Data consultancy

The Pixel Scientists

Insight scientists and data storytellers. We turn complex, messy data into decisions people trust, and tell the story so they act on them.

Scroll to sort it

Most data arrives like this. Behind this text, coloured streaks fill the screen like noise.

Bright, loud and hard to read. The answer you need is in there somewhere.

The noise, sorted

The same 42 streaks, measured four ways. Data tells many stories. We find the one that answers your question.

Measure the streaks by

Magenta has the most lines, 11 of the 42. Laid end to end, cyan would cross the screen 4.2 times, from only 9 lines. Yellow covers the most ground, with 52% of all the colour on screen, from only 8 lines. Violet draws the longest lines, 69% of the screen width on average, from only 3 lines.

Show the numbers
Streaks in the opening frame, measured four ways, by colour
ColourLinesLength (screens)AreaAvg length
Magenta111.95%18%
Cyan94.215%47%
Yellow82.852%35%
Blue61.812%30%
Orange51.411%28%
Violet32.15%69%

What we do

From raw data to a decision

We work across the whole pipeline, so nothing gets lost between the people who build the data and the people who use it.

  1. 01 Engineer

    Pipelines, warehouses and the infrastructure under them. Extract and load through Fivetran and Airbyte, with DataOps and MLOps on Google Cloud and AWS.

  2. 02 Model

    The transform in ELT: raw data shaped into tested, documented models in the warehouse. Then statistical and machine-learning models your team can question and understand.

  3. 03 Analyse

    Analysis, reporting and business intelligence. Statistics that stand up to scrutiny, delivered where your team will use them: in Tableau, Looker or Power BI, in the tools you already have, or in a bespoke dashboard.

  4. 04 Explain

    Data visualisation, scientific and technical writing, and C-suite reporting. Interactive graphics in D3.js and Shiny when a static chart is not enough.

Each strip above is the same twelve pixels, a little more ordered at every step.

Skills and stack

The methods and tools behind the work

We are scientists first, and hands-on across the whole stack. These are the methods and tools we use most. On every project we start from your question and the tools your team already has.

  • Statistics and survey science

    Careful inference from surveys and experiments, with the uncertainty stated plainly.

    • R
    • Python
    • Survey design and weighting
    • Design effects
    • Margins of error
    • Power and sample size
    • Regression and GLMs
    • Multilevel and mixed models
    • Robust and clustered errors
    • Bayesian methods
    • Causal inference
    • SAS
    • Julia
  • Machine learning and forecasting

    Predictive models that are tested honestly, calibrated and explained.

    • Python
    • scikit-learn
    • Gradient boosting
    • Random forests
    • Regularised regression
    • Cross-validation
    • Calibration
    • Forecasting and nowcasting
    • Clustering and PCA
    • XGBoost
    • PyTorch
  • AI and language models

    Language models put to work with guardrails. They answer from your data, and people make the decisions.

    • LLM applications
    • Text-to-SQL
    • Retrieval-augmented generation
    • Prompt and evaluation design
    • Agentic workflows
    • Model Context Protocol
    • Claude
    • Gemini on Vertex AI
    • OpenAI
    • AI governance
  • Data engineering and warehousing

    Pipelines and warehouses that keep data tested, documented and on time.

    • SQL
    • BigQuery
    • Dataform
    • dbt
    • Dimensional modelling
    • Incremental ELT
    • Data quality tests
    • PostgreSQL
    • DuckDB
    • Fivetran
    • Airbyte
    • Amazon Redshift
  • FinOps and financial analysis

    Responsive models and predictions on large financial data, refreshed as the numbers move, so finance teams act on what is happening now.

    • FinOps
    • Financial analysis
    • Near-real-time data models
    • Prediction at scale
    • Big data
    • BigQuery
    • Vertex AI
    • Google Cloud
  • Visualisation and data storytelling

    Charts, maps and interactive stories that make complex results clear at a glance.

    • D3.js
    • Observable Plot
    • Plotly
    • three.js
    • Maps and GIS
    • ggplot2
    • matplotlib
    • Shiny
    • Tableau
    • Looker
    • Power BI
  • Product and experimentation analytics

    Cohorts, funnels and experiments that show what is changing, and why.

