You put a lot of work into the data strategy, working all hours around the day job to get it done. Then it gets ignored, or criticised, and quietly left on the shelf. I have seen plenty of data strategy failures over the years, and most of them come from a handful of avoidable causes. This guide pulls together what I have learned about writing a data strategy that the business actually acts on.

At its heart, a data strategy is a vision of where data needs to be in your organisation, an honest picture of where it is today, and a costed plan to close the gap.

1. Start with the business strategy

The single biggest reason data strategies fail is that they are not seen as relevant to the business. If the recommendations don’t logically follow the executive’s understanding of the business and its plan, the strategy looks self-serving and its investments look discretionary.

“But we don’t have a business strategy” is something I hear a lot. In my career I have never found that to be true. There may not be a published document, but that is not the same as having no strategy. You can build the business context from:

• the organisation’s mission, or simply what it does, who it serves and how it is funded
• the CEO’s priorities, set out in reports to shareholders and communications to employees
• executives’ messages: HR on engagement, marketing on customers and positioning, and so on

You won’t see future acquisitions or new products in development, but that is a small part of the picture. More in IT or Data Strategy blocked by missing Business Strategy.

2. Make the vision compelling

What makes a data vision compelling is its link to the business strategy, and therefore the imperative to achieve it. Executives are passionate about the business strategy, and anything that enables it is likely to get their backing. That link might be analytics that enable new services or a step change in efficiency, or risk avoidance through data governance.

A compelling strategy needs three things:

1. A vision tied to the business strategy.
2. A full assessment of the current state, so data professionals believe the vision is achievable.
3. A multi-year strategic plan, clear on the outcomes for the business, how long they take, and the investment required.

More in A Compelling Data Strategy.

3. Balance all four concerns

Some parts of data are more exciting than others, and that excitement can lead us to focus on one area. A data strategy must consider all four:

• Data analytics and data assets
• Data platforms and tools
• Data governance and curation
• People, organisation and skills

Focusing on one disenfranchises everyone who cares about the others, and puts the plan at risk. We can’t deliver new analytics without the right platform, more data is no use unless it is governed and curated, and only an organisation with skilled data citizens can take full advantage of data. More in Data Strategy Fails.

4. Cover the whole organisation

Data teams are often embedded in the high data value areas, which biases the strategy towards them and can leave the rest of the business behind, or create competing data strategies. Large federated organisations may need a formal federation model that is clear about who is responsible for what. Some areas will get less attention in the plan, but that should be communicated as a priority call, not an oversight.

How do you know you have found everything? Organisation charts are a poor guide, because hidden pieces of business process sit in unexpected teams. My best friend here is a capability model: a high-level functional model of the business, perhaps a few days’ work and a few dozen functions. Map your data subjects and assets to it. Every function produces data, so a function with no data mapped to it is a gap you must either fill or justify. More in Unknown Unknowns in Data and IT Strategy.

5. Treat governance as an enabler

We tend to think of governance as red tape, and in data that misses the point. I liken governance to climbing pitons: they stop you falling far, for a small cost. There are old climbers and bold climbers, but no old, bold climbers.

Governance also enables data use. Stewards manage data quality and help analysts get up to speed on unfamiliar datasets. And curation makes data like a good museum collection: catalogued, described, with guides for every level of visitor, from the data tourist to the data scientist extending the descriptions. More in Data Governance at the Data Museum.

6. Put people at the heart

The people dimension is often left out, or squeezed in at the end. It has two audiences.

The data team. Data people are in demand. Retaining them takes fulfilling roles, the right tools, and real opportunities to grow their skills and careers.

Data citizens. These are everyone who uses data to make decisions, sometimes reaching partners and customers. I think of data literacy as six skills:

1. Identify useful data, including its quality and completeness
2. Analyse data with basic statistics and simple charts
3. Explain its meaning in business context
4. Apply it to decisions
5. Tell the data narrative
6. Use it ethically, aware of bias and within company guidelines

Often the barrier isn’t skills but culture: do decisions require data, and are results measured? Data democratisation, making data easy to access, is a key enabler. More in People are at the heart of the Data Strategy.

7. Mop up the data puddles

Data puddles are small pools of data, each with a local analyst team, built for local needs. Locally they are a success. The trouble comes with use cases that span many puddles, which get bogged down joining and importing data.

The fix starts with a data catalogue, which depends on the local teams’ knowledge, then consistent technology, structures and keys so the puddles can be joined. That may lead to a lake or warehouse, but it doesn’t have to. Every step adds work for local teams with no benefit to the analytics they already deliver, so expect pushback. A top-down mandate rarely works. Find the killer use cases that the local teams and their stakeholders want, and let those drive the change. More in Mopping Up the Data Puddles.

8. Time it for the budget

The data strategy often sits on the “soon but not yet” pile until it is too late. The moment that matters is the budget cycle: the strategy provides the arguments for your budget requests, especially for skills and headcount, which get the most scrutiny.

Budget expectations are set surprisingly early. I aim to have the strategy ready by around the fifth month of the financial year, for the second-half forecast, which means starting the update at the start of the second quarter at the latest. An update should take eight to ten weeks. A first strategy in years, or one after a major restructure, needs more time, so start at the beginning of the year. More in Start Your IT or Data Strategy Today.

9. Communicate change with data

Any change of strategic direction needs three messages to bring people with you: a shared understanding of where we are today, often best shown as a clear data story; a shared picture of where we need to be; and a plan showing how the change is delivered and the benefit realised. Without them you get debate instead of action, as the UK’s 2022 mini budget showed. More in Data Must Underpin a Mini Budget.

How far data strategy has come

Data strategy hasn’t always been a discipline in its own right. For a light-hearted look at how IT and data strategy evolved from 1952 to 2022, see IT and Data Strategy through the Jubilees.

Go further

I walk through the whole process step by step, including building the business context and the capability model, in my Udemy course Getting Started on the Data Strategy. If you are writing an IT strategy too, see my companion guide, How to Write an IT Strategy Aligned to the Business.

• Enterprise Architecture