Etterby Analytics
Menu

Independent analytics consultancy

Senior analytics for complex data and decisions.

Etterby combines analytics engineering, measurement, experimentation, forecasting and applied modelling to solve complex analytical problems end-to-end.

Talk through an analytics problem View representative work

From the underlying data to the decision it supports.

01 / Anatomy of a decision

What would change
the decision?

  1. 01
    Define the evidence

    Question → metric → measurement

  2. 02
    Test the explanation

    Data model → analysis → uncertainty

  3. 03
    Make it operational

    Decision rule → workflow → monitoring

Review what happened. Refine the question.
A working method: connect each analytical step to a decision and an accountable owner.
  • Founder-led
  • Senior practitioner throughout
  • Analytics + engineering
  • Strategy through production
  • UK-based

01 / Expertise

From reliable data to practical use.

Engineering, analysis and implementation belong together. Start with the problem; bring in the depth it needs.

01

Analytics Engineering & Measurement

Can you trust the foundations?

Build trustworthy analytical foundations around the questions your organisation actually needs to answer.

Analytical modelling · Pipelines & orchestration · Reliability · Metric architecture · Semantic layers · Instrumentation

Reliable data. Consistent definitions. Traceable metrics.

02

Advanced Analytics & Decision Science

What does the evidence support?

Apply the right analytical method to important decisions, with assumptions and uncertainty made explicit.

Experimentation · Forecasting · Pricing · Behavioural analysis · Segmentation · Causal analysis · Scenario modelling

A testable explanation and a clearer basis for action.

03

Applied ML, Risk & Decision Systems

Will it work in practice?

Put analytical and predictive logic into reliable operational use, with evaluation and monitoring built in.

Behavioural detection · Fraud & abuse analytics · Feature engineering · Model evaluation · Threshold design · Scoring

Models connected to workflows, monitoring and ownership.

02 / Representative work

Make the problem concrete.

Representative examples of analytical problems, approaches and intended outcomes.

01 / Analytics engineering & measurement

From fragmented reporting to a shared metric layer.

Problem
Different reports give different answers to the same question.
Why difficult
Definitions, grain and refresh timing differ across systems; a single reconciled number can hide those differences.
Approach
Trace a decision back to its metric, agree the definition and reconcile the underlying events.
Outcome sought
An intended result: a metric that teams can explain, reproduce and maintain.

Metric lineage · illustrative

  1. Source events
  2. Tested model
  3. Shared definition
  4. Decision metric
A metric is only as dependable as its definition and the path back to source.
System, implementation & handover
Analytical system
Versioned analytical models, tested transformations and a shared semantic layer.
Implementation
Build the model incrementally, test key assumptions and expose freshness and reconciliation checks.
Handover
Metric definitions, model lineage, test ownership and a runbook for resolving discrepancies.

02 / Experimentation & decision science

From an uncertain change to a decision you can defend.

Problem
A proposed product or pricing change has an unclear effect.
Why difficult
Behaviour varies over time; selection effects and competing outcomes make before-and-after comparisons unreliable.
Approach
Frame a testable hypothesis, choose the allocation unit and specify the decision criteria before testing.
Outcome sought
An intended result: a repeatable way to decide whether to adopt, revise or stop a change.

Decision protocol · illustrative

  1. Hypothesis
  2. Allocation & exposure
  3. Effect & uncertainty
  4. Adopt / revise / stop
Define the decision before seeing the result; retain uncertainty in the readout.
System, implementation & handover
Analytical system
An experiment workflow joining assignment, exposure, outcomes and guardrail metrics.
Implementation
Validate measurement, check allocation and estimate effects with uncertainty made explicit.
Handover
A documented design, reusable readout and ownership of follow-up measurement.

03 / The analytical system

The analysis is one part of the system.

A useful result depends on sound measurement, dependable data and a method that fits the question. It also needs a workflow, monitoring and an owner.

Establish the evidence

  1. 01
    QuestionName the decision.
  2. 02
    MeasurementDefine the evidence.
  3. 03
    Data modelMake it dependable.

Choose the method

  1. 04
    Analysis / experiment / modelTest and quantify.

Put it to work

  1. 05
    Decision workflowPut it into use.
  2. 06
    MonitoringCheck it still works.
  3. 07
    OwnershipEquip the team.
Feedback connects monitoring to the original question. Ownership keeps that loop working.

04 / How Etterby works

Direct involvement. From framing to handover.

The person helping define the problem is also involved in doing the work. Senior judgement stays close to implementation.

  • Hands-on delivery alongside your internal team.
  • Clear reasoning and documentation of what is built.
  • Capability transfer, so your team can own the result.

05 / Technical depth

Built to be used. And maintained.

Tools follow the problem and your existing environment. The result must be something your team can understand, operate and maintain.

Model & measure
SQL · dbt · BI / semantic layers
Build & operate
Python · ClickHouse · Airflow
Analyse & evaluate
Experimentation · Forecasting · Modelling

06 / Founder-led, directly delivered

Senior judgement.
Personal accountability.

You work directly with the senior practitioner responsible for framing and delivering the work, from the first question through implementation and handover.

Etterby connects analytical thinking with hands-on engineering, and works alongside your team to leave the reasoning and the system in your hands.

Start a direct conversation

07 / Insights / Lab

A closer look at the thinking.

Selected technical notes on data architecture, model evaluation and practical delivery.

Analytics engineering / Technical note

When ClickHouse shines

Connect analytical storage choices to investigative questions, modelling requirements and operational ownership.

Applied decision systems / Technical note

From heuristics to ML

Evaluate predictive controls alongside existing rules, with explicit thresholds, review processes and rollback paths.

Start with the question

Where does the work need to go next?

Unclear measurement, a forecast you cannot rely on, or a model that needs to reach production? Start with a direct conversation.

Talk through an analytics problem View representative work