Data foundations for citizen-scale public services

Butterfly Data helps UK public sector organisations get national-scale, citizen-level data accurate, governed and connected, so you can use analytics and AI responsibly. By combining the deep expertise of our UK-based specialists with global-scale technology and innovation, we help you make every decision with confidence, knowing your data reliable and ready to drive impact.

Our trusted clients and partners

Who we are

Butterfly Data is an employee-owned, B Corp certified data consultancy and technology solutions provider. For over 20 years we have worked on some of the UK’s largest and most sensitive citizen datasets for government departments, defence, policing and health.

Our specialism is the work that has to happen before AI can be relied on, such as data governance, data quality, interoperability and migrating data off legacy systems. With those foundations in place, we build the analytics and AI that sit on top.

Our services

Turn untapped potential into continuous improvement

Data quality, governance, and privacy

Ensure your data is accurate, well-governed, and safeguarded for evolving privacy standards, whilst establishing a trusted foundation for AI.

Data engineering, integration, and cloud adoption

Design and implement scalable data platforms that enable seamless integration, automation and cloud-based operations to support modern analytics and AI solutions.

Data analytics and visualisation

Transform complex datasets into clear, interactive visual insights that support smarter, faster decision-making.

Data science and AI solutions

Apply advanced AI and machine learning to unlock predictive insights, automate workflows, and drive measurable business value.

Our experience

Why Butterfly Data?

Proven expertise

For more than 20 years our data specialists have worked on national datasets holding millions of citizen records, in secure and compliant environments. You get that depth without the overheads of a larger firm.

Abstract design of curved golden lines forming a flowing wave pattern on a black background.

Innovative technology

We use cutting-edge technologies from leading vendors like SAS, Databricks, and Snowflake to boost performance and accelerate business transformation.

Abstract yellow curved lines forming a flowing wave pattern on a black background.

A personalised approach

The people you meet are the people who deliver. We start with discovery, so the work is shaped by your data and your objectives rather than a standard package, and we work alongside your in-house team from the first conversation to handover.

Spiral pattern made of small golden dots on a black background.

Data for good

As a proud B-Corp, we use the power of data for good – partnering and collaborating with organisations that align with our core values to create a positive impact.

Abstract flowing yellow mesh lines forming wave-like curves on a black background.

Measurable results

Chosen by industry leaders for our agility and commitment to excellence, we let the data speak for itself.

Yellow curved parallel lines forming an abstract wave pattern on a black background.

Simple procurement

Easily procure our services, either directly or via key public sector frameworks, including G-Cloud, DOS, Spark, ACE, and NVfI.

Abstract geometric design of overlapping yellow square lines forming a diamond shape on a black background.

“The invaluable work that Butterfly Data have undertaken with a key collaborator of mine will feed directly into my work, making it both simpler and faster and enabling me to better identify data gaps. Incredibly useful. Thank you."

Abstract dark background with smooth, curved yellow light streaks.
Butterfly DATA guide

70+ projects. 35+ organisations. One brochure.

If you are evaluating data consultancies, our brochure sets out our services, our track record and how we work.

Laptop showing a Butterfly Data report titled Put data behind every decision
resources

Insights to power better decisions

Sara Boltman, founder and CEO of Butterfly Data, has been nominated as a candidate for techUK’s Central Government Council.

The techUK Central Government Council is made up of 22 industry representatives elected by techUK members, 11 of which are SMEs. It shapes techUK’s work with central government and acts as a link between the tech sector and government departments. It also promotes open, practical dialogue that helps government get better value from technology and gives suppliers of every size, including SMEs like Butterfly Data, a fair chance to compete.

Sara founded Butterfly Data in 2003 and has grown it from two people to around 50, with offices in Cardiff and Manchester. Butterfly Data delivers data and analytics work across central government, including for HMRC, DWP and FCDO. Sara also serves on the DAMA UK committee and mentors female founders.

