“You’ve likely seen your CEO’s excitement about AI projects promising better customer insights or streamlined operations. But whilst 99% of executives (UK-based) claim their organisations are AI-ready, you also see the reality – duplicate customer records, inconsistent supplier names and outdated contact lists.
According to Semarchy’s 2026 research, which focuses on UK organisations but reflects challenges that are universal, 29% of organisations don’t even measure data quality before feeding it into AI systems. When AI trains on messy data, it doesn’t clean it up. It amplifies the mess at scale.
Because EAs work across the organisation – booking travel, managing executive contacts, coordinating with suppliers and handling communications – you’re often the first to spot these inconsistencies.
That cross-functional visibility puts you in a unique position to prevent data issues from undermining AI projects.
Why messy data is AI’s Achilles heel
AI is powerful but it’s not magical. When your systems list ‘John Smith at ABC Ltd’ in one place and ‘J. Smith, ABC Limited’ in another, AI doesn’t know they’re the same person and company.
It treats them as separate entities, resulting in duplicate outreach, confused analytics and recommendations built on incomplete data.
The real-world consequences are significant. Imagine your CEO using AI-generated insights to make a major supplier decision, unaware that the system has counted the same supplier three times under different names.
Or an AI tool recommending outreach to a ‘new prospect’ who’s actually a long-standing client, placing a carefully nurtured relationship at risk.
Once AI is trained on poor data, fixing the problem isn’t as simple as cleaning a spreadsheet. It requires retraining models, re-running analyses and explaining to leadership why the insights they’ve been acting on are unreliable.
The solution is master data management (MDM) – creating a single source of truth by consolidating duplicates, standardising naming conventions and maintaining one authoritative version of critical data.
The EA’s unique role in data quality
Data quality issues show up in your world every day: contact records missing key details, duplicate entries for the same person scattered across different systems, inconsistent naming conventions and outdated information no one has bothered to update.
You’ve probably already dealt with the fallout – booking travel to the wrong office, sending invitations to outdated contacts or spending hours reconciling conflicting records before important meetings.
Now imagine AI making thousands of automated decisions using that same unreliable data. The small glitches you manually fix quickly become systemic failures.
This is where your strategic value comes in. Only 5% of UK organisations have appointed a dedicated head of AI, with most relying on already stretched CTOs, CIOs or even CEOs to manage AI strategy alongside everything else. There’s a leadership vacuum you’re uniquely positioned to help fill.
Practical steps EAs can take now
You don’t need to be a data engineer to improve data quality.
Start with what you already manage:
- Audit key databases: Review the contact lists, CRMs, calendars and supplier records you use. Flag duplicates, outdated entries and inconsistent names in a simple tracking sheet you can share when advocating for better systems.
- Standardise within your control: Set clear, uniform naming conventions and use a basic template to capture complete information. When you see errors or duplicates, correct or flag them rather than work around them.
- Champion MDM with leadership: When your executive discusses new AI initiatives, ask: “Do we have a single source of truth for this data?” Use concrete examples of inconsistencies you’ve seen to show how they could mislead AI. Frame MDM as essential groundwork, not admin clean-up.
- Build cross-functional data allies: Connect with other EAs across departments. Agree on informal shared standards and jointly raise systemic issues with leadership. Your collective voice carries more weight.
- Measure and share wins: Track improvements – ‘reduced duplicate supplier records by 40%’ for example – and report these to your executive. It demonstrates strategic thinking and quantifiable value.
EAs as AI enablers
As AI initiatives accelerate, many organisations are unknowingly building on unreliable data.
EAs who step in now to advocate for data quality are playing a strategic role in protecting executive decisions, safeguarding AI investments and preventing costly failures.
The organisations that succeed with AI won’t be those with the flashiest tools. They’ll be those with solid data foundations – and EAs are uniquely placed to make that foundation a reality.”