    • Cohort and retention analysis
    • Funnels
    • A/B testing
    • Churn and revenue metrics
    • Event data
    • Counterfactual baselines
  • Data products and platforms

    Accessible apps, dashboards and APIs, deployed securely and easy to hand over.

    • TypeScript
    • JavaScript
    • React
    • Next.js
    • Node.js
    • Docker
    • Google Cloud
    • AWS
    • Cloudflare
    • GitHub Actions
    • Terraform
    • Playwright
    • Accessibility (WCAG 2.2)
  • Design, publishing and interactive

    Design systems, print-ready reports and interactive work, produced from the same data.

    • Figma
    • Design systems
    • InDesign
    • Illustrator
    • Report automation
    • LaTeX
    • Motion design
    • Godot

How we work

Substance is style. People, not programs.

Substance is style

Good design starts with getting the numbers right. We do the analysis properly first, then shape the chart, report or dashboard so the right reading is the easy one. We never fudge the numbers to fit a story. Data quality problems are raised with you, discussed, fixed and recorded, so every change is in the open.

People, not programs

You work directly with the scientists doing your analysis. We choose tools to suit the problem and the people who will use the results, and we explain every choice. We won't oversell you or lock you in. We train your team, and we trust them to run things.

Bottom up

We start with the people at the coalface, who do the work every day. We learn how it really happens, build on the tools they already use, and add something new only when it earns its place. Nothing lands from on high as future tech debt or an expensive licence nobody opens.

Open source

Fairfold

Our mantra, written as software. Fairfold is the open-source side of The Pixel Scientists. Any organisation can use it. We build it for the ones whose money should go on their mission, such as charities, foundations and funders, community groups, social enterprises and CICs, housing associations, membership bodies, councils and university research offices.

Their digital budgets are tight, and too much of them goes on software subscriptions built for someone else. That software is often expensive, complex, hard to leave and a poor fit for the work. Small teams pay for features they never use, while the real work stays in spreadsheets.

Fairfold is open source and free. Run it yourself and build on it in house, with a team you trust, or have us (or someone else) run it for you. Every feature is in the open, and your data is yours to take with you.

Substance is style here too. Each tool does one job properly, with accessibility and data protection built in from the start. Every tool shares one platform, so you switch on only what you need and enter each fact once.

23 tools, one platform

  • Accounts and permissions
  • Audit log
  • Shared record of organisations and people
  • Data warehouse
5 in development 6 in discovery 12 planned

Fairfold Grants leads the way, with a complete funding round on synthetic data due at the end of October 2026. The first live round is planned for March 2027, and every tool by the end of 2027. These are plans, and dates can move: the Fairfold timeline has the latest.

Licence
AGPL-3.0-or-later for code, CC BY 4.0 for docs
Will run
Self-hosted with Docker Compose, or as a managed service, from 2027
What the evidence says
  • 34%of small UK charities say software and licences are one of the things they most need funding for. For large charities it is 37%.UK, 2026
  • 38%of large UK charities say the limits of their CRM, such as being hard to use, poor reporting or not connecting to other systems, are one of their biggest barriers.UK, 2026
  • 22%of small UK charities got any funding towards digital costs in the past year, against 37% of large charities.UK, 2026
  • 13.2%of their budget goes on technology, on average, for small US non-profits (under $1 million), against 1.5% for the largest. The typical (median) small non-profit spends 2.6%, against 1.0%, so the cost falls hardest on the smallest.US, 2017

Sources: Charity Digital Skills Report 2026 (UK). NTEN Nonprofit Technology Staffing and Investments Report 2017 (US). In the UK report, squeezed finances are the biggest barrier to digital for small and large charities alike.

Fairfold Grants

In development

A complete funding round on synthetic data is due at the end of October 2026. Planned for launch in March 2027.

Runs a funding round from start to finish, from setting up a programme and taking applications to private decisions, releasing outcomes and exporting data to the 360Giving Data Standard.

modules/grants

Fairfold CRM

In development

The shared record it builds on is being built now. Planned for launch in March 2027.

Relationships, interactions and consent for the organisations and people you work with.

modules/crm

Fairfold Data Protection

In development

Its audit log, data classification and retention rules are part of the platform build. Planned for launch in March 2027.

Subject access requests, breaches, impact assessments and records of processing, using how every tool classifies its data.

modules/privacy

Fairfold Forms

In development

Its form engine is being built for Grants. Planned for launch in March 2027.

