The Analyst Role Is Changing—And That’s a Good Thing
If AI takes the busywork, what’s left might be the best part of the job.
I’ll never forget the first month I spent learning SQL. I took a three-hour SQL course and then spent weeks working on what is STILL—years later—the hardest SQL I’ve ever had to write. There was much deep sighing, face desking and victorious cheering when I learned how to solve another complex problem (time series imputation with window and lag functions?!)
For years I wrote a lot of SQL, and R, and fiddled in Tableau…and even built reporting in spreadsheets. Today I rarely take the time to write queries from scratch. In part because I lead product teams building data & AI platforms—so pulling data isn’t a core job function anymore—but also because it’s just not efficient to write them manually.
Today when I need data, I conversationally work with ChatGPT on syntax, drop in screenshots from the data catalog, manually add context to revise, and eventually land on a precise query. I can even ask for help optimizing the query for efficiency if it’s scanning a lot of data and I want it to run faster/cheaper. It’s all done in far less time and it feels like magic.
As a data product leader I’ve spent a lot of time thinking in recent years about how the way businesses use data will change because of AI. How the systems, processes, and people might evolve.
For analysts, whose value has historically come from technical execution and niche expertise, I’ve observed that this shift can feel threatening. If AI can pull context from the catalog, generate SQL, build dashboards, and even identify trends, will the business even need analyst specialists? If your days have been filled with hunting down data, piecing together missing context, holding knowledge of known data issues in your head, and fine-tuning queries and reports — what happens when agents can do lots of these steps?
In my opinion - the job gets a lot more interesting, more strategic, and more impactful. Most analysts are so busy just gathering and preparing accurate data they have little time to do more than descriptive statistics and basic reports.
Diagnostic analytics - rarely.
Predictive analytics - no time for it.
Prescriptive analytics - can’t remember the last time I got to work on those.
But what if suddenly you were working with an analyst co-pilot and could knock out the descriptive analytics work in 10% of the time?
Maybe you could spend more time with the business on:
Precision metrics: What’s the *right* north start metric? Do all the stakeholders who care about this agree on what it should mean? Is our current calculation a good match for the business intent?
Diagnostic deep dives: Remember how you spent months building that dashboard and as soon as you showed the prototype to the business they were happy, but also had a dozen new question?
Hmmm, that’s surprising - can we figure out why there’s a spike in December in the US?
This location is an outlier - I’m really surprised with how low those satisfaction scores, what’s going on there?
I wonder how these numbers would look different if we sliced these numbers by X (sales region, rep, item, month, etc)
Predictive analytics: How might things change if we try scenario A vs scenario B?
That’s interesting - if we double our ad spend here I wonder how that will impact our funnel
Shoot - this supplier has been back ordered for a while, if we add a second supplier how will that impact output?
User churn has been spiking, if we added feature X to the silver tier how might that affect churn?
Prescriptive analytics: Rather than waiting for the business to ask questions or for new forecasts, what if we proactively mined for insights and made recommendations? What if we created agents that proactively scanned and brought those opportunities to us? What if we operationalized insights into push notifications that nudged humans take different actions? What if we trusted those discoveries enough to take action automatically?
Doesn’t this sound better? Like super fun? I write about it and it makes me miss being an analyst. But in my experience, many analysts aren’t seeing this. They feel instead a vague threat from AI, from change, from being asked to completely upend the way they work and move out of their comfort zone.
So data product managers have a role to play helping convey this vision for how things could be. And helping analysts and their managers move in that direction. It requires telling the same story over and over, but in new ways that resonate and call out assumptions (conscious and unconscious).
Below are a few myths, false beliefs I think a lot of us in data hold - whether we realize it or not. And some reframes for how we might bust those myths for good and move toward a much more exciting future as data teams and companies.
Myth #1: “SQL is my superpower.”
Then:
You were the go-to person for “figuring out the data.” You knew which tables were reliable (and which ones weren’t), how to join them, and how to pull exactly the right slice for someone’s weekly report or executive slide deck. You were fast, accurate, and indispensable.
Now:
AI is getting surprisingly good at generating SQL. Tools like ChatGPT, dbt Cloud’s AI assistant, and natural language querying in modern BI platforms reduce the need to start from scratch.
