Career Roadmap: Analyst to Scientist to Lead

The classic career roadmap from analyst to scientist to lead reads like a ladder with management at the top. Real data careers look different. Experience as an analyst counts towards data science roles, "lead" can mean a technical leader rather than a manager, and most experienced data professionals never manage anyone. The analysis below uses answers from 1,925 data engineers, scientists, analysts and AI/ML engineers in the 2025 Stack Overflow survey (Stack Overflow survey 2025), alongside a public job ladder.

import pandas as pd

so = pd.read_csv("results.csv", low_memory=False)          # 2025 Stack Overflow Developer Survey
schema = pd.read_csv("schema.csv", encoding="utf-8-sig")
data = so[so.DevType.isin(["Data engineer", "Data scientist", "Data or business analyst", "AI/ML engineer"])
          & so.ICorPM.notna()].copy()
data["track"] = data.ICorPM.map({"Individual contributor": "IC", "People manager": "manager"})
data["experience"] = pd.cut(data.WorkExp, [0, 2, 5, 10, 15, 20, 100], labels=["1-2", "3-5", "6-10", "11-15", "16-20", "21+"])

print("L1 share of data professionals who manage people, by years of experience")
print("   " + " | ".join(f"{b}: {g.track.eq('manager').mean():.0%}" for b, g in data.groupby("experience", observed=True)))
print(f"   all experience levels: {data.track.eq('manager').mean():.0%} of {len(data):,}")

senior = data[data.WorkExp >= 10]
print("\nL2 data professionals with 10+ years: individual contributors vs managers")
for track, g in senior.groupby("track"):
    us_pay = g[g.Country.eq("United States of America")].ConvertedCompYearly.dropna()
    print(f"   {track:<8} n={len(g):>4} | job satisfaction {g.JobSat.mean():.1f}/10 | US median pay ${us_pay.median():,.0f} (n={len(us_pay)})")

print("\nL3 job satisfaction factors in the top 3, data professionals with 10+ years")
cols = [c for c in so.columns if c.startswith("JobSatPoints_") and not c.endswith(("TEXT", "_15"))]
names = {r.qname: r.sub for r in schema.itertuples() if r.qname in cols}
ranked = senior.dropna(subset=cols)
top3 = ranked.groupby("track")[cols].apply(lambda g: (g <= 3).mean()).T.rename(index=names)
top3["difference"] = top3["manager"] - top3["IC"]
for name, row in top3.sort_values("difference").iterrows():
    print(f"   IC {row['IC']:>4.0%} | manager {row['manager']:>4.0%}  {name}")
print(f"   (n: IC {ranked.track.eq('IC').sum()}, manager {ranked.track.eq('manager').sum()})")

L1 share of data professionals who manage people, by years of experience
   1-2: 6% | 3-5: 6% | 6-10: 10% | 11-15: 16% | 16-20: 20% | 21+: 15%
   all experience levels: 11% of 1,925

L2 data professionals with 10+ years: individual contributors vs managers
   IC       n= 743 | job satisfaction 7.3/10 | US median pay $175,000 (n=148)
   manager  n= 150 | job satisfaction 7.4/10 | US median pay $195,000 (n=24)

L3 job satisfaction factors in the top 3, data professionals with 10+ years
   IC  47% | manager  33%  Competitive pay and benefits
   IC  47% | manager  39%  Autonomy and trust to manage your own tasks
   IC  36% | manager  32%  Solving real-world problems
   IC  21% | manager  17%  Job stability and career growth with a single employer
   IC   7% | manager   6%  Recognition from leadership for your work
   IC   5% | manager   5%  Expert mentors
   IC   6% | manager   6%  Opportunities to mentor and lead junior-level coworkers
   IC   9% | manager   9%  Recognition from peers for your work
   IC  17% | manager  18%  Developing specialized expertise
   IC  23% | manager  26%  Working with new technologies and tools
   IC  22% | manager  26%  You like your manager
   IC  13% | manager  17%  Collaboration and support from a team
   IC  31% | manager  38%  Innovation through solving challenging and complex problems
   IC  15% | manager  27%  Control over the level of quality in projects
   (n: IC 421, manager 66)

Stage 1 to 2: analyst experience counts

Analyst work is not a detour on the way to data science; good job ladders count it. GitLab publishes its data science job family, and an intermediate data scientist there needs either four or more years in an analytics role or two or more years in predictive analytics or data science. A senior data scientist needs six years in analytics or four in data science (GitLab: data science job family). Two years of analytics roughly equals one year of modelling work.

The way to make analyst years count more is to do data science work inside the analyst job: forecasts, experiments, segmentation and simple models that someone uses. From data analyst to data scientist covers that transition step by step.

Stage 2 to 3: "lead" has two meanings

After senior level, the path splits.

• Technical lead (individual contributor). Staff or principal data scientist or engineer. You set technical direction, take on the hardest problems, review others' work and mentor, without managing people. GitLab says its staff and principal analytics engineer levels exist precisely to give deep technical people a career path without moving into management (GitLab: analytics engineering job family).
• People lead (manager). Data team lead or manager. You hire, set priorities, run reviews, unblock people and represent the team. Your output becomes the team's output.

What the data says about the two tracks

Most data professionals stay individual contributors. Only 6% of those with up to five years of experience managed people, rising to 16% at 11 to 15 years and 20% at 16 to 20. Overall, 11% were managers. Even among people with ten or more years, 743 of 893 were individual contributors.

Satisfaction was the same. Experienced individual contributors rated their job 7.3 out of 10; managers 7.4. Management is not the only road to a satisfying career.

Pay was somewhat higher for managers, at least in the United States: a median of $195,000 against $175,000 for experienced individual contributors. The manager figure comes from only 24 people, so treat it as a hint, not a rule. Many companies pay staff and principal individual contributors at the same bands as managers.

They value different things. Among experienced data professionals, managers more often ranked control over quality in their top three job satisfaction factors (27% against 15%), and less often competitive pay (33% against 47%) or autonomy over their own tasks (39% against 47%). A manager's satisfaction comes through the team's work; an individual contributor's through their own.

How to choose

Try both before you commit. Ask yourself which of these you would rather spend a week on:

• Leading a project technically: designing the approach, writing the hardest parts, reviewing others' code and teaching. That points to the technical track.
• Hiring and growing people: running one-to-ones, resolving conflicts, negotiating priorities with stakeholders and seeing others succeed. That points to management.

Ways to test it without changing your title: lead a project with two or three colleagues, mentor an intern or new joiner, run the team's planning for one quarter, or own a relationship with a business stakeholder. Many people find out what they like by doing a little of each.

Tell your manager which track interests you; they can only offer the right opportunities if they know. The switch is not one-way. Plenty of people try management and return to a technical track, or the reverse, and good companies allow both.

Skills for each stage

For the analyst stages in more detail see the data analyst career path, and for getting the next title in your current team see how to get promoted in a data team.

Statistics, modelling and experiment design for the scientist stage are part of the Data Science course.

Sources

• Data Science job family (GitLab Handbook) — GitLab
• Analytics Engineering job family (GitLab Handbook) — GitLab
• 2025 Stack Overflow Developer Survey — Stack Overflow