Methodology

How we turn live consumer behaviour into institutional intelligence.

A transparent, versioned methodology covering data capture, anonymisation, cleaning, validation, segmentation and reporting. Reviewed quarterly, open to client challenge.

1. Data source

All Wealth Intelligence data originates from a national network of free, consumer-facing financial planning tools we operate. Households use these tools to model real decisions across ISA, pensions, mortgages, protection and inheritance.

Because the tools are free, no-login and openly accessible, the sample reflects organic UK consumer behaviour rather than the self-selected universe of any single platform, advice firm or provider book.

2. Anonymity at source

No personally identifiable information is collected from any user of the network. There is no login, no name, no email, no national insurance number, no postcode finer than partial outward code, and no device-level identifier captured for analytics.

What is captured

Anonymous inputs to a calculator, the option chosen, the modelled outcome, and a coarse timestamp.

What is not captured

No names, emails, addresses, IPs, device IDs, payment data, or any field that could re-identify an individual.

3. Behavioural capture

Each tool emits a structured event for every meaningful step of a modelling journey: the inputs supplied, the assumptions selected, the result rendered, and any subsequent re-modelling. We measure actual decisions, not stated intent.

  • Event-level granularity, with no cross-session linkage.
  • Coarse temporal resolution to prevent inference.
  • No A/B test fingerprinting that could identify a session.

4. Cleaning & validation

Raw events are passed through a deterministic cleaning pipeline that removes test traffic, automated agents, abandoned sessions and clearly impossible inputs. Validation rules are versioned and published with each report.

Bot & test removal

Server-side rules and behavioural heuristics strip non-human traffic before any aggregation.

Plausibility rules

Inputs outside published HMRC limits or otherwise impossible are flagged and excluded.

5. Segmentation

Cleaned behaviour is segmented along dimensions that matter to wealth managers and product teams: life stage, household wealth band, region, product category, and decision archetype. All segmentation operates on aggregated cohorts, never on individuals.

6. External benchmarks

Distributions are continuously benchmarked against ONS Wealth and Assets, HMRC pension and ISA statistics, FCA retail investor data and Bank of England household balance sheets. Where the network sample diverges materially, divergence is reported transparently rather than corrected silently.

7. Confidence & sample sizes

Every metric in every Wealth Intelligence output carries an explicit sample size and a confidence band. Cells with insufficient sample to support an institutional claim are suppressed rather than reported with false precision.

  • Minimum cell size before publication: configurable, with a default floor.
  • Confidence bands published alongside point estimates.
  • Material month-on-month moves cross-checked for spurious volatility.

8. Governance & review

The methodology is owned by the Wealth Intelligence research team and reviewed quarterly. Material changes are versioned, dated and communicated to clients. Clients can request the underlying methodology pack and challenge any element of it.

Request the full methodology pack