AI Already Took a Data Analyst's Job. Here's What Actually Happened.

The Email That Arrived Mid-Interview

Ivan Ureña-Valdes was interviewing a job candidate for Block when the email from Jack Dorsey landed. A coworker messaged him: "Hey, are you okay?" From those four words, he already knew. He had to stop the interview and tell the candidate: "I was just let go from the company. I probably won't be able to submit your feedback in time."

Ivan was a Business Intelligence Analyst at Square, Block's payments arm. He had survived three previous rounds of layoffs. He was not underperforming — he was mid-way through the two largest projects of his career. He had embraced AI tools enthusiastically, exactly as Dorsey had pushed the company to do. And in February 2026, Block cut more than 4,000 employees and cited AI explicitly as the reason.

What makes Ivan's story more useful than a warning is that he had watched it happening. "I could see in my own work very quickly how much of it was already being automated," he told Business Insider. "So much of the data analyst world is finding the right dataset, writing something that will allow you to pull the data set that you want, and then generating output. Every single one of those steps is significantly faster and easier because of AI."

He was not replaced by a careless technology. He was replaced because the layer of his work that AI could automate happened to be the layer his employer valued most visibly. That distinction — between what AI can automate and what it still cannot — is the only thing that actually matters for your career right now. Not the projections, not the hot takes. The honest picture is more specific than any of those, and it points toward moves you can make this week.

What AI Is Actually Eating — And Where It Still Fails

AI has taken a real and measurable bite out of traditional analyst work. According to Kissmetrics' 2026 landscape analysis, AI has automated roughly 30–40% of traditional analyst tasks — specifically SQL writing, data cleaning, and standard reporting. These are the tasks that fill a calendar. They are not the tasks that appear on a business's profit and loss statement.

To understand what that boundary looks like in practice, consider what happened when a pseudonymous working analyst called "Data Mind" spent six weeks in early 2026 trying to replace their entire workflow with AI agents. The experiment had two phases, and they could not have been more different.

Phase 1 was data cleaning and merging. The AI succeeded instantly. Data Mind's internal reaction: "This is the beginning of the end of my career." If AI could demolish the most time-consuming foundational work in seconds, what was left?

Phase 2 was building DAX measures and a date table in Power BI — tasks that feel mechanical but require understanding the business logic behind the numbers. The AI spent two hours producing broken logic. Data Mind completed the same task in ten minutes. The verdict: "The parts of my job I wanted to automate were automated perfectly. The parts I thought would be easy for AI were the hardest." The dashboard they ultimately built, using AI for cleaning and their own judgment for everything else, was better than anything they had built alone.

AI is not going to replace me. An analyst who uses AI will.
— Data Mind, Data Analyst

Gartner's May 2026 analysis reinforces this from the organizational side: AI-driven layoffs "may create budget room but do not deliver returns." A Towards AI case study documented three companies that attempted full analyst replacement with AI agents. All three experienced business errors or had to rehire senior analysts. The failure point in every case was identical — AI lacked the domain context to interpret what its own outputs meant.

The threat is real and bounded simultaneously. If your primary value to your employer is the mechanical layer — running queries, generating standard reports, maintaining dashboards — that value is under direct pressure. If your primary value is the interpretive layer — knowing which question to ask before running any query, explaining to a stakeholder why a number is moving — that value is currently AI-proof.

Run a quick mental audit of last week's work. How many hours were mechanical output versus contextual interpretation? That ratio is your personal threat score, more useful than any industry projection.

This applies beyond data analytics in the narrow sense. Marketing analysts building weekly performance reports, HR analysts running headcount dashboards, finance analysts generating variance summaries — the same 30–40% automatable layer exists in any role where finding the data and generating the output is the visible deliverable.

Same Disruption, Two Outcomes

Ivan's own diagnosis, delivered in plain language to Business Insider, is precise: his daily work was "finding the right dataset, writing something that will allow you to pull the data set that you want, and then generating output." That is a description of the automatable 30–40%. He was good at it. He embraced the AI tools that eventually did it faster. His employer, facing cuts, saw a layer that could be consolidated. This is not a moral failing — it is a structural exposure.

Now consider Kedeisha Bryan. Before data analytics, she was a US Navy Aviation Electronics Technician managing a team of 15 technicians, then spent time delivering pizzas before pivoting into the field. She has since coached more than 15,000 students into data roles. Her coaching thesis, stated plainly on LinkedIn: "AI was meant to kill data analyst jobs. But the data tells a very different story."

