Quick answer
Mixed, with no clear difference overall. On Indeed, Sales and Customer Service postings are down less than software postings on both start dates, and Marketing sits closer to software than they do. Marketing is the sector that does not fit a simple story. Postings on one job site are not jobs or hires.
Key takeaways
- What is counted: postings on one job site (Indeed), seasonally adjusted, by Indeed's own title-based sector label, for the US as a whole.
- Feb 2020 base first, then Dec 2021 to Nov 2022 base: Software Development -29% and -65%; Sales -6.4% and -28%; Marketing -23% and -53%; Customer Service -11% and -33%.
- Marketing is the sector that does not fit: on new postings, Feb 2020 base, it reads -23% against -17% for software.
- Software is not alone: Data & Analytics, IT Systems & Solutions and IT Infrastructure also sit among the largest declines.
Did Sales and Marketing postings fall less than software postings?
Mixed, with no clear difference overall. Sales and Customer Service postings are down less than software's on both start dates (Sales -6.4% and -28%, software -29% and -65%, Feb 2020 base first), and Marketing (-23% and -53%) sits closer to software than they do on both. Marketing is the sector that does not fit a simple story, and its section below says why.
What is counted is the number of postings on one job site, Indeed, seasonally adjusted, by Indeed's own sector label, nationally. Postings are not jobs or hires, and AI agents are not measured here.
How did we get these numbers?
You can read the result as postings in Oct 2025 to Sep 2026 compared with two starting points: Feb 2020, and the average of Dec 2021 to Nov 2022. The source is the Indeed Hiring Lab Job Postings Index, which is seasonally adjusted and sets the number of postings on 1 February 2020 at 100. Averaging days into months gives 47 sectors over 80 months (Feb 2020 to Sep 2026); October 2026 (2 days) is dropped.
Attribution: Indeed Hiring Lab, Job Postings Index, https://github.com/hiring-lab/job_postings_tracker (CC BY 4.0), data to 2026-10-02. Indeed adopted a seasonal-adjustment method developed by the Deutsche Bundesbank in November 2024 and says it revised its history, so values depend on the files retrieved on 2026-10-06 and can differ from figures published earlier.
I expected selling sectors to fall less than software. Both windows, the sector groups and the decision rules were fixed before any sector was compared, and the direction for software was visible before the method was fixed, so this is a check in form, not a blind test. No margin of error exists for this index: Indeed publishes none, so each figure is a plain reading of what happened on Indeed, not an estimate with a range.
Technical detail (skippable): "change from the reference year" is the current mean divided by the reference mean, minus one; "versus Feb 2020" is the current mean minus 100. The rank correlation across the 47 sectors between the two measures is rho = 0.83, which shows how far the base choice reorders sectors. Software's monthly readings are so persistent (lag-1 correlation 0.99 in levels) that the 80 carry far less information than 80 separate ones. The 80 months and the 1,806 ordered pairs of other sectors behind the gap positions are not a sample size, and the pairs share sectors, so they are not independent.
What do the four sectors show on each base?
On the Feb 2020 base software is lowest of the four at -29%, and Sales is nearest to its starting level at -6.4%.
| Sector | Change vs Feb 2020 | Change vs Dec 2021 to Nov 2022 |
|---|---|---|
| Software Development | -29% | -65% |
| Sales | -6.4% | -28% |
| Marketing | -23% | -53% |
| Customer Service | -11% | -33% |
The index, in Oct 2025 to Sep 2026 and then in Dec 2021 to Nov 2022, was 71 and 202 for software, 94 and 129 for Sales, 77 and 162 for Marketing and 89 and 134 for Customer Service; in that reference year the four sat 102%, 29%, 62% and 34% above Feb 2020, in the same order, a postings boom year that was highest for software. The Feb 2020 base is the cleaner comparison of levels, and neither base is the true one.
Ranks among the 47 sectors (1 = largest fall), Feb 2020 base then the other base: software rank 5 of 47 and rank 2 of 47, Sales 16 and 35, Marketing 7 and 8, Customer Service 12 and 27. Software and Marketing are among the ten largest falls on both bases; Sales and Customer Service are not.
How large are the gaps?
Each gap is the change for the selling sector minus the change for software, in points, and all three are positive on both bases. "+22 points" means the change for Sales was 22 points higher than the change for software.
| Selling sector minus software | Feb 2020 base | Dec 2021 to Nov 2022 base |
|---|---|---|
| Sales | +22 points | +37 points |
| Marketing | +5.3 points | +12 points |
| Customer Service | +18 points | +31 points |
In plain terms, Sales' and Customer Service's lead over software is large next to the usual gap between two other sectors on the Dec 2021 to Nov 2022 base, and ordinary on the Feb 2020 base. Gaps this size for Sales and Customer Service are in the upper tail of the gaps between pairs of other sectors on the Dec 2021 to Nov 2022 base (3.8% and 6.3% of pairs are at least as large) and not unusual on the Feb 2020 base (26% and 31%); this is a position among other gaps, not a test. Five points is a line we drew in advance for what counts as a difference worth noting, and Marketing's Feb 2020 gap barely clears it.
