Put data behind every decision

Designed for public sector leaders, data professionals, and AI governance teams, this 20-minute talk focuses on real-world implementation, not theory, to help you assess whether your data is truly fit for AI use.

What you will learn:

• Why data provenance is essential for trustworthy AI
• Common data risks in public sector AI projects
• How to evaluate data readiness for AI initiatives
• Practical steps to improve data governance and quality
• A “farm-to-table” framework for ethical AI data

There is a question that doesn’t get asked often enough in AI projects: “Where did this data actually come from?”

Not “what does it contain” or “how clean is it” — those matter too — but the more fundamental question of origin, ownership and handling. Without a clear answer, you are not building on a foundation. You are building on assumptions.

Consider how the best restaurants approach their ingredients. The farm-to-table movement didn’t take off because people suddenly cared about carrots — it took off because provenance became a proxy for quality and trust. Diners started asking which farm, which season and which supplier. And chefs who could answer those questions with confidence built reputations that those who couldn’t simply couldn’t match.

The same principle applies to AI. Data provenance — the ability to trace the origin, ownership and handling of every dataset used in a model — is the farm-to-table standard for responsible AI. And just like in food, the organisations that can’t account for where their ingredients came from are the ones most likely to end up with problems on their hands.

This idea was explored in more depth during a recent webinar based on Maja’s presentation for Digital Leaders AI Public Sector Week, where the discussion highlighted how provenance is rapidly becoming a core requirement for trustworthy AI — not just a “nice to have” for governance teams.

The farm-to-table standard for data

In food, farm-to-table means complete transparency: you know which farm your tomatoes came from, when they were picked and how they were transported. In data terms, this is provenance — and data lineage takes it further still, tracing every transformation the data has undergone on its journey into your model.

In practice, this means being able to answer questions like the following: Who collected this data? Under what conditions? Has it changed hands? Has it been filtered, merged or modified? Is the consent still valid for the way we’re now using it?

These aren’t bureaucratic questions. They are the difference between data you can rely on and data that quietly undermines everything built on top of it.

When provenance is unclear, data stops being an asset and becomes a potential liability. Organisations working in regulated environments — government, healthcare, defence, finance — know this all too well. “Dark data” (unlabelled, unused, untracked, unverified or poorly governed) is the equivalent of ingredients with no label and no known supplier. A chef who used them would lose their kitchen. An organisation that builds AI on them risks much the same.

Clean data is not the same as trusted data

This is a distinction worth making clearly, because the two are often confused.

Clean data has been processed by data quality rules: duplicates removed, formats were standardised and outliers were handled. That is genuinely important work. A dataset where “Male”, “M” and “1” all coexist in the same column is going to cause problems downstream. Consistency matters — it is the data equivalent of mise en place, making sure everything is prepared and in order before you start cooking.

But a vegetable can be perfectly scrubbed, peeled and sliced, and still be dangerous if it was grown in contaminated soil. That is the limit of cleaning. It removes surface-level problems, but it can’t fix what’s built in from the start.

Trusted data goes further: it’s verifiable, sourced through transparent channels, with a clear audit trail and an ethical basis for the way it is being used. You can have perfectly formatted data that was collected without proper consent or that was originally gathered for a completely different purpose. Cleaned up, it still looks fine. But it carries risks that no amount of standardisation can remove.

The question isn’t just “is this usable?” It is “is this appropriate for what we’re building?”

The data bias problem starts earlier than you think

A lot of the conversation around AI bias focuses on the model itself — on fine-tuning, on output testing and on fairness metrics. And those things matter. But bias is often introduced much earlier, at the data collection stage, and it is harder to fix after the fact.

If your training data over-represents certain demographics, geographies or time periods, your model will reflect that. If it was collected during an unusual period – such as a global pandemic or a period of economic disruption – it may not generalise well to normal conditions. Think of it like a restaurant that only sources ingredients from one small region: the menu might be excellent, but it won’t represent the full range of what’s out there and it will be brittle when that one supplier has a bad season.

