Data quality and Microsoft Copilot: why trusted information comes first

Most businesses don't have an information problem they can point to. They have a hundred small ones.

Someone spends twenty minutes hunting for the latest version of a contract. Two departments produce monthly figures that don't agree. A customer record sits in three systems with three different phone numbers. A policy document from 2021 still turns up first in search, and nobody's sure whether it's been replaced.

None of that feels like a crisis. It just quietly slows the business down.

That's the honest starting point for any conversation about data quality. It isn't a technical topic that belongs with IT. It's an operational one that affects how quickly your people can answer a question, how confident you are in your reporting, and how consistently customers are looked after.

Poor information costs you whether or not you use AI

It's tempting to treat data quality as something you sort out before rolling out Microsoft Copilot. In practice, the cost is already being paid.

• Productivity. Time spent searching, checking and re-checking is time not spent on the work itself. In a 40-person business, even 20 minutes a day per person adds up to something you'd notice on a P&L.
• Customer service. Out-of-date contact details, duplicate records and conflicting notes lead to slow responses and repeated questions the customer has already answered.
• Reporting. When finance, sales and operations each build figures from their own source, board meetings turn into debates about which number is right rather than what to do about it.
• Decision making. Leaders either delay decisions while they verify the data, or make them on information they privately don't trust.
• Compliance. If you can't say what information you hold, where it lives or how long you keep it, responding to a data request or an audit becomes a scramble.
• Operational efficiency. Automation and integrations built on inconsistent data create rework rather than removing it.

Fixing this is worth doing on its own merits. Microsoft Copilot simply makes the case harder to ignore.

Copilot tends to reveal problems rather than create them

When a business turns on Microsoft Copilot across Microsoft 365, people start asking questions of their own information for the first time in a general way - "what did we agree with this client", "summarise the last three project updates", "pull the current pricing".

If there are four versions of the pricing document in different SharePoint sites, that shows up quickly. If a team site was opened to everyone five years ago for a one-off project and never closed, that shows up too.

The underlying issues were already there. What changes is visibility. A search box surfaces what a person happened to look for; an assistant that draws across your content surfaces what's actually in there.

That's a useful thing, not a threat. It gives you a practical reason to tackle work that's been on the list for years.

A caveat worth stating plainly: data quality isn't the only factor in how useful an answer is. How the service is configured, what it's permitted to see, and how the question is asked all matter. Clean, well-owned information gives you the best chance of a reliable answer - it isn't a guarantee on its own.

Storing information isn't the same as managing it

Most SMEs store plenty. Files sit in SharePoint, OneDrive, Teams, a CRM, a finance system, a shared drive somebody never quite retired, and a handful of inboxes.

Storing means the information exists somewhere. Managing it means:

• someone is accountable for it
• there's one version that counts
• it's labelled so people know what it is and how sensitive it is
• the right people can reach it and the wrong people can't
• it gets removed or archived when it's no longer needed

The gap between those two states is where most day-to-day frustration lives. It's also why buying more storage or another system rarely helps.

What trusted information looks like in practice

You don't need a formal data governance programme to know whether your information can be trusted. A few practical tests:

• There is one obvious place to look. People don't have to guess between three locations.
• The newest version is the one you find first. Superseded documents are archived, not left alongside the current one.
• It has an owner. A named person or role in the business - not "IT" - is responsible for keeping it right.
• The figures reconcile. Sales, finance and operations start from the same source, or the differences are understood and documented.
• Records are unique. One customer, one record, across the systems that matter.
• Sensitivity is clear. Commercially sensitive or personal information is labelled and access is limited accordingly.
• It's current. Documents have a review date and someone acts on it.

Where those things hold, reporting gets faster, onboarding gets easier and business intelligence tools produce numbers people accept.

A practical example

A 60-person distribution business we've described this pattern to had three separate places recording customer delivery preferences: the CRM, a shared spreadsheet the transport team maintained, and notes held in individual inboxes.

Nothing was obviously broken. But two or three times a month a delivery went to the wrong site or arrived outside a customer's booking window, and each one took a couple of hours of somebody's day to unpick. Monthly service reporting was built from the CRM, so those failures didn't appear in the figures - the leadership team believed performance was better than it was.

The fix wasn't clever. They agreed the CRM was the single source for delivery preferences, gave the transport supervisor ownership of that field, moved the spreadsheet content in, and retired it. Service reporting started to reflect reality, and the pattern behind the failures became visible.

That's data stewardship in its most practical form: someone owns the information, there's one version of it, and the business can see what's actually happening.

Common data quality challenges in SMEs

• Multiple versions of key documents across SharePoint, OneDrive and local drives
• Duplicate customer or supplier records across CRM and finance systems
• Inconsistent naming, so the same client appears three different ways
• Departments reporting the same metric with different definitions
• Historic permissions left open from projects that finished years ago
• No owner for shared reference material - price lists, templates, policies
• Documents with no review date and no retention rule
• Free-text fields used where a structured field should be
• Information held in personal mailboxes rather than shared locations
• Reporting rebuilt manually each month rather than drawn from a maintained source

Where to start

Start by understanding what you have, not by buying anything.

1. Pick the information that matters most. Customer records, pricing, contracts, standard operating procedures, board reporting. Ignore the rest for now.
2. Find out where it actually lives. Ask the teams, not the systems diagram. You'll usually find at least one location nobody had accounted for.
3. Name an owner for each. A business role, not IT. Ownership is the single change that makes the rest stick.
4. Agree the single source. For each type of information, decide which location or system wins, and say so out loud.
5. Deal with the duplicates. Merge, archive or delete. Leaving a second copy "just in case" defeats the exercise.
6. Review access. Who can see it, who should, and what's open that shouldn't be.
7. Set retention and review dates. Even a simple rule beats no rule.
8. Fix the reporting definitions. Write down what each headline metric means and where it comes from.

Then repeat, one information area at a time. This works far better as an ongoing operational habit than as a one-off project.

A checklist you can use this week

• Duplicate information - are there multiple copies of the documents and records people rely on?
• Out-of-date documents - can you tell which version is current, without asking someone?
• Ownership - is there a named business owner for each key information set?
• Classification - is sensitive and personal information labelled?
• Access controls - do permissions still reflect who actually needs access?
• Retention - do you know what you keep, for how long, and why?

If you can't answer five of those six confidently, that's your starting point - and it's a more valuable piece of work than any technology decision you'll make this year.

How Axon helps

As a Managed Intelligence Provider, our job goes beyond keeping systems running. We help organisations understand what information they hold, where it lives, who owns it, whether it can be trusted, and how well it supports the outcomes the business cares about.

In practice that means working through your Microsoft 365 estate and business systems to identify duplication, unclear ownership and permissions that have drifted; agreeing sensible information governance that people will actually follow; and putting the day-to-day practices in place that keep information accurate rather than letting it decay again.

The benefit shows up in ordinary things. Reporting people believe. Faster answers to customer questions. Automation that behaves predictably. Better visibility for the leadership team. And when you do adopt Microsoft Copilot, it has trusted data to work from - which is what separates a useful rollout from a disappointing one.

Successful digital transformation rarely comes down to the technology. It comes down to whether the information underneath it is well managed and well owned.

If you'd like a clear view of your current information quality, ownership and data governance - and a practical plan to improve it - get in touch with the Axon team. We'll start with what you've got, not with what we'd like to sell you.