Is Marketing the sector that does not fit?
Yes. Marketing fell 23% against Feb 2020 (rank 7 of 47) and 53% against Dec 2021 to Nov 2022 (rank 8 of 47), among the larger declines of all 47 sectors. Its gap over software is +5.3 points on the Feb 2020 base and +12 points on the other; 45% and 27% of the other-sector pairs have a gap at least as large. On the Dec 2021 to Nov 2022 base its change sat 17 points below the other-sector average of -35%.
New postings (on Indeed for 7 days or fewer) change the picture. On the Feb 2020 base and then the Dec 2021 to Nov 2022 base they read: software -17% and -58%, Sales -10% and -36%, Marketing -23% and -49%, Customer Service -12% and -48%. Marketing's advantage over software disappears on new postings, Feb 2020 base: -23% against -17%.
How do the three selling sectors compare with the other sectors?
It depends on the base. The unweighted average of the other 43 sectors is 8.1% on the Feb 2020 base and -35% on the Dec 2021 to Nov 2022 base; these are means of sector readings, not all-economy figures. On the Feb 2020 base all three selling sectors (-6.4%, -23%, -11%) are below the average of 8.1%. On the Dec 2021 to Nov 2022 base Sales is 7.8 points above the average of -35% and Customer Service 2.1 points above it, while Marketing is below it.
Is software the exception, or part of a wider pattern?
Software is among the largest declines, but several tech-labelled sectors sit with it: on the Dec 2021 to Nov 2022 base this is a tech-wide pattern in Indeed's labels, not software alone. On the Dec 2021 to Nov 2022 base, Data & Analytics is -66% (rank 1), software -65% (and -29% on the Feb 2020 base), then IT Systems & Solutions at -59% and IT Infrastructure, Operations & Support at -55%.
What can the data not show?
They cannot show who does the work or why postings moved.
- One job site, postings only. Postings are not jobs, openings filled or hires.
- National only. There is no sector-by-state, metro-by-sector or seniority split. Sector is a title-based label, and Software Development is a title class, not an industry.
- 47 sectors are the unit. Each selling-versus-software contrast is one sector against one sector.
- No standard errors, intervals or tests. The numbers are descriptive readings, from files with data to 2026-10-02 and October 2026 dropped.
- The windows were fixed in advance. Other windows were not tested.
- The question is about who does the work, which postings cannot show. AI, interest rates and the 2021-22 postings boom are not measured here, and the design cannot show what lies behind a move.
- Metro lines are context only. For context, all-occupation postings in the San Francisco and San Jose areas fell 47% and 43% from Dec 2021 to Nov 2022 to Oct 2025 to Sep 2026, and 32% and 22% against Feb 2020, and the median across the 593 areas in Indeed's metro file is a fall of 27% on the first of those bases only. These figures cover all occupations and say nothing about software, sales or any sector in California.
How can you check it yourself?
Download the two Indeed Hiring Lab files under sources and average the daily index by month. In the sector file, filter the variable to total postings, pick a display name, average the daily values for Oct 2025 to Sep 2026 and compare with 100, then with the Dec 2021 to Nov 2022 average.
The companion piece on sales and marketing occupations vs software occupations uses employment and wage data, a different dataset from these postings. For other cautious readings of public data, see the junior developer payroll study and the Census piece on AI use.
What can you do with this?
When a headline says one kind of posting fell or rose, check the start date before you repeat it. "Software postings down 65%" is true only from the Dec 2021 to Nov 2022 base; from Feb 2020 the same series is down 29%, so a headline that gives one figure without the start date fails the check. Also ask whether the starting period was a peak, whether the label is an industry or a title class, and whether the count is postings or people.
Analysis by Deivy Hernandez, from public data files.
FAQ
Which start date should I trust?
Neither is the true one. Feb 2020 is the date Indeed sets to 100. In Dec 2021 to Nov 2022 the software index sat 102% above that level, so the -65% starts from a postings-boom level and the -29% starts from Feb 2020. Read both side by side.
Does this mean software jobs are disappearing?
No. The data count postings on one site, Indeed, not jobs, openings filled or hires. A fall in postings says nothing about how many people hold software jobs.
Does ‘Software Development’ on Indeed cover every tech employer?
No. It is a title class: Indeed gives each posting a title-based sector label, 47 sectors in all. It is not an industry and not the government's occupation list.
Can I use these numbers to decide whether to hire a salesperson?
Not directly. They are national and from one site, so they say nothing about applicants or competing employers in your town.