AI for Humans: Data Foundations

In this second instalment of our series, we look at ways to strengthen your data foundations without major investment or disruption.

AI projects rely on data. But the assumption that you need perfect data to get started often delays or derails useful progress. In reality, most businesses begin with fragmented, inconsistent and underutilised datasets. The question is not whether your data is perfect. It is whether you understand what you have, how it is structured and what it can support.

This post provides an approach to enhancing your data environment without requiring a rebuild. These minor adjustments can reduce risk, increase clarity and enable more informed AI decisions over time.

Understanding Your Current Data

Before improving your data, take time to understand its current state. A simple, focused inventory will reveal more value than you might expect.

Quick Data Inventory Exercise

For the business problem you prioritised in Part 1, document the following:

What relevant data do you currently collect?Examples: customer details, sales transactions, website interactions, operational metrics, service records.
Where does this data live?Examples: CRM systems, spreadsheets, accounting software, email, surveys, paper records.
How good is the data?Ask yourself:

Is it complete? Are there gaps?
Is it consistent? Is similar data recorded in the same way across systems?
Is it current? How often is it updated?
Is it accessible? Can you retrieve and use it easily?

These questions give you a clear view of what can be used now, what needs improvement and what should be archived or retired.

Sustainability check: As you map out your current data, take note of systems or tools that are rarely used but still consuming resources. Retiring unused platforms or consolidating overlapping systems can reduce digital waste and energy usage, particularly in cloud environments.

Four Quick Wins You Can Implement This Month

Rather than trying to fix all your data issues, focus on a few changes that reduce friction and improve usability.

1. Standardise Key Fields

Challenge: Inconsistent data entry (e.g. customer names entered in different formats) makes reporting unreliable and messy.

Solution: Pick a few fields that are most important to your business (like customer name, job title, or product category). Create a short guide for how these should be entered. Then ask your team to follow the same rules.

Example: One of our clients standardised industry types across their CRM and billing system in a week, giving them clearer insight into customer trends with no new tools.

2. Implement Basic Data Validation

Challenge: Incorrect or missing information in forms or systems causes follow-up issues like failed deliveries or broken reports.

Solution: Use the built-in settings in tools like your CRM or website form to add checks (e.g., an email contains “@” or a postcode matches the UK format).

Example: A retailer we worked with added postcode validation to their checkout form and reduced failed deliveries immediately.

3. Create a Single Source of Truth for Key Metrics

Challenge: Teams often calculate the same thing in different ways, which leads to confusion, rework, or bad decisions.

Solution: Pick a few key metrics (like “monthly revenue” or “on-time delivery”) and agree on how they should be defined and calculated. Write them down and make sure everyone uses the same version.

Example: We worked with a wholesaler who standardised the “on-time delivery” metric across sales, ops, and customer service. This instantly cleared up misaligned reporting.

4. Begin Consolidating Siloed Data

Challenge: Related data often sits in separate systems (e.g. sales data in one place, feedback in another), making it hard to spot patterns.

Solution: Start simple. Export two related data sets and combine them in a spreadsheet. Even just looking at them side-by-side can reveal valuable insights.

Example: A healthcare provider we supported combined appointment scheduling and satisfaction scores into one spreadsheet. This helped them identify drop-off patterns that no single system showed on its own.

Sustainability check: Manual exports might feel low-tech, but they can highlight opportunities to reduce duplication or unnecessary data processing. Optimising how and where data is handled helps cut energy usage over time, especially in businesses relying on multiple systems running in parallel.

Simple Data Governance for Small Teams

Governance often sounds complex or results in bloated processes, but it does not need to be. These light-touch practices will help maintain your progress:

This does not guarantee clean data but it does create conditions where progress can be tracked and improvements can be sustained. Which is often enough to support your first few AI initiatives.

Sustainability check: Good data governance improves accuracy and reduces inefficiency. By improving clarity and reducing rework, your team uses less time, energy, and computational power, all small steps toward more sustainable operations.

Try This: A Two-Week Data Sprint

Here is a suggested action plan to help you make solid progress quickly.

Week 1: Assessment and Planning

Days 1 to 2: Complete your data inventory for the chosen use case
Day 3: Identify the 3 to 5 most critical data fields
Days 4 to 5: Document any issues with these fields

Week 2: Quick Wins Implementation

Days 6 to 7: Create and share basic standards for your key fields
Days 8 to 9: Add simple validation where possible
Day 10: Assign clear ownership for your critical datasets

Summary and Next Steps

Improving your data does not require an overhaul. It starts with clarity, knowing what you have, how it is being used and what you want it to support. Small, deliberate actions can remove obstacles to AI adoption and increase the return on future investment.

In the next post, we will explore low-risk ways to begin applying AI using tools that are already available and how to evaluate whether they are worth implementing.

If you would like support with strengthening your data foundations or planning your next steps with AI. Our Digital Product and AI service.