Forms and surveys for feedback, registrations and more, built on the same form engine as grant applications.

modules/forms

Fairfold Reporting

In development

The shared data warehouse it reports from is part of the platform build. Planned for launch in March 2027.

Reports for trustees and regulators from the shared data warehouse.

modules/insight

Fairfold Due Diligence

In discovery

Planned for launch in March 2027.

Checks on applicants and grantees, from registry lookups and sanctions screening to fraud and safeguarding checks and monitoring visits.

modules/assure

Fairfold Payments

In discovery

Planned for launch in June 2027.

Approved grant payments, donations and money owed, sent through regulated payment providers and into your accounting system. It will not hold a ledger or card details.

modules/finance

Fairfold Impact

In discovery

Planned for launch in June 2027.

Outcomes, indicators and evidence from grants and activities, so trustees can see what difference the money makes.

modules/impact

Fairfold Governance

In discovery

Planned for launch in June 2027.

Trustee boards, board packs, decisions, conflicts of interest, risks, policies and incidents in one place.

modules/governance

Fairfold Fundraising

In discovery

Planned for launch in December 2027.

Supporters, donations, Gift Aid and campaigns, for charities that raise money as well as give it.

modules/fundraising

Fairfold Case Management

In discovery

Planned for launch in December 2027.

Casework for service teams, with referrals, assessments, plans and outcomes for the people a charity supports.

modules/cases

Fairfold Integrations

Planned

Planned for launch in June 2027.

A plugin API and connections to other systems, with tools to bring data in from what you use today.

modules/connect

Fairfold Communications

Planned

Planned for launch in June 2027.

Newsletters, feedback and stories, sent only to people who have agreed to hear from you.

modules/engage

Fairfold Activities

Planned

Planned for launch in September 2027.

A charity's own direct charitable activities, recorded beside its grants in the terms of the charity SORP, with procurement, contracts and support for grantees.

modules/dca

Fairfold Field App

Planned

Planned for launch in September 2027.

A mobile app for work away from the office, such as monitoring visits, that keeps working without a signal.

modules/field

Fairfold Recruitment

Planned

Planned for launch in September 2027.

Fair recruitment first, with adverts, applications and shortlisting that reduces bias.

modules/people

Fairfold Volunteers

Planned

Planned for launch in September 2027.

Recruiting, checking, scheduling and thanking volunteers.

modules/volunteers

Fairfold Fund Accounting

Planned

Planned for launch in September 2027.

Restricted and unrestricted funds, budgets and commitments, so a finance team can see what is spent and promised against each fund.

modules/funds

Fairfold Bid Manager

Planned

Planned for launch in September 2027.

Finding funders and managing bids, for charities that apply for grants.

modules/finder

Fairfold Membership

Planned

Planned for launch in December 2027.

Members, subscriptions, events and continuing professional development, for membership bodies.

modules/membership

Fairfold Assistant

Planned

Planned for launch in December 2027.

A governed assistant that helps staff draft, summarise and find things, and leaves every decision to people.

modules/assist

Fairfold Funding Map

Planned

Planned for launch in December 2027.

Who funds what, and where, built from open grants data.

modules/atlas

Fairfold Organisation Profile

Planned

Planned for launch in December 2027.

An organisation's details, given once and reused with every funder it applies to.

modules/passport

Where we come from

Scientific rigour, any data

Science is our background. Between us we hold PhDs in immunology, physiology and human genetics, so we were trained to design careful studies, test our own results and say plainly how sure we are.

We bring that rigour to every kind of data. Our clients range from biotech and pharma to media, education and automotive software.

  • Lab data sequencing, assays, clinical studies
  • Product data app events and usage
  • Audiences media and content
  • Surveys polls and panels
  • Sales commerce and operations
  • Financial data transactions, KYC, audit trails
  • Funding grants and programmes

Clients

Trusted in science, media and technology

  • BBC
  • Berg
  • Labster
  • Redx Pharma
  • Azure Biosystems
  • Genefood
  • Keyloop
  • Boundless

Team

The people you will work with

Contact

Tell us what you need to find out

Send a short note about your data and the question it needs to answer. One of us will get back to you.

Email

Registered office

The Pixel Scientists Ltd
3rd Floor, 86–90 Paul Street
London EC2A 4NE