This is why provenance and representativeness need to be considered together. Understanding where data came from helps you understand what it might be missing — and whether those gaps matter for the task at hand.

Asking the right questions before you build

Good data governance means asking harder questions at the start of a project, not after something goes wrong. Before feeding a dataset into a model, it is worth working through a few fundamentals:

•        Is the origin verified? Was this data acquired through transparent, documented channels?

•        Is it fit for this specific purpose? Data collected for one use case doesn’t automatically transfer to another. Consent and intended use both matter.

•        Is it still current? Data has a shelf life, just like produce. A model trained on population data from five years ago may produce conclusions that no longer hold — and stale data, like stale ingredients, can quietly ruin the final dish.

•        Could the people behind the data see this outcome? It is a useful sanity check. If the answer gives you pause, that’s worth paying attention to.

Why data provenance matters more as AI scales up

There is a compounding effect here. The larger the model, the more data it needs and the harder it becomes to maintain a clear audit trail across all of it. That is a problem that doesn’t get easier over time — it gets harder.

Organisations that invest in data provenance early are building something genuinely valuable: the ability to explain their models. Explainability is increasingly a regulatory expectation, particularly in public sector contexts and increasingly a commercial differentiator too. People and institutions want to work with AI systems they can trust and trust requires transparency about what went in.

The UK Government’s Data Quality Framework, GDPR and sector-specific governance standards all push in the same direction: know your data, document it, and be able to demonstrate that it was ethically sourced and appropriate for the purpose.

Final thoughts

Building AI on poorly understood data isn’t just a technical risk. It is a credibility risk. The farm-to-table movement taught the food industry that people care deeply about where things come from — not just how they are presented. The same shift is happening in AI. The organisations getting this right aren’t necessarily those with the biggest datasets — they are the ones who can clearly account for what they have, where it came from and why it’s appropriate for the job.

Great chefs don’t just cook well. They know their supply chain. That is what data provenance is really about.

Whether you are managing a government department, a contact centre or a frontline service team, one thing is true: gut instinct alone isn't enough anymore. Here is why operational analytics is fast becoming non-negotiable - and what it looks like in practice.

The gap between data and decisions

Most organisations today are not short of data. They have performance reports, spreadsheets, case management systems, CRMs and dashboards. Quite often more of them than anyone can keep track of. 

The real problem is that very little of this data reaches the people making day-to-day operational decisions, in a format they can act on, at the time they need it.

This gap - between data that exists and decisions that are genuinely informed by it - is where operational performance suffers. Teams default to instinct. And leaders wait for the monthly report. Problems that could have been spotted on Tuesday aren't identified until the end-of-quarter review.

Data-driven decision making isn't just a buzzword. It is a practical discipline: giving operations teams the right data, in the right format, at the right time - so that decisions are grounded in evidence, not assumption.

What is operational analytics?

Operational analytics is the application of data analysis to the day-to-day running of an organisation. It focuses on the data that drives performance right now - not just historical trends or strategic forecasts, but the metrics that determine whether services are functioning as they should, today.

This includes things like:

• How long are customers or citizens waiting for a response?
• Where are the bottlenecks in a process?
• Which teams or regions are underperforming, and why?
• Are caseloads distributed fairly and efficiently?
• What does tomorrow look like based on current patterns?

The goal isn't data for data's sake. It is faster, more confident decisions, made by the people closest to the work.

Why it matters in the public sector

For government departments and public services, the stakes around operational decision making are particularly high. Poor data leads to misallocated resources, delayed services and ultimately, worse outcomes for citizens.

Across central and local government, health, defence, justice and beyond, operations teams are often working with fragmented data sources, manual reporting processes and limited analytical capability. The result is a reliance on lagging indicators: by the time the problem is visible in the data, it has already had an impact.

