Retail Salespersons

BLS National Employment Matrix 2024-34

→ 2034

BLS Employment Projections — industry-occupation matrix + labor productivity assumptions. The 2024-34 cycle projects -0.5% employment change for 41-2031, equivalent to approximately -19,600 positions — from 3,936,700 (2024) to 3,917,100 (2034). This is classified as "little or no change" against an all-occupations average of +4%. The BLS methodology models continued e-commerce displacement and self-checkout expansion as the primary headwinds, offset by continued demand for in-person retail in categories resistant to online substitution (cars, luxury, high-complexity products). The projection does not explicitly model the Amazon Go closure or the pace of AI shopping-assistant adoption, which could shift the number in either direction by a few percent.

BLS Employment Projections 2024-34 — retail trade sector

→ 2034

BLS projects the retail trade sector (NAICS 44-45) to decline -1.2% overall from 2024-2034 — the largest absolute employment decline of any major sector in the current projection cycle. The sector-level projection is slightly more pessimistic than the occupation-specific 41-2031 projection because it captures store closures in department store formats and general merchandise that specifically eliminate floor-sales positions. Reported here as a cross-check against the occupation-level projection; the sector and occupation numbers should move in the same direction but may differ by format mix.

Eloundou et al. — "GPTs are GPTs" (2023)

→ 2028

GPT-4 task-by-task LLM exposure labeling on O*NET tasks for Sales and Related Occupations. Retail salespersons score in the low-to-medium range for LLM exposure — the dominant tasks (demonstrating merchandise, advising on purchases, processing transactions, stocking shelves) require physical presence and situational judgment that LLMs cannot provide from a data center. The important contrast to Frey & Osborne: Eloundou measures LLM-specific exposure, not general automation. E-commerce and self-checkout displaced retail salespeople; language models are a secondary threat via AI shopping assistants that reduce the number of in-store visits rather than directly replacing the in-store role. The -8% estimate reflects this indirect channel: AI-assisted product discovery at home reduces foot traffic, which reduces floor staff headcount.

Frey & Osborne (2013)

→ 2033

Gaussian-process classifier on O*NET task features. Frey & Osborne placed Retail Salespersons among the highest-risk occupations in their 2013 study — at over 92% probability of computerization, in the company of cashiers and waitstaff. The F&O bottleneck analysis found that retail sales tasks — querying product information, processing transactions, advising customers on standard purchases — presented no meaningful barriers to automation from pattern-recognition or procedural-automation technologies. This was a reasonable prediction in 2013; what F&O did not model was the deployment friction (Amazon Go closure), consumer preference for physical retail in specific categories, and the role's ongoing adaptation toward the tasks AI cannot substitute. The -92% here represents the F&O probability as an implied ceiling on displacement, not a realized forecast. Actual 2013-2024 employment declined ~15% from the 2015 peak, far below F&O's scenario.