If elected, Sara will focus on three areas. The first is making sure SMEs that deliver for government every day can compete on quality, not be screened out by requirements that are disproportionate to the work. The second is bringing the voice of suppliers across the UK’s nations and regions to the Council, not just London. The third is sharing practical experience of making data and AI work responsibly at scale, as departments move AI from pilots to live services.

“I will bring the perspective of an SME that delivers for Whitehall every day,” said Sara. “I want to help techUK push for proportionate requirements and earlier, more open market engagement, so departments can access the full breadth of UK innovation.”

Voting is open until 30 November. Each techUK member company votes through its nominated representative, who can vote using the form here. Read Sara’s full candidate statement.

Picture police forces right across the nation, every crew pulling together toward a shared mission to protect the public. Yet when teams work out of disconnected boats, even brilliant local innovations stay trapped on local shores. We see incredible ingenuity went into bespoke regional tech, happening simultaneously alongside an urgent need for live, cross-border data.

We aren't off course, but we can move so much faster in formation. Instead of paddling in standalone circles, policing can lash its expertise together, sharing resources, connecting networks, and navigating as one unified fleet. 

Taking full stock of the fleet's cargo: scaling local invention 

Instead of every boat building its own gear from scratch, we need to inspect what’s already on deck.

Digital teams across forces regularly build smart local solutions, from automated AI redaction pipelines to predictive risk calculators. But without a shared inventory, neighboring crews spend precious time re-engineering the exact same tools.

  • Central Deck Storefront: Host all shared software in a central, secure cloud hub managed by the Police Digital Service. This gives every force instant access to proven, pre-vetted digital tools, allowing local teams to roll out new capabilities in minutes rather than spending months building them from scratch.
  • Deck-to-deck Sharing: Let pioneering forces share their proven digital innovations directly with neighboring teams. This cuts out duplication of processes/innovation/purchases and allows smaller forces to immediately operate with top-tier capabilities.

Navigating from one map: standardising data language

Lashing ships together only works if every captain reads from the exact same chart.

Right now, cross-border intelligence hits a wall because regional platforms speak different digital dialects. By standardising how we label and structure information across the country, we ensure vital signals pass cleanly across every vessel in the fleet, instantly, seamlessly, and without loss in translation. The same applies to digital evidence, where fragmented workflows across forces slow down forensic processing. We look at this in more detail in how policing can improve its capture and analysis of digital evidence.

  • A Universal Navigation Chart: Adopt a single, standardised blueprint for how every system records core details: Person, Object, Location, and Event. When every force structures its records using the same format, software can cross-reference facts automatically across regional borders, stripping away the friction of manual data sorting.
  • Lighting Up the National Radar: Connect local system tags directly into the central Law Enforcement Data Service. That way, when an officer logs a "Vulnerability" or "Person of Interest" flag in South Wales, that critical insight instantly lights up on the operational radar of a crew in London, eliminating phone calls, email chains, and dangerous delays.

Securing the rigging: connecting the fleet in real time

A fleet stays coordinated through strong, dynamic connections, not manual signals or slow paper messages sent deck to deck.

Layering modern API connections over existing infrastructure acts as modern rigging, streaming vital intelligence directly to frontline mobile officers.

  • Secure Real Time Channels: Connect existing core platforms directly across force borders so officers can stream live intelligence to their mobile devices. Smart security is baked into every query, instantly verifying an officer's role and live situation so sensitive data stays protected without putting the brakes on frontline response.
    ‍

‍

By taking stock of our shared inventory, aligning our navigation charts, and securing our digital connections, we can turn disconnected boats into a fully synchronised fleet. This shift is not about replacing what works locally, but about linking our strengths so that good ideas in one region benefit the entire country.

When policing operates on a shared current, information flows without delay, resources go further, and frontline officers get the modern support they deserve to keep communities safe.

Digital evidence is defined as information of investigatory value that is stored or transmitted electronically. Sourced from hardware such as hard drives, mobile phones, and CCTV systems, this umbrella term spans across data artifacts including electronic mail, contact directories, communication logs, images, and audio visual recordings.