Initiatives like the UK Government's Data Quality Framework and the DDaT profession have placed renewed emphasis on ensuring that data is not just collected but used effectively. Yet the leap from 'better data' to 'better decisions' still requires investment in the tools, skills and culture that make operational analytics possible.

The good news is that this is increasingly achievable - even within the constraints of legacy infrastructure and stretched teams.

Why it matters in the private sector

In commercial organisations, the pressure is different but equally compelling. Competition, customer expectations and the pace of change mean that operational inefficiency is costly and highly visible.

Whether it is a logistics team tracking delivery performance, a financial services firm monitoring risk indicators, or a retailer managing stock across sites, operational analytics provides the situational awareness that leaders need to act quickly and accurately.

The companies pulling ahead are not necessarily those with the biggest data teams. They are the ones that have embedded data into the rhythm of operations - where every team lead has access to clear, current performance data and knows how to use it.

What good looks like: from reporting to insight

There is an important distinction between operational reporting and operational analytics. Reporting tells you what happened. Analytics tells you what it means and often, what to do next.

From manual reporting to automated insight

Many operations teams are still spending significant time producing reports manually: pulling data from multiple systems, reformatting it in spreadsheets and distributing it by email. This is slow, error-prone and backwards-looking.

Automating this process - through performance dashboards, scheduled data pipelines and self-service analytics tools - frees up time and shifts the focus from compiling data to acting on it. In government settings, this can dramatically reduce the burden of management information production while improving its accuracy and timeliness.

From dashboards to decision support

A well-designed performance analytics dashboard doesn't just display metrics. It surfaces the right information to the right people, highlights anomalies and supports the decisions that need to be made at that level of the organisation.

For a frontline manager, this might mean a daily view of team workload and outstanding cases. For a senior leader, it might mean a strategic overview with drill-down capability. The key is designing analytics around the decisions that need to be made, rather than around the data that happens to be available.

From hindsight to foresight

The most mature form of operational analytics is predictive: using historical patterns to anticipate what's coming and enabling teams to respond proactively rather than reactively.

In public services, this could mean predicting demand spikes in call centres, identifying cases likely to escalate or forecasting resource gaps before they become crises. In the private sector, it might mean anticipating churn, optimising stock levels or modelling the impact of operational changes before they're implemented.

Predictive analytics for operations is no longer reserved for organisations with large data science teams. With the right data foundations and tooling, it is becoming increasingly accessible and increasingly expected.

How we apply this at Butterfly Data

Within Butterfly Data, operational analytics is not just something we advise clients on, but it is how we run our own business.

Internally, we use Collide Hub, our self-built analytics platform, to bring together operational, delivery and commercial data into a single environment. Like many organisations, we previously had data spread across multiple tools: project tracking systems, CRM records, financial data and internal reporting spreadsheets. Individually these sources were useful, but they did not always provide a clear operational picture in real time.

Collide Hub allows us to combine these data sources and surface the metrics that matter most to the team running the work day to day.

For example, we use it to track:

• project delivery performance and utilisation across teams
• emerging delivery risks before they impact timelines
• employee engagement levels
• operational and departmental budgets
• internal operational capacity and workload balance

Instead of waiting for end-of-month reporting, team leads can see the current state of delivery and make adjustments early - reallocating resources, addressing bottlenecks or prioritising work based on real data.

One of the most important design principles behind Collide is that analytics should support decisions, not overwhelm people with metrics. Dashboards are built around the questions teams need to answer: Where are we today? What needs attention? What is likely to happen next?

Using our own platform internally also allows us to continuously refine how operational analytics tools should work in practice.

Common barriers and how to address them

Despite the clear value, many organisations struggle to embed data-driven decision making into operational practice. The barriers tend to cluster around three areas:

Data quality: Analytics is only as good as the data behind it. If frontline teams are entering inconsistent or incomplete data, the insights generated will be unreliable. Improving data quality at source - through better systems, clearer standards like those of DAMA, and cultural change - is a prerequisite for effective operational analytics.

