Career Growth: When to Stay vs When to Switch

"Should I stay or switch jobs?" is usually framed as loyalty against ambition. Data gives a more useful frame. Over 29 years of US wage records, people who changed jobs usually saw their pay grow faster, but how much faster depended heavily on the job market at the time. And when data professionals were asked what makes a job satisfying, pay came first only narrowly, with autonomy close behind. The analysis below uses the Atlanta Fed's wage tracker (Atlanta Fed: Wage Growth Tracker) and answers from data roles in the 2025 Stack Overflow survey (Stack Overflow Developer Survey 2025).

import pandas as pd

# Atlanta Fed Wage Growth Tracker: median 12-month wage growth, 3-month moving average, US CPS data
wgt = pd.read_excel("wage-growth-data.xlsx", sheet_name="data_overall", header=1)
wgt = wgt.rename(columns={wgt.columns[0]: "month"})[["month", "Job Stayer", "Job Switcher"]].dropna(subset=["month"])
wgt[["Job Stayer", "Job Switcher"]] = wgt[["Job Stayer", "Job Switcher"]].apply(pd.to_numeric, errors="coerce")
missing = wgt["Job Stayer"].isna()
print(f"S0 months published as '.' (no estimate): {missing.sum()}, {wgt.month[missing].min():%b %Y} to {wgt.month[missing].max():%b %Y}")
wgt = wgt[~missing]
wgt["gap"] = wgt["Job Switcher"] - wgt["Job Stayer"]
print(f"S1 wage growth, switchers vs stayers: {len(wgt)} months, {wgt.month.min():%b %Y} to {wgt.month.max():%b %Y}")
print(f"   switchers ahead in {(wgt.gap > 0).mean():.0%} of months | average gap {wgt.gap.mean():.1f} points")
for label, start, end in (("2009-2010 (after the financial crisis)", "2009-01", "2010-12"),
                          ("2022 (hot labour market)", "2022-01", "2022-12"),
                          ("latest 12 months", wgt.month.max() - pd.DateOffset(months=11), wgt.month.max())):
    w = wgt[(wgt.month >= pd.Timestamp(start)) & (wgt.month <= pd.Timestamp(end))]
    print(f"   {label:<40} stayers {w['Job Stayer'].mean():.1f}% | switchers {w['Job Switcher'].mean():.1f}% | gap {w.gap.mean():.1f}")

# Stack Overflow Developer Survey 2025: what makes a tech job satisfying, ranked 1 (most) to 15
so = pd.read_csv("results.csv", low_memory=False)
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"])]
cols = [c for c in so.columns if c.startswith("JobSatPoints_") and not c.endswith("TEXT")]
names = {r.qname: r.sub for r in schema.itertuples() if r.qname in cols}
ranked = data[cols].dropna()
share_top3 = (ranked <= 3).mean().rename(names).drop("Other (please specify):").sort_values(ascending=False)
print(f"\nS2 data professionals who ranked each factor in their top 3 (n={len(ranked):,})")
for name, share in share_top3.items():
    print(f"   {share:>4.0%}  {name}")

print("\nS3 job satisfaction (0-10) by whether they considered a change in the past year, data roles")
sat = data.dropna(subset=["JobSat", "NewRole"]).groupby("NewRole").JobSat.agg(["median", "mean", "count"])
for label, row in sat.sort_values("mean", ascending=False).iterrows():
    print(f"   median {row['median']:.0f} | mean {row['mean']:.1f} | n={int(row['count']):>4}  {label}")

S0 months published as '.' (no estimate): 1, Oct 2025 to Oct 2025
S1 wage growth, switchers vs stayers: 356 months, Jan 1997 to Sep 2026
   switchers ahead in 86% of months | average gap 0.7 points
   2009-2010 (after the financial crisis)   stayers 2.3% | switchers 2.2% | gap -0.2
   2022 (hot labour market)                 stayers 5.5% | switchers 7.5% | gap 2.1
   latest 12 months                         stayers 3.5% | switchers 4.4% | gap 0.9

S2 data professionals who ranked each factor in their top 3 (n=1,245)
    42%  Competitive pay and benefits
    41%  Autonomy and trust to manage your own tasks
    36%  Solving real-world problems
    32%  Innovation through solving challenging and complex problems
    22%  Working with new technologies and tools
    21%  Job stability and career growth with a single employer
    21%  You like your manager
    19%  Developing specialized expertise
    16%  Collaboration and support from a team
    16%  Control over the level of quality in projects
    10%  Recognition from peers for your work
     9%  Expert mentors
     9%  Recognition from leadership for your work
     6%  Opportunities to mentor and lead junior-level coworkers

