
Analysis and forecasts for those who have to decide
In an insurer's claims management, for example, not every claim notification deserves the same amount of time: a model ranks them by complexity and fraud risk and assigns each one to the right claims handler, so the handler spends time where it is needed and the decision on the claim remains theirs. This is one of many cases in which we build analysis and predictive models with Data Intelligence.
Data Intelligence · What it contains
- Reporting and management analysis
- Dashboards for management
- Advanced analytics and segmentation
- Customer segments
- Predictive and machine learning models
- Demand, risk, anomalies
- Scoring and time series
- Scores and historical series
- Natural language querying
- Questions in your own language, answers with the source
What it contains
On the company's data we build the models decision-makers need, for example to anticipate customers at risk of leaving, pick up the signs of credit deterioration, forecast liquidity or work out which product to offer to whom. In all these cases the model provides estimates and priorities, while the decision stays with people.
01
Reporting and management analysis
Dashboards and analysis for management, built on metrics and analytics certified in Data Foundation, so everyone reads the same number.
02
Advanced analytics and segmentation
We combine the organisation's data with open and third-party data to find relationships that help decisions, for example in customer segmentation.
03
Predictive and machine learning models
Models that estimate demand, propensity, risk and anomalies, with scenario simulations to see what changes if the assumptions change.
04
Scoring and time series
Scores that rank the work and route it, and time series through which the models learn from the history of each process.
05
Natural language querying
People query the data in their own language, without jargon, and get immediate answers based on certified data, with the source in view.
06
Model monitoring
Performance, drift and explainability of the models in production, readable by those who decide and those who audit.
The accounts to chase, ranked by risk
The models read the payment history, the amounts overdue and the reminders already sent, and put the highest-risk accounts at the top with the reasons alongside. The repayment plan remains the credit manager's decision, and the errors they report are used to correct the model, to the benefit of the accounts that follow.
- The models read the payment history
- They put the highest-risk accounts at the top
- The credit manager approves the repayment plan
- Reported errors correct the model
Illustrative example
Frequently asked questions
We already have a business intelligence tool. Do we need more dashboards?
We usually work inside the tools you already have, because people know them and replacing them would cost time and money; our contribution lies in the certified metrics and the models those tools display.
Do you need agents to use Data Intelligence?
No. A forecast or a score is just as useful to people working without agents, and many projects start here.
Does a score decide in place of the analyst?
No, it ranks the work. Decisions that affect a customer are taken and justified by a person, and the model that contributed is recorded alongside, with the version in use.
How do we notice that a model is getting things wrong?
We monitor the performance and drift of the models in production and receive an alert when results move away from what was expected.
Start from a decision taken today without reliable numbers
If you have a disputed metric or a forecast that is done by hand today, we can look at them together and tell you what it would take to build them on your data.