Tooling and access: Operations teams often lack access to the analytical tools they need or the tools they have require specialist skills to use. Investing in accessible, well-designed dashboards and self-service analytics removes this friction and puts insight directly into the hands of decision makers.

Culture and capability: Technology alone is not enough. Organisations that succeed with data-driven decision making invest in building data literacy at every level - not just amongst analysts, but amongst the managers and leaders who need to interpret and act on the data.

A note on AI and what it requires

There is a lot of enthusiasm right now around AI-powered operational tools - and rightly so. From intelligent scheduling to automated anomaly detection, the potential is significant.

But AI tools are only as effective as the data foundations beneath them. An AI model trained on incomplete, inconsistent or poorly governed data will produce unreliable outputs - and in operational contexts, unreliable outputs can have real consequences.

Practical AI adoption for operations teams starts with getting the basics right: clean, well-structured, well-governed data that reflects reality accurately. For organisations that invest in their data foundations now, the path to meaningful AI-enabled operations becomes considerably shorter.

This is why at Butterfly Data, we talk about data readiness before AI readiness. The two are inseparable.

Getting started: five questions to ask your operations team

If you are not sure where your organisation stands, these five questions are a useful starting point:

• How long does it take to produce your standard operational reports, and is that time well spent?
• Do your frontline managers have access to real-time or near-real-time performance data?
• When a problem emerges, how quickly can you identify the root cause using data?
• Are your operational decisions based on current data, or last month's report?
• Do your teams trust the data they're working with?

If the answers are uncomfortable, you are not alone. Most organisations have significant room to improve and significant value to unlock by doing so.

The bottom line

Data-driven decision making is not just about having the most sophisticated technology or the largest analytics team. It is about building the conditions in which operations teams can make better decisions, more quickly, with greater confidence.

For public sector organisations, that can mean less manual reporting, better visibility of frontline performance and the ability to respond to demand before it becomes a crisis. For private sector businesses, it can mean sharper operational insight, faster course correction, and a meaningful competitive edge.

The organisations that invest in operational analytics now - in the right tools, the right data and the right culture - will be far better placed for whatever comes next, including the AI-enabled future that's already starting to take shape.

Need support?

We work with public and private sector organisations to build the data foundations, analytical tools and operational insights that make data-driven decision making a reality. If you would like to explore what that could look like for your team, get in touch for a free discovery call.

‍

‍

FAQs: Data-driven decision making and operational analytics

What is data-driven decision making?

Data-driven decision making is the practice of using accurate, timely data,  rather than intuition or assumption, as the primary basis for operational and strategic decisions. It requires the right data infrastructure, analytical tools, and organisational culture to be effective.

What is operational analytics?

Operational analytics refers to the use of data analysis techniques applied to the day-to-day running of an organisation. It focuses on real-time or near-real-time performance data to help teams monitor progress, identify problems and make faster, more informed decisions.

Why do public sector organisations need operational analytics?

Public sector organisations often rely on lagging indicators and manual reporting processes, which limit their ability to respond quickly to service pressures. Operational analytics enables departments to monitor performance in real time, reduce manual reporting burden, allocate resources more effectively, and improve outcomes for citizens.

What is the difference between operational reporting and operational analytics?

Operational reporting tells you what has happened. It records historical data in a structured format. Operational analytics goes further, helping organisations understand what the data means, identify trends and anomalies, and, in more advanced applications, predict what is likely to happen next.

How does data quality affect operational decision making?

Poor data quality directly undermines the reliability of operational decisions. If the data entering a system is incomplete, inconsistent or inaccurate, any analysis built on it will be similarly flawed. Investing in data quality at the point of capture is therefore a prerequisite for effective analytics.

What do organisations need to do before adopting AI in operations?

Before deploying AI tools in operational settings, organisations need to ensure their data is clean, well-structured, consistently governed and representative of the processes they want to improve. Strong data foundations are the single most important enabler of practical AI adoption.

‍