Applied Behavioral Economics
Predictive models for retention, conversion and customer experience built on behavioral economics.
The challenge
Customer decisions do not follow the rational logic assumed by traditional retention and conversion models.
Our approach
- Identification of biases and decision moments along the journey
- Predictive models of churn and propensity
- Redesign of communication, offer and touchpoints
- Controlled experimentation and impact measurement
Deliverables
- Predictive churn and conversion model
- Journey redesign and contact scripts
- Experimentation program with control metrics
- Quantified impact on retained revenue
Decision Design
The customer doesn't decide the way they say they decide.
Before changing the offer, we change the decision context: it is almost always cheaper and faster.
Decision moments
We identify the points in the journey where the customer is truly won or lost, and the biases operating at each one.
- Journey and friction map
- Bias inventory by moment
- Abandonment analytics
Predictive model
We build churn, conversion and value propensity models with behavioral variables, not just demographic ones.
- Churn and propensity model
- Behavioral segmentation
- Prioritization by value at risk
Decision redesign
We redesign offer, communication, defaults and contact scripts so the right choice is the easiest one.
- Default and framing redesign
- Contact scripts and content
- Choice architecture
Experimentation
We test with control groups, measure real incrementality and scale only what proves impact.
- A/B test design with control
- Incrementality measurement
- Scaling and monitoring
DD·NS methodology path
Sequence of phases, with the analytical focus of each one and the decision point that closes it.
Fig. 1 · Phase flow
Real effect versus the control group
NS ranges come from interventions measured at scale: the real effect is around +1.4 points, far below the +8.7 points typically reported in academic literature.
Fig. 3 · Adoption rate: control vs. intervention at scale
Reference ranges behind the framework
Index values from the NS engagement base that we use to calibrate hypotheses and targets before measuring with the client's own data.
true average effect of a nudge across 126 large-scale experiments with 23 million people (+8.0% over control)
NS Base · Behavioral Economicseffect reported in academic publications; ~70% of the gap is explained by publication bias and low statistical power
NS Base · Behavioral EconomicsFig. 4 · NS reference ranges
Deliverables map
Each deliverable is anchored to the phase where it is produced and validated with the client team.
Fig. 2 · Deliverables by phase
The figures use reference ranges and index values from the NS method, calibrated with our project base and market tracking. Each project's results are reported with the client's actual figures.
Lead partner
José Manuel Noriega
Business economics: price, growth, margin and capital.
See partner profilesAI as an accelerator in this practice
We combine behavioral economics with predictive models to anticipate customer decisions.
- Churn, purchase and offer-acceptance propensity models trained on real behavior
- Unsupervised behavioral segmentation to design nudges by profile
- Design and automated reading of A/B experiments with statistical significance
- Next Best Action engine that prioritizes the intervention per customer
We work with data governance, confidentiality controls and human validation on every deliverable: AI accelerates the analysis, it never replaces the decision.
Impact cases
portfolio churn
Predictive churn model based on behavioral economics, implemented with the Customer Service Department.
Leading insurer · Peru
Insights from this practice
Frameworks and criteria we use on projects, explained in detail.
Why customers who said they were satisfied still leave
Stated satisfaction is a poor predictor of retention. What predicts well is observed behavior at the critical moments of the journey.
6 min readBehavioralChoice architecture: how to present options so people choose well
How alternatives are presented changes the decision as much as the content of the alternatives themselves.
5 min readBehavioralRunning experiments with a real control group
Without a control group there is no learning: there's a story told with the numbers that survived.
6 min readLet's talk about your specific case
A first conversation is enough to size the opportunity.