Help wanted, eventually

7-minute read

In brief. Job postings for scientific research roles in America bottomed in September 2025 and have risen 18% since. Hiring has not caught up: Massachusetts biotech research employers took on 6.7 people for every 100 on staff each quarter in 2025, the fewest since 2013, and drug companies have nearly stopped filing restructurings, so fewer experienced people are coming onto the market. Ads come before hires, so competition for candidates comes next, starting with manufacturing roles, where postings are already 20% above their 2020 level. Ads for data analysts remain 39% below 2020, but the share of all ads asking for AI skills has doubled in a year, to a record 7.2%. Managers who fill roles now do so before the market tightens.

Anyone who runs a lab, a plant or a data team at a biotech has spent two years hiring in a slump. Job postings for scientific research and development roles in America fell in 41 of the 42 months after March 2022, to 69% of their pre-pandemic level by September 2025.11Indeed Hiring Lab, Job Postings Index, sector, state, metro and national files, seasonally adjusted, February 1, 2020 = 100. Data as of October 20, 2026 (repository commit 76edff1, verified by checksum); licensed CC BY 4.0. Monthly figures are means of daily values. https://github.com/hiring-lab/job_postings_trackerPrimarythe publisher's open data Since then they have risen in ten of 12 months, to 82%. The ads have turned. The hires have not, yet.

Figure 1. Off the floor. United States, job postings for scientific research and development roles on Indeed, monthly average, Feb 1st 2020=100, log scale. Red: the twelve months since the September 2025 low.11Indeed Hiring Lab, Job Postings Index, sector, state, metro and national files, seasonally adjusted, February 1, 2020 = 100. Data as of October 20, 2026 (repository commit 76edff1, verified by checksum); licensed CC BY 4.0. Monthly figures are means of daily values. https://github.com/hiring-lab/job_postings_trackerPrimarythe publisher's open data Drawn by figures/fig-1.py from figures/fig-1.csv.

Ads come first, hires later

Hiring runs in a fixed order. A company approves a role, advertises it, makes an offer, and only then adds a person to the payroll. Job ads move first, hires later, head count last. Indeed does not publish science postings by state, so the ads here are national; the hires and staff counts are for Massachusetts biotech research employers.

The hires have not moved yet. Those employers took on 6.7 new people for every 100 on staff each quarter in 2025, down from 9.9 in 2019, a fall of 32% (95% confidence interval 3% to 53%) and the lowest rate since 2013.22US Census Bureau, Quarterly Workforce Indicators, Massachusetts, private employers, NAICS 541714 (research and development in biotechnology); hires divided by beginning-of-quarter employment; not seasonally adjusted, so years and like quarters are compared. Series 2010Q2 to 2025Q4, retrieved October 21, 2026. https://www.census.gov/data/developers/data-sets/qwi.htmlPrimaryadministrative records The official count ends with the last quarter of 2025, when they hired 3,661 people, almost exactly the 3,692 of a year earlier. Head count is still shrinking: they had 4.8% fewer staff in early 2026 than a year before.33US Bureau of Labor Statistics, Quarterly Census of Employment and Wages, Massachusetts, private ownership, NAICS 541714. Retrieved October 20, 2026; latest quarter 2026Q1. https://www.bls.gov/cew/Primaryadministrative records MassBio, the state's industry body, counts 2025 as the first year of falling biopharma employment in more than two decades.44MassBio, 2026 Industry Snapshot: Massachusetts biopharma employment down 3.1% in 2025 to 113,503, "the first annual decline in the more than two decades MassBio has tracked this data". https://www.massbio.org/industry-reports/industry-snapshot/Secondarytrade association report

Together the numbers describe the bottom of a hiring cycle. Employers have started advertising again but are not yet adding people, which is when a hiring manager faces the least competition for candidates.

Figure 2. Ads first, hires later. Scientific R&D. Top: job postings on Indeed, United States, Feb 1st 2020=100.11Indeed Hiring Lab, Job Postings Index, sector, state, metro and national files, seasonally adjusted, February 1, 2020 = 100. Data as of October 20, 2026 (repository commit 76edff1, verified by checksum); licensed CC BY 4.0. Monthly figures are means of daily values. https://github.com/hiring-lab/job_postings_trackerPrimarythe publisher's open data Middle: hires per 100 staff each quarter, Massachusetts biotech research employers.22US Census Bureau, Quarterly Workforce Indicators, Massachusetts, private employers, NAICS 541714 (research and development in biotechnology); hires divided by beginning-of-quarter employment; not seasonally adjusted, so years and like quarters are compared. Series 2010Q2 to 2025Q4, retrieved October 21, 2026. https://www.census.gov/data/developers/data-sets/qwi.htmlPrimaryadministrative records Bottom: staff at the same employers, thousands.33US Bureau of Labor Statistics, Quarterly Census of Employment and Wages, Massachusetts, private ownership, NAICS 541714. Retrieved October 20, 2026; latest quarter 2026Q1. https://www.bls.gov/cew/Primaryadministrative records Red: the latest value. Drawn by figures/fig-3.py from figures/fig-3.csv.

