated — EU Directive 2004/113/EC, as implemented in Portuguese law, prohibits using gender as a rating factor in insurance. Others are our own choice. Age is a statistical predictor of risk at the population level, but individual data — actual kilometres driven, actual parking conditions, actual claims history — does a better job of predicting individual risk than demographics do. We use the better inputs.
This also means the model is not penalising young drivers simply for being young. A 23-year-old who parks in a private garage, drives 7,000 km per year, and has held their licence for three years without a claim will receive a price that reflects those facts. They may still pay more than a 45-year-old with identical circumstances, because they have less claims history for the model to work with — but the penalty is smaller than under a pure demographic approach.
How the price locks in and the policy is issued
Once you confirm your quote, the price is locked for 24 hours. During that window, if you proceed, your NIF (Número de Identificação Fiscal) is used to generate the policy documents under the underwriting partner's regulatory registration. The policy is issued digitally, with a unique policy number registered with the ASF's database of active policies — the same database that traffic authorities check when a vehicle is stopped. A PDF of your full policy schedule arrives by email within seconds. There is no period where you are waiting for documents to clear a compliance queue.
The three-minute total time — from starting the quote to having a valid, registerable policy in your email — is a function of having no manual steps in the middle. The model runs, the price is confirmed, the payment clears, and the documents generate. What takes weeks at a traditional insurer is the human queue, not the regulatory process.
The technology is the means. The goal is a price that is actually yours.