A property valuation in one second, built on messy data
- Seconds from address to home valuation in the app
- Messy data BBR cleaned before the customer sees a figure
- Around the clock including when advisers are closed
- The data is public, but it cannot be used as it is.
- An answer has to arrive now, not when the calculation finishes.
- One wrong valuation costs more trust than ten right ones earn.
That is what it looks like when public data has to carry a customer-facing feature. These are some of the challenges we have worked on with Jyske Bank.
What was Jyske Bank facing?
Jyske Bank wanted to give customers a property valuation directly in the "Bedste Lån" app, based on public property data from the Danish building register. The idea is simple. The data is not.
Building register data is heavily fragmented: fields are filled in differently, information is missing, and some records contradict each other. On top of that came the requirement for fast response times and stable operation, no matter how many people use the app at once.
The customer gets a valuation immediately, instead of waiting for an adviser.
A property valuation required getting in touch, and public data sat available but unusable in practice.
An accurate valuation in the app in seconds, built on cleaned data and the bank's own criteria.
Part of the advice happens automatically, and the platform can carry more users without being rebuilt.
The work lies in making the data usable
- Cleaning
Fragmented and contradictory register entries are validated, so only data that holds together enters the calculation.
- Sampling
The algorithms find the most relevant data points rather than calculating on everything, which is both faster and more accurate.
- Failure scenarios
Known exceptions are collected and structured, so complex data relationships can be grasped and new errors spotted early.
The result is a valuation the bank can stand behind in front of the customer. It is not the algorithm that decides the quality; it is the discipline in what goes into it.
Jyske Bank's own developers had to be able to build on it themselves
Jyske Bank has capable app developers of its own. The task was not to build their app, but to give them something that was easy to build on, and that did not tie them to us.
A finished component for the app
- Faster to get something visible in the app
- But the bank's developers are tied to our choices
- The feature can only be used in one place
A developer-friendly and flexible API
- The bank's own developers control the user experience
- The same service can be used by several solutions
- A clear separation between our responsibility and theirs
Three moves that hold response time and stability
Jyske Bank knows the customers, the credit assessment and the business behind "Bedste Lån". Our contribution was the technical part: getting public data to deliver reliable answers fast enough to feel instant.
- 01
Caching with judgement
Heavy calculations are reused where it is sound to do so, keeping response times down without compromising correctness.
- 02
Scalable infrastructure
The service can handle more requests as the user base grows, without the architecture having to be redone.
- 03
Proactive error monitoring
Failure scenarios are collected and visualised, so problems are spotted by us before they are spotted by a customer.
Responsibility for operations, not just delivery
We are technical partner to Jyske Bank on this part of the solution, and that includes responsibility for keeping the service running. An API that delivers valuations to customers of a bank must not be something someone has to remember to keep an eye on.
So monitoring and error handling are part of the delivery, not something that comes after. 98.5 per cent stable operation is not a number you hit by being lucky.
A foundation that can carry more than one app
Because the valuation sits in a service of its own, it can be used by other solutions in the bank without being built again. And because the infrastructure is scalable, more users do not cost a new project.
- Industry
- Banking and finance
- Platform
- API and server solution
- Our role
- Development partner with operational responsibility under a service agreement
- Capabilities
- System development
- Technologies
- Kotlin, Go, Postgres, Directus CMS
- The collaboration
- Ongoing collaboration on the engine behind the app
Do you have data that needs to carry a customer-facing solution?
- A reply from one of us by the next business day.
- A no-obligation review of your situation.
- Clarity on which options are relevant to you.
- A concrete plan, if you choose to go ahead.