Today, the digital forensic landscape operates within a framework that is increasingly strained by the sheer volume of incoming electronic evidence. As investigators routinely seize hardware ranging from mobile devices to surveillance infrastructure, laboratories face the complex task of isolating relevant digital artifacts from overwhelming quantities of non-probative data. This persistent mismatch between laboratory resources and the scale of seized media has resulted in a staggering national backlog, with wait times now routinely stretching past the 12-month mark for over 20,000 queued devices.

The Systemic Bottlenecks in Forensic Processing

As highlighted by the NPCC , over 90% of reported offences now feature a digital component. However, reflecting on the seizure process, the initial collection of this evidence is predominantly conducted by frontline officers rather than Subject Matter Experts (SMEs). This practice generates unnecessary processing delays during intake, as un-triaged devices are submitted in bulk to specialised laboratories.

Delving deeper into the issue, police forces face severe SME retention deficits, outdated technical infrastructure, and complex legal frameworks where administrative authorisation timelines frequently exceed execution times. Qualified specialists are routinely overwhelmed by manual, repetitive extraction tasks, leading to workload fatigue and high staff turnover. These internal pressure points are made worse by a rising wave of external cyber threats, with national data showing that 19% of UK businesses and 14% of charities fall victim to cybercrime each year. As capabilities lag behind criminal technology, law enforcement requires systemic overhauls, including centralised national digital platforms and dedicated AI centers, to manage the expanding workload efficiently.

A Technical Roadmap for Digital Forensics

Addressing this force-wide bottleneck requires a fundamental paradigm shift toward automated triage, edge processing, and a standardised digital data architecture. Data science and automated engineering present clear opportunities to transform this crisis into an efficient, scalable pipeline:

Automation and AI Integration

Integrating artificial intelligence within the primary digital data sweeping phase, specifically utilising automated text extraction and computer vision algorithms, allows systems to automatically flag incriminating data artifacts. Image recognition models and natural language processing can scan raw media for specific keywords, visual indicators, or other needed material. While human-in-the-loop oversight remains mandatory for court admissibility, automating this initial pass dramatically minimises manual man-hours.

Standardised Workflows and Analytics

Fragmented workflows across different forces create inconsistent outcomes and difficulty to share resources across the different forces. Implementing standardised, automated ingestion scripts ensures evidence is systematically categorised, sorted, and routed to the correct department without continuous manual intervention. Furthermore, deploying force-wide analytical software allows investigators to automatically cross-reference identifiers, phone numbers, nicknames, and digital signatures across separate cases, uncovering hidden connections that could otherwise be missed.

Targeted Upskilling 

Technology must be supported by foundational operational skills. Establishing core digital etiquette across all frontline staff ensures evidence integrity from the moment of seizure. Embedding tactical "Digital Champions" within standard investigative units bridges the knowledge gap between frontline officers and expert tech laboratories

‍

‍

Conclusion

Ultimately, resolving the digital forensics crisis requires closing the gap between raw data collection and actionable intelligence. By modernising legacy infrastructure, standardising ingestion pipelines, and deploying automated triage tools, police forces can eliminate processing backlogs, support frontline personnel, and accelerate the administration of justice.

‍

Questions about data quality, migration and AI readiness

What to know about data quality and getting data ready for AI.

Where do data quality problems start?

Most start at the point data enters a system: a form, a feed from another organisation, or a migration from a legacy platform. Fixing quality there costs far less than cleaning it later, which is the logic of the 1-10-100 rule. In government, much of that incoming data comes from other departments, banks or suppliers, so quality checks need to sit at the boundary.

Why does AI need data foundations first?

An AI model learns from and acts on the data it is given. If records are duplicated, incomplete or poorly matched, the model repeats those errors at scale. Data quality, clear ownership and documented lineage come first, so AI outputs can be explained and trusted.

Ready to transform your data?

Book a free discovery call to explore how our tailored data solutions can help you manage complex datasets, gain actionable insights and drive measurable results.