But that doesn’t mean SQL doesn’t matter anymore. It just means writing it from memory is no longer the differentiator. Your edge is knowing what to ask for, how to validate it, and how it fits into the broader business context.
Soon:
Analysts might not be writing raw SQL every day. Instead, they’ll spend more time framing metrics, defining data models, and guiding teams on how to interpret results. AI will handle the “how”; analysts will own the “what” and “why.”
Reframe: “Consulting the business on how to build the right metrics & models is my superpower”
Myth #2: “I’m the reporting expert.”
Then:
You built dashboards, updated KPIs, and kept the reporting machine running. You were the Looker/Tableau/Power BI expert, the person who could wrangle requests from every function into a clean visual.
Now:
AI is starting to generate entire dashboards with just a prompt. Tools are getting better at automatically suggesting visualizations, detecting anomalies, and even generating commentary. Are they ready for prime time? Largely no, IMO—but they will be soon.
And even when that time comes, someone still needs work with the business to understand the need and collaborate with AI to design those dashboards (and review builds). In fact, now would be a great time to finally study up on UX/UI and data storytelling skills - going beyond just getting the dashboard up and running and spend more time designing something intuitive and sticky.
The value is shifting away from building reports and toward designing, interpreting, explaining, and iterating on them until they become sticky, useful, and easy to use. Fewer dashboards, well built.
Soon:
It won’t be enough to build exactly what the business asks for. We need to build analytics that drive impact, which means making them usable, flexible, and improving them over time.
Reframe: “I’m the expert in creating intuitive, sticky, data experiences that the business finds delightful to use.”
Myth #3: “My job is to answer questions.”
Then:
You worked in reactive mode. Someone in marketing, ops, or product would ping you with a question. You’d go off and come back with a number, a chart, or a dashboard.
Now:
You’re probably still answering a lot of those questions. But you’re also starting to notice patterns. You’re pushing back on questions that are too vague or poorly framed. You’re asking, “What decision will this inform?”
You’re starting to shift from being a data server to a data partner - someone who helps shape the question, not just answer it.
Soon:
The most impactful analysts won’t just answer questions—they’ll help shape them. They’ll work upstream with product, strategy, and leadership to:
Define success metrics
Design experiments
Set the right expectations around measurement
You’ll be the person helping the business get clearer about what it actually wants to know. And you’ll be helping the data product managers and engineers build better data products that enable self-service and reduce ad-hoc data requests over time.
Reframe: “My job is to help the business get the answers it needs today and make it easier to get the answers it needs in the future too.”
Doesn’t this sound exciting? Analyst jobs will get much more interesting with AI in the picture, IF and only if they take charge of reframing their role amid change. They can still be indispensable to the business, even more so, if they focus on adding value in strategic ways.
What AI can’t do (yet)
None of us can predict where AI goes in the long term, or even in the medium term, but it seems clear that for the foreseeable future even the best AI tools can’t:
Navigate ambiguity and proactively clarify vague business questions
Understand organizational context, goals, and tradeoffs
Decide which metrics matter and why
Communicate insights clearly and credibly
Influence decisions across functions
These are the skills that set great analysts apart—and they’re only becoming more important. In fact, as AI handles more of the tactical work, these strategic and interpersonal skills become even more important.
Additionally, the AI powered self-service experiences vendors are demoing today (Snowflake intelligence anyone?) only work with data teams working behind the scenes to build the context, data products, and metric definitions that allow for trustworthy, authoritative answers. The interface and natural language capabilities are ready, the data models and business context organization largely are not. Analysts can play a key role in facilitating the roll-out of powerful AI-enabled analytics experiences, while at the same time focusing on the things they uniquely can provide.
It’s not about replacing analysts. It’s about redefining the role so it’s less about execution and more about thinking, guiding, and communicating.
That’s good for analysts—and even better for the organizations that depend on them.
So what now?
If you’re an analyst, ask yourself:
What parts of my work are repeatable or automatable?
What skills do I bring that AI can’t (yet)?
How can I work more closely with stakeholders, earlier in the process?
If you’re leading a team, ask:
Are we measuring analysts on deliverables or on impact?
Are we giving them space to ask questions, not just answer them?
Are we rewarding strategic thinking—or speed?
AI will change analytics work. That much is inevitable. But how it changes it—and whether that change is good or bad—depends a lot on how we respond.