Fewer laid-off scientists are coming onto the market

The other source of candidates is drying up. SEC filings show drug and biologics companies reported eight, nine and eight restructurings (form 8-K, Item 2.05) in the first three quarters of 2026, against 29, 22 and 19 a year earlier and 38, 26 and 31 in 2023, a fall of 73% on 2023 (95% confidence interval 16% to 91%).55US Securities and Exchange Commission, EDGAR full-text search: form 8-K filings whose reported items include 2.05 (costs of exit or disposal), filers in industry codes 2834 (pharmaceutical preparations) and 2836 (biological products), counted by quarter on October 21, 2026. https://efts.sec.gov/LATEST/search-index?q=%22Item%202.05%22&forms=8-KPrimaryregulatory filings BioSpace, a trade publication, counted 2,568 people affected by biopharma layoffs in the third quarter of 2026, against 24,891 a year before.66BioSpace, "Q3 layoffs down YOY, with 90% fewer biopharma employees affected", October 8, 2026: 2,568 people in the third quarter of 2026; the 2025 figure for the same quarter, 24,891, is from BioSpace's 2026 Employment Outlook Report. https://www.biospace.com/job-trends/q3-layoffs-down-yoy-with-90-fewer-biopharma-employees-affectedSecondarytrade press tally Fewer layoffs mean fewer experienced people looking for work, just as demand starts to recover.

Figure 3. Fewer pink slips. United States, form 8-K filings reporting a restructuring (Item 2.05) by listed drug and biologics companies, per quarter. Red: 2026.55US Securities and Exchange Commission, EDGAR full-text search: form 8-K filings whose reported items include 2.05 (costs of exit or disposal), filers in industry codes 2834 (pharmaceutical preparations) and 2836 (biological products), counted by quarter on October 21, 2026. https://efts.sec.gov/LATEST/search-index?q=%22Item%202.05%22&forms=8-KPrimaryregulatory filings Drawn by figures/fig-4.py from figures/fig-4.csv.

Manufacturing roles will get harder to fill first

Indeed's job categories cover every employer, so they show who a biotech is really competing with for each kind of worker. Postings for production and manufacturing roles are 20% above their 2020 level and have been rising by about 1.6% a month (95% confidence interval 0.2% to 2.8%). Postings for data analysts and data scientists are 39% below 2020 and flat; software development is 24% below.11Indeed Hiring Lab, Job Postings Index, sector, state, metro and national files, seasonally adjusted, February 1, 2020 = 100. Data as of October 20, 2026 (repository commit 76edff1, verified by checksum); licensed CC BY 4.0. Monthly figures are means of daily values. https://github.com/hiring-lab/job_postings_trackerPrimarythe publisher's open data The slump in data ads began before the chatbots: they peaked in March 2022 and had already fallen 29% by the time ChatGPT launched that November.11Indeed Hiring Lab, Job Postings Index, sector, state, metro and national files, seasonally adjusted, February 1, 2020 = 100. Data as of October 20, 2026 (repository commit 76edff1, verified by checksum); licensed CC BY 4.0. Monthly figures are means of daily values. https://github.com/hiring-lab/job_postings_trackerPrimarythe publisher's open data Manufacturing hires are getting harder already.

Figure 4. Help wanted, selectively. United States, job postings on Indeed by occupation, all employers, monthly average, Feb 1st 2020=100, log scale. Black: the occupation; grey: all postings; red: the September 2026 value.11Indeed Hiring Lab, Job Postings Index, sector, state, metro and national files, seasonally adjusted, February 1, 2020 = 100. Data as of October 20, 2026 (repository commit 76edff1, verified by checksum); licensed CC BY 4.0. Monthly figures are means of daily values. https://github.com/hiring-lab/job_postings_trackerPrimarythe publisher's open data Drawn by figures/fig-2.py from figures/fig-2.csv.