S3 job satisfaction (0-10) by whether they considered a change in the past year, data roles
   median 8 | mean 7.8 | n= 491  I have neither consider or transitioned into a new career or industry
   median 8 | mean 7.8 | n= 173  I have transitioned into a new career and/or industry voluntarily
   median 8 | mean 7.6 | n=  27  I have transitioned into a new career and/or industry involuntarily
   median 7 | mean 6.8 | n= 375  I have somewhat considered changing my career and/or the industry I work in
   median 6 | mean 6.0 | n= 228  I have strongly considered changing my career and/or the industry I work in

What the wage data shows

The tracker follows the same people a year apart and reports the median change in their hourly pay. Switchers here means people who moved to a different occupation or industry within the year, or changed employer or duties in the past three months. Across 356 months of data, switchers' pay grew faster in 86% of them, by 0.7 percentage points on average.

The averages hide the important part. In 2009 and 2010, after the financial crisis, the switcher premium disappeared: switchers' wage growth was slightly below stayers'. In 2022, when employers competed hard for staff, switchers were 2.1 points ahead, 7.5% against 5.5%. In the most recent year it was 0.9 points. The value of switching depends on demand: when companies are hiring aggressively, moving pays; when they are cautious, the move itself is the main thing you get.

Two caveats. This is US data, and it measures pay growth, not total reward, so equity, bonuses and learning are not in it. In the Indian market the same logic applies, but check current demand for your skills with real job postings, rather than assuming a switch brings a 30% rise because it did for a friend in 2022.

What data professionals value

Pay matters, but it is not everything. Of 1,245 data engineers, data scientists, analysts and AI/ML engineers who ranked 14 job factors, 42% put competitive pay and benefits in their top three. Almost as many, 41%, put autonomy and trust to manage their own work there, and 36% solving real-world problems. "Job stability and career growth with a single employer" made the top three for 21%, the same share as liking your manager.

Satisfaction tracked with intent. Data professionals who had not considered a career change rated their job 7.8 out of 10 on average; those who had strongly considered one rated it 6.0. Those who had already moved voluntarily were back at 7.8, which suggests a move that fixes the real problem can work.

A way to decide

Use the data above as a checklist for your own situation, not as a rule.

1. Name the real problem. Write down what is wrong in one sentence. Pay below market, no autonomy, work that does not matter, a manager you cannot work with, or no new skills are all different problems with different fixes.
2. Test whether it can be fixed where you are. Pay can be raised in a review if you bring market evidence; autonomy and project choice can often be negotiated; a bad manager sometimes leaves first. Ask directly, with a deadline in your own mind, before you assume nothing will change.
3. Check the market, not anecdotes. Count current postings for your role and skills, and talk to two or three people who were hired recently. A hot market rewards switching; a cautious one rewards patience and internal growth.
4. Price the hidden costs. A new job resets your reputation and often your access to the interesting projects. Unvested bonuses, notice periods and a probation period are real costs too.
5. Decide what you will do in each case. If the problem can be fixed by a date, stay and review it then. If not, start searching while employed, which keeps your negotiating position.

When staying is the better move

Staying usually wins when you are learning fast, when a promotion or a strong project is close, or when you have been in the role less than a year and the problem is newness rather than the job. Building a track record of finished work is also easier in one place. The data analyst career path shows what progression inside a team looks like.

When switching is the better move

Switching usually wins when your pay is clearly below market and the company will not move, when you have stopped learning, when the work you want does not exist where you are, or when the manager or culture is the problem and will not change. It is also the usual route into a different specialism, such as from analytics into engineering, which How to make a career pivot without a pay cut covers.

Questions to ask yourself

• Would I take this job again today, knowing what I know?
• What will I be able to do in a year that I cannot do now?
• Is the problem the company, the team, the manager or the role?
• Have I asked clearly for what I want?

For a longer horizon, see how to plan a 5-year career in data.

The skills that keep your options open are taught in the Data Analytics course.

Sources

• Wage Growth Tracker (Federal Reserve Bank of Atlanta) — Federal Reserve Bank of Atlanta
• 2025 Stack Overflow Developer Survey — Stack Overflow