The Job Description Is Ten Years Behind the Job

It happens after the session ends, after the chief has left the room and the chairs are getting stacked. Someone hangs back — usually the analyst who asked the sharpest question of the day — and waits until it's just the two of us. Then they ask some version of the same thing.

"Do you think I'll still have this job in five years?"

I've been asked that in enough agencies now that I've stopped treating it as a nervous aside. It's the real question. And I understand where it comes from, because I have sat in that chair. The one always evolving because the work, the budget, the mission changed. Every one of those transitions arrived with a technology attached to it and a rumor that the technology was coming for the work.

Here's what I tell them, and it is not reassurance. The role isn't disappearing. It's splitting. And the version of the job that survives is not the one most agencies are currently advertising for.

The wrong question, asked in good faith

Every time a new technology arrives, someone asks whether it will eliminate their job. It's a fair instinct. It's also the wrong frame, because roles rarely vanish outright. They get reshaped by forces that act on different parts of the same job at the same time.

For crime analysis, the forces that created the role are still running. Policing generates more data every year, not less. Compliance obligations — traffic-stop reporting, use-of-force data, consent decrees — still require someone who can produce, verify, and defend the numbers. Communities still expect their agencies to explain what's happening and why. None of that is going away.

But the forces that erode roles are running too, and they are landing on one specific layer of this job: the routine reporting, the standard mapping, the recurring statistics produced days after the fact. That layer is being absorbed into software. It is the substitutable part.

So the honest answer to the analyst in the empty room is not "you're fine." It's "half of what you do is being automated, and the other half is about to be worth more than it has ever been worth — and whether you get to move into that half depends largely on decisions your leadership is making right now, mostly without realizing they're making them."

That's the part I want leaders to sit with.

What the hiring data actually shows

I pulled and analyzed 850 analytical positions in public safety, posted between July 2022 and February 2025. Not opinions about the field. Postings — the documents agencies write when they tell the market what they want.

Two findings should stop you.

First, the crime analysis postings themselves. Across 483 of them, machine learning appears in 0.4 percent. Artificial intelligence appears in 1.0 percent. Meanwhile the private-sector analytical roles we compete against for the same candidates ask for Python in 54.6 percent of postings and machine learning in 37.6 percent. We are recruiting from the same labor pool with a description written for a different decade.

Second, the real-time crime center postings. Positions tied to that function came from 28 different organizations, clustered into a supervisory layer and an analyst layer, with some agencies reposting year after year. That is not experimental hiring. That is a standing function. And the technology profile is entirely different from the traditional analyst role: records management systems in 86.5 percent of postings, GIS and mapping in 76.9 percent, camera systems and license plate readers each at 50.0 percent, social media monitoring at 44.2 percent.

Same job title on both sets of postings. Two different jobs.

Bifurcation, not extinction

Put the forces together and the pattern is a split into two tiers.

The commodity tier does the routine reporting, the standard maps, the recurring statistics. Its relationship to technology is stewardship — running the software that someone else designed. Its exposure to substitution is high, and its demand trajectory is shrinking as the tools mature. It will remain a job. It will remain a low-paid administrative one.

The elevated tier governs the data rather than just consuming it. It owns quality and governance. It builds the extraction, transformation, and loading that makes anything else possible. It questions what the vendor's model is actually doing. It translates analysis into a decision a commander can act on, sometimes in real time. Its exposure to substitution is low, because this is the layer software cannot replace. Its demand is growing.

Here's the uncomfortable part. The elevated tier is also the tier your current hiring practice is least prepared to fill — because your postings don't describe it.

Three beliefs worth reframing

BELIEF: AI is going to replace my analysts. REFRAME: AI will amplify whatever your data already is. If your records are incomplete and your governance is undocumented, faster output means faster wrong answers. The analyst who can govern the data is the reason the technology works at all.

BELIEF: We already have a crime analyst job description. We'll just repost it. REFRAME: Reposting the old requisition is the mechanism that creates the mismatch. Every hiring cycle you run it, the gap between the posting and the work compounds.

BELIEF: This is a technology problem. Once the platform is in, we're set. REFRAME: The technology is a purchasing decision. The workforce transition — the hiring, the reskilling, the steadying of a worried team — is a leadership decision. Those are not the same budget line, and they are not the same skill.

A word about the fear

Expect your people to react to this the way people react to any change they did not choose. Denial first. Resistance next. Acceptance and commitment much later, and only if someone walks them through it rather than around it.

The message that steadies an analyst is not that nothing will change. They already know that's false, and telling them otherwise costs you credibility you'll need later. What steadies them is a visible path — that the role is evolving, that there is a tier worth moving into, and that you intend to help them get there.

Fear left unaddressed becomes attrition. And the analysts who leave first are the ones with the most transferable skills — the exact people you needed for the elevated tier.

What to do with this on Monday

Decide which tier you're hiring for. A real-time center analyst, a chief's-office analyst, and an investigative analyst are three different jobs. Stop running one requisition for all of them.

Rewrite the requisition where the work has already moved. Name the data governance, the integration work, and the modern analytical skills the role actually requires now.

Build the bridge for the people you already have. Your current analysts have the domain grounding the elevated tier needs — the institutional knowledge of why the data looks the way it does. That is expensive to hire and slow to rebuild. Develop it before the commodity layer makes it irrelevant.

Invest in your people. Substitution and skill mismatch are your live risks. Consolidation is a watch item. Offshoring, despite the pilots, has largely not held — because this work depends on a person embedded in your agency, close to the operation and close to the data.

That last point is the whole argument. You can buy the platform in a quarter. You cannot buy the person who knows what your data means.

The full analysis — the five forces that create and erase roles, the complete hiring dataset, and the two-tier breakdown with the specific language to put in your next requisition — is in the IDEA Analytics white paper, Roles That Emerge, Roles That Disappear: Crime Analysis as a Case Study in a Changing Labor Market.

Interested in reading more?

Click here to request the white paper.

Then go read your last analyst posting next to it. If the posting describes a job your agency stopped doing three years ago, you already know what to fix first.

Prepared for IDEA Analytics DTAC Program

Proven. Not Promised.

Source: Herbert, J. (2026). Analytical Staffing in Policing: An Examination of 850 analytical positions in Public Safety, July 2022–February 2025. IDEA Analytics — DTAC Program, Phoenix, AZ. ISBN: 979-8-9936123-5-5