AI skills are the scarcest of all

Demand for AI skills runs the other way, though not on the schedule most people assume. Employers did not rush to hire for AI when the chatbots arrived. OpenAI released ChatGPT on November 30, 2022,77OpenAI, "Introducing ChatGPT", November 30, 2022. https://openai.com/index/chatgpt/Primarythe company's announcement and Anthropic released Claude on March 14, 2023.88Anthropic, "Introducing Claude", March 14, 2023. https://www.anthropic.com/news/introducing-claudePrimarythe company's announcement At ChatGPT's launch, 2.3% of American job ads mentioned AI, machine learning or data science, already down from 3.2% in March 2022, when the tech-hiring boom peaked. The share kept falling after both launches, to 1.7% in June 2023. It did not regain its launch level until September 2024, and passed its 2022 high only in August 2025, nearly three years after ChatGPT.99Indeed Hiring Lab, AI Tracker: share of US job postings containing AI-related terms such as "Machine Learning", "Data Science" and "Artificial Intelligence", seven-day trailing average; data to September 2026 (repository commit bca9fe5); licensed CC BY 4.0. Monthly figures are means of daily values. https://github.com/hiring-lab/ai-trackerPrimarythe publisher's open data

Then it took off. In September 2026, 7.2% of ads mentioned AI, double a year earlier and the highest since the tracker began in 2019; the past 12 months ran 83% above the 12 before (95% confidence interval 49% to 126%).99Indeed Hiring Lab, AI Tracker: share of US job postings containing AI-related terms such as "Machine Learning", "Data Science" and "Artificial Intelligence", seven-day trailing average; data to September 2026 (repository commit bca9fe5); licensed CC BY 4.0. Monthly figures are means of daily values. https://github.com/hiring-lab/ai-trackerPrimarythe publisher's open data Employers are not hiring more analysts. They are asking for AI skills in every other job, which puts a biotech in competition with every industry for scientists and engineers who can use these tools. The tools arrived in 2022; the competition for people who can use them arrived in 2025, and it is still rising.

Figure 5. Wanted: AI. United States, share of job postings on Indeed that mention AI-related terms, monthly average, %. Lines: the launches of ChatGPT (November 30, 2022) and Claude (March 14, 2023). Red: the past twelve months.99Indeed Hiring Lab, AI Tracker: share of US job postings containing AI-related terms such as "Machine Learning", "Data Science" and "Artificial Intelligence", seven-day trailing average; data to September 2026 (repository commit bca9fe5); licensed CC BY 4.0. Monthly figures are means of daily values. https://github.com/hiring-lab/ai-trackerPrimarythe publisher's open data Drawn by figures/fig-5.py from figures/fig-5.csv.

Which roles to fill now

• AI skills: hire now and expect to pay for them. The share of job ads asking for AI has doubled in a year, so this is where competition is fiercest.
• Project and program managers: hire now. Ads across all US employers are 13% above 2020 and rising 1.7% a month (95% confidence interval 0.7% to 2.7%), the clearest rise among the roles named in the plan before any data was pulled.
• Manufacturing: open requisitions early and budget more, because postings are already above 2020 and rising.
• Research scientists: postings are still 18% below 2020 but rising faster than the job market as a whole, so expect competition here to return next.
• Data analysts: a buyer's market, with ads 39% below 2020.
• Benchmark: compare your own hiring rate with 6.7 hires per 100 staff a quarter, the 2025 rate for Massachusetts biotech research employers.
• Timing: watch two numbers: Indeed's science postings, updated weekly, and the Census hiring count for the first quarter of 2026. When hires start rising, the window is closing.

BigBio puts a 46% probability on science postings reaching 86 on Indeed's index by March 2027, about 5% above September's level, against a 35% share of six-month stretches since 2020 with a gain that large. The forecast will be scored in public in April 2027.1010The evidence pack: the pre-registration and its dated amendment, the analysis and calibration-test code, the numbers sheet, the results file and the registered forecast. Statistics follow Harvey Motulsky, Intuitive Biostatistics (4th edition, Oxford University Press, 2018): effect sizes with 95% confidence intervals, tests fixed in advance, later additions labelled. evidence packPrimarythis analysis

The official count of hires runs nearly ten months behind: in October 2026 its latest figures were for the last quarter of 2025. By the time it shows the turn, the roles worth filling early will have been filled.