The analysts she sees landing and advancing are not the ones who outrun AI on technical execution speed. They are the ones competing on a different plane entirely — translating data into decisions that a non-technical stakeholder will actually act on. Her formulation of the durable skill is not "which tool" but "which question to ask before touching any tool."

AI was meant to kill data analyst jobs. But the data tells a very different story.
— Kedeisha Bryan, Data Analytics Career Coach

Ivan and Kedeisha faced the same technological disruption. The variable is positioning: Ivan's visible value was in the automated layer; the survivors' value is in the interpretive and communicative layer that AI still fails at — the same layer that broke down in Data Mind's Power BI experiment.

The market has already priced this in. PwC's 2025 Global AI Jobs Barometer analyzed nearly a billion job advertisements across six continents and found that workers with AI skills now command a 56% wage premium — up from 25% just the prior year. Wages in AI-exposed industries are rising twice as fast as in industries least touched by the technology. The market is not telling analysts to simply "learn AI." It is repricing the role in real time based on whether the analyst can orchestrate AI toward a business outcome, not just run it.

This bifurcation is happening in business intelligence, marketing analytics, HR analytics, and financial reporting — anywhere AI can now generate the output layer. The durable differentiator in every case is the same: domain expertise deep enough to know when the AI is wrong, and communication skill sufficient to make a decision-maker act on what is right.

The Part the Optimists Are Quietly Ignoring

The augmentation argument is real. But it coexists with a market that has quietly removed 35% of entry-level analyst postings — and that removal has consequences running up the entire career ladder.

Revelio Labs data, cited by the World Economic Forum in March 2026, found that entry-level job postings in the US have fallen 35% in the 18 months leading into early 2026, "in large part because of AI." CNBC framed the implication starkly in September 2025: "AI isn't just ending entry-level jobs. It's ending the career ladder." ADP Research confirmed the demographic effect: employment for workers aged 22 to 25 in high-AI-exposure jobs fell 6% between late 2022 and July 2025, while employment for older workers in the same categories remained stable.

Indeed's January 2026 labor market data put numbers on what this looks like in the data field specifically — a 13% year-over-year drop in data and analytics job postings. At the same time, nearly 45% of remaining data and analytics postings now contain AI-related terms. The market is contracting for purely traditional roles while the surviving roles require AI proficiency from day one.

Lis Cooper, a 30-year-old data analyst in Melbourne, quit their tech job in early 2026 after company leaders announced they were rebuilding the data warehouse to be "optimized for AI." Cooper asked directly: "We are data analysts. How does that fit?" The answer — AI would "spin up the graphs" — was sufficient. Cooper sold their house to eliminate a mortgage they could no longer afford to carry into an uncertain job market. Their observation on what has changed: "It was so easy to leave a company and find a role. My colleagues and I would have companies messaging us all the time, asking to poach us. It is not like that anymore."

The WEF observation is acute: when entry-level work is automated, the work doesn't disappear — it gets pushed upward onto mid-career and senior staff, who absorb it while also being expected to deliver strategic insight. Mid-career analysts who think "I'll be fine because I'm experienced" may be underestimating how much invisible junior work they are about to absorb, and how little runway they have before that overextension becomes burnout.

The entry-level collapse is not unique to data analytics. The same pattern is hitting marketing analytics, business intelligence, financial reporting, and HR analytics. In every field where AI can handle the baseline output layer, the junior role that once trained people to eventually do senior work is thinning.

What to Do Before Your Employer Does the Audit for You

Ivan told Business Insider he remains "more pessimistic about the industry as a whole than many people probably are." That pessimism is worth holding — not as paralysis, but as honest navigation fuel. He watched the automated layer consume his role in real time. He named it clearly. That clarity is the thing to borrow from his story.

The analysts who are advancing right now are not the ones who escaped the automation wave. They are the ones who identified their own automatable layer before their employer did — and deliberately shifted their visible value to the interpretive and communicative work that Gartner, three failed corporate replacement attempts, and one broken Power BI date table all confirm AI still cannot do.

This week's move: spend 30 minutes listing every recurring task in your current role, then mark each one as mechanical output (data pull, standard report, routine query) or contextual interpretation (stakeholder question, anomaly explanation, decision framing). Calculate your personal ratio. That ratio is more informative than any industry projection about where your specific exposure sits — and it tells you exactly which skills to build next, in priority order.

The analysts who are fine are not the ones who saw AI coming. They are the ones who ran the audit one week before everyone else.

Recommended Tools & Resources