Comparison
Imagine All The People vs Customer Twins
Being faithful to your customers is not representing your market.
A Customer Twin aims to reproduce a customer, or a set of known customers, from observed data. Imagine All The People starts from the target population: customers, non-customers, citizens, employees or stakeholders, including those absent from your proprietary databases.
Four differences, stated plainly.
Thesis
Fidelity to observed individuals and representativeness of the target population are two distinct objectives.
- Customer Twin
Fidelity to observed individuals.
Imagine All The PeopleRepresentativeness of the target population.
- Customer Twin
CRM data, transactions, browsing, interactions and first-party data.
Imagine All The PeopleStatistical data, distributions, constraints and, where relevant, business data.
- Customer Twin
Excellent for understanding and anticipating the behaviour of known customers.
Imagine All The PeopleAlso makes it possible to explore people absent from proprietary databases.
- Customer Twin
Coverage bias in the source data can be reproduced at scale.
Imagine All The PeopleThe target population is explicitly constructed and controlled on the modelled dimensions.
What changes, dimension by dimension.
| Dimension | Customer Twin | Imagine All The People |
|---|---|---|
| Base unit | Customer TwinAn observed customer or individual | Imagine All The PeopleA synthetic individual within a target population |
| Starting point | Customer TwinHistorical individual data | Imagine All The PeopleDecision + population data + constraints |
| Coverage | Customer TwinThe population present in the source data | Imagine All The PeopleThe target population defined for the decision |
| Personal data | Customer TwinOften first-party and individual | Imagine All The PeopleNo individual personal data required to build a population |
| Bias | Customer TwinMay reproduce selection and coverage biases of the source data | Imagine All The PeopleDepends on the data, constraints and explicitly modelled dependencies |
| Non-customers | Customer TwinHard to represent when absent from the data | Imagine All The PeopleCan be included in the target population |
| Collective dynamics | Customer TwinOften centred on individual behaviour | Imagine All The PeopleIndividuals, groups, interactions, influence, cascades |
| Scenarios | Customer TwinDepending on the architecture | Imagine All The PeopleSeveral scenarios on the same population |
| Best use | Customer TwinPersonalisation and customer prediction from what is known | Imagine All The PeopleSimulating a decision across all stakeholders |
Population-centric vs customer-centric.
A Customer Twin first aims to faithfully reproduce observed customers. Imagine All The People first aims to build the population that a decision affects.
This makes it possible to explicitly include people absent from the CRM, non-customers, light users or other populations that are hard to observe.
This does not automatically make the population perfect: its quality depends on reference data, constraints and modelled dependencies.
An illustrative example.
Decision
Should we close a distribution channel?
Customer Twin
Simulate the impact on identified customers who use, or could use, that channel.
Imagine All The People
Also include occasional customers, non-customers, prospects and indirectly affected populations, then simulate collective reactions.
Each approach has its own domain of relevance.
Choose a Customer Twin if…
- personalisation;
- next-best-action;
- CRM;
- churn;
- customer journey;
- optimisation on a known base.
Choose Imagine All The People if…
- new offer;
- new market;
- public policy;
- reputation;
- acceptability;
- decision involving several stakeholders;
- population that goes beyond the CRM base.
What these pages do not claim to prove.
The statistical robustness of the population does not automatically transfer to behavioural answers. Building the population and validating the answers are two distinct layers.