Key figures

Series: science, project management and production postings and the AI share, last 24 months; hiring rate, 2015-2025; staff, 2019-2026; filings, 2021-2026. Confidence intervals from a time-series model that lets each month or quarter carry over into the next. The first three rows were the tests fixed before any data was pulled, and project management was named in the same plan; the rest were added later.1010The evidence pack: the pre-registration and its dated amendment, the analysis and calibration-test code, the numbers sheet, the results file and the registered forecast. Statistics follow Harvey Motulsky, Intuitive Biostatistics (4th edition, Oxford University Press, 2018): effect sizes with 95% confidence intervals, tests fixed in advance, later additions labelled. evidence packPrimarythis analysis

How the analysis was done. BigBio fixed the question, the tests and the pass mark before pulling any data, and time-stamped them. The data came straight from the record-keepers: Indeed's open postings data, the Census Bureau's hiring records, the Bureau of Labor Statistics' staff counts and the SEC's filings, with press figures kept separate. The method for the confidence intervals was checked on 300 simulated markets with a known answer before it touched the real data, and every number was checked back to its source file. The first report took under half an hour from the locked question; that speed is what makes a weekly refresh, cut to one company's roles and sites, affordable. The data, code and plan are published.1010The evidence pack: the pre-registration and its dated amendment, the analysis and calibration-test code, the numbers sheet, the results file and the registered forecast. Statistics follow Harvey Motulsky, Intuitive Biostatistics (4th edition, Oxford University Press, 2018): effect sizes with 95% confidence intervals, tests fixed in advance, later additions labelled. evidence packPrimarythis analysis

Footnotes

1. Indeed Hiring Lab, Job Postings Index, sector, state, metro and national files, seasonally adjusted, February 1, 2020 = 100. Data as of October 20, 2026 (repository commit 76edff1, verified by checksum); licensed CC BY 4.0. Monthly figures are means of daily values. https://github.com/hiring-lab/job_postings_tracker (Primary -- the publisher's open data) ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶

2. US Census Bureau, Quarterly Workforce Indicators, Massachusetts, private employers, NAICS 541714 (research and development in biotechnology); hires divided by beginning-of-quarter employment; not seasonally adjusted, so years and like quarters are compared. Series 2010Q2 to 2025Q4, retrieved October 21, 2026. https://www.census.gov/data/developers/data-sets/qwi.html (Primary -- administrative records) ↩ ↩²

3. US Bureau of Labor Statistics, Quarterly Census of Employment and Wages, Massachusetts, private ownership, NAICS 541714. Retrieved October 20, 2026; latest quarter 2026Q1. https://www.bls.gov/cew/ (Primary -- administrative records) ↩ ↩²

4. MassBio, 2026 Industry Snapshot: Massachusetts biopharma employment down 3.1% in 2025 to 113,503, "the first annual decline in the more than two decades MassBio has tracked this data". https://www.massbio.org/industry-reports/industry-snapshot/ (Secondary -- trade association report) ↩

5. US Securities and Exchange Commission, EDGAR full-text search: form 8-K filings whose reported items include 2.05 (costs of exit or disposal), filers in industry codes 2834 (pharmaceutical preparations) and 2836 (biological products), counted by quarter on October 21, 2026. https://efts.sec.gov/LATEST/search-index?q=%22Item%202.05%22&forms=8-K (Primary -- regulatory filings) ↩ ↩²

6. BioSpace, "Q3 layoffs down YOY, with 90% fewer biopharma employees affected", October 8, 2026: 2,568 people in the third quarter of 2026; the 2025 figure for the same quarter, 24,891, is from BioSpace's 2026 Employment Outlook Report. https://www.biospace.com/job-trends/q3-layoffs-down-yoy-with-90-fewer-biopharma-employees-affected (Secondary -- trade press tally) ↩

7. OpenAI, "Introducing ChatGPT", November 30, 2022. https://openai.com/index/chatgpt/ (Primary -- the company's announcement) ↩

8. Anthropic, "Introducing Claude", March 14, 2023. https://www.anthropic.com/news/introducing-claude (Primary -- the company's announcement) ↩

9. Indeed Hiring Lab, AI Tracker: share of US job postings containing AI-related terms such as "Machine Learning", "Data Science" and "Artificial Intelligence", seven-day trailing average; data to September 2026 (repository commit bca9fe5); licensed CC BY 4.0. Monthly figures are means of daily values. https://github.com/hiring-lab/ai-tracker (Primary -- the publisher's open data) ↩ ↩² ↩³

10. The evidence pack: the pre-registration and its dated amendment, the analysis and calibration-test code, the numbers sheet, the results file and the registered forecast. Statistics follow Harvey Motulsky, Intuitive Biostatistics (4th edition, Oxford University Press, 2018): effect sizes with 95% confidence intervals, tests fixed in advance, later additions labelled. evidence pack (Primary -- this analysis) ↩ ↩² ↩³