Data analysts and business analysts work at the intersection of data and decisions. The job is to find signal in noise — and then communicate it clearly enough that something actually changes. AI is particularly well-suited to accelerating the mechanical parts of this work: writing queries, cleaning data, drafting reports, and explaining findings to non-technical stakeholders. The judgment about what matters and why remains the analyst’s job.
Where AI Adds Real Value for Analysts
1. SQL Query Writing and Optimisation
Writing SQL is often the bottleneck between a business question and an answer. AI can write first-draft SQL queries from a plain-English description of what you need: “give me monthly revenue by region for the last 12 months, excluding cancelled orders, sorted by region.” You review, adjust for your schema specifics, and run it.
For complex queries — window functions, CTEs, recursive queries — AI can draft structures you’d otherwise spend an hour constructing. It can also explain existing queries you’ve inherited but don’t fully understand, suggest optimisations for slow-running queries, and help you translate between SQL dialects (MySQL to PostgreSQL, T-SQL to BigQuery).
2. Python and R Code Assistance
For analysts working in Python (pandas, numpy, matplotlib, seaborn) or R, AI functions as a capable pair programmer. Describe what you want to do with your dataframe, and AI drafts the code. Ask it to explain an error message and it explains it in plain English and suggests a fix. This is particularly valuable for analysts who are competent but not expert programmers — AI closes the gap between knowing what you want and knowing how to write it.
3. Report and Presentation Writing
The hardest part of analysis is often the last mile: turning findings into a document that a non-technical audience will actually read and act on. AI can draft executive summaries, key findings sections, and recommendation narratives from your analysis notes and data outputs. You provide the insight; AI structures it into clear, professional prose.
For business analysts writing requirements documents, process maps, or business cases, AI can draft structured documents from your raw notes — ensuring consistent formatting and coverage of standard sections.
4. Data Cleaning and Transformation Logic
Describing a data cleaning problem to AI and getting code back is often faster than Googling the right pandas function. “Remove duplicates based on customer ID keeping the most recent record”, “standardise phone numbers to NZ format”, “split a full name column into first and last name” — all of these can be coded by AI in seconds and tested in your environment.
5. Power BI and Excel AI Features
Microsoft’s Copilot features inside Power BI and Excel are increasingly capable. In Power BI, Copilot can generate DAX measures from plain-English descriptions, create report summaries, and suggest visualisations. In Excel, Copilot can write complex formulas, generate pivot table configurations, and explain what a formula does. These are available through Microsoft 365 Copilot licences — worth evaluating if your organisation already uses Microsoft’s stack.
6. Stakeholder Communication
Business analysts often spend significant time in meetings, requirements gathering sessions, and cross-functional communication. AI can help draft meeting agendas, summarise requirements from interview notes, write user stories from functional requirements, and prepare stakeholder update emails. The communication overhead of the BA role is significant — AI reduces the time spent on document production so you can spend more time on the actual analysis and facilitation work.
7. Hypothesis Generation and Exploratory Questions
When you’re starting a new analysis, AI can help you structure the problem: what hypotheses are worth testing, what data would you need, what analysis approaches are appropriate, what questions are you likely to be asked by stakeholders. This is more about using AI as a thinking partner than a code generator — and it’s often where the most value lies for senior analysts who already know how to do the work.
Will AI Replace Data Analysts?
This is the question most analysts have in the back of their mind. The honest answer: AI is replacing the most routine, mechanical parts of analysis — the query-writing, the data cleaning, the report formatting. These were never the highest-value parts of the job.
What AI can’t do: understand the business context that makes one finding important and another irrelevant, build the stakeholder relationships that make analysis actionable, exercise judgment about data quality problems that will invalidate an analysis, or take accountability for a recommendation that turns out to be wrong.
The analysts most at risk are those doing purely mechanical work with little business judgment involved. The analysts best positioned are those who use AI to accelerate the mechanical work and reinvest the time in the higher-value activities: deeper analysis, better stakeholder communication, more strategic questions.
Data Privacy for NZ Analysts
Analysts regularly work with data that includes personal information — customer records, employee data, health data, financial records. Under the NZ Privacy Act 2020, this data must be handled appropriately. Key considerations when using AI with data:
- Never paste real personal data into consumer AI tools (ChatGPT, Claude.ai). Use anonymised or synthetic data for testing AI-assisted queries and code.
- Generate synthetic test data with AI — ask it to create a realistic-looking dataset with the same schema as your real data. Test your queries against synthetic data, then run against real data in your secure environment.
- Check your organisation’s AI use policy before using AI tools with any business data. Many NZ organisations are still developing these policies — here’s how to write one.
- Enterprise agreements matter: GitHub Copilot for Business, Microsoft Copilot with a commercial licence, and similar enterprise tools have data processing terms that differ from consumer tools. Understand what you’re using before using it with real data.
Getting Started
The fastest win: take your next SQL query or data cleaning task and write a plain-English description of what you need before you start coding. Paste it into an AI tool and see what it produces. Compare it to what you’d have written. See how much time it saves and where it needs correction.
Most analysts find the first few interactions require significant correction — the AI doesn’t know your schema, your data quirks, or your business context. After a few sessions where you correct and refine, you develop a sense of how to prompt effectively for your specific environment, and the value compounds.
For a structured approach to AI capability across your analytics team — covering tools, workflows, data governance, and training — an AI Assessment provides a practical roadmap tailored to how NZ data teams actually work.
Frequently Asked Questions
Can AI write SQL for me?
Yes — AI can write first-draft SQL from plain-English descriptions. It doesn’t know your specific table names and column names unless you tell it, so you’ll need to provide your schema or correct the output. For standard query patterns (joins, aggregations, window functions), it’s often accurate enough to run with minor edits. For complex business logic specific to your environment, it provides a useful starting structure that needs refinement.
Is GitHub Copilot worth it for analysts?
For analysts working regularly in Python or R, GitHub Copilot (or similar code completion tools) can materially speed up coding. It’s most useful for analysts who are competent but not expert programmers — it closes the gap on syntax and library usage. For analysts who primarily work in SQL and Excel, the Microsoft Copilot features inside Power BI and Excel are more relevant.
Can I use AI to analyse my data directly?
Some AI tools (ChatGPT’s Advanced Data Analysis, Claude’s analysis features) allow you to upload data files and perform analysis directly. These can be useful for quick exploratory work on non-sensitive datasets. For production analysis involving personal data, sensitive business data, or data governed by your organisation’s policies, you should use AI to write the code and run it in your secure environment rather than uploading raw data to external AI tools.
