Why the doorstep stays with a person
Ask whether AI will replace door to door sales workers and the answer starts with where the work happens. It happens on sidewalks, porches and street corners. Someone walks a route, knocks, reads a stranger’s face in the first three seconds and decides whether to keep talking. Software can draft the pitch. It cannot stand there and be trusted.
Two duties carry most of the weight. The first is the cold approach: finding the house or the passer-by, opening a conversation nobody asked for, and handling the brush-off without losing the next door. The second is closing on the spot, which often means taking cash or a card at a cart, a stand or a front step, and sorting out a complaint before the customer walks away. Both are physical, social and local. They depend on reading a person, not a dataset.
There is a third piece that does move. Route planning, lead lists, order records, receipts and follow-up messages are all paperwork, and paperwork is where current tools are strongest. That is the honest shape of the pressure here: tasks peeling away from the job rather than the job disappearing. You can see how the three questions behind the figures above are built in how we score jobs.
What AI does, what it helps with, what people keep
The tasks our data marks as things AI can do on its own are the record-keeping and outreach ones: building and sorting prospect lists, logging sales and orders, and writing the follow-up contact after a visit. Across this job, the share of task time in that group is 29%. None of it removes the visit. It removes the evening spent writing things down.
The helper group is larger and more interesting. Suggesting which streets or stalls to work next, prepping answers to common objections, pricing and change-making, and translating a pitch for a customer who speaks another language all go faster with software in a pocket. Our split puts 11% of task time here. The worker still makes the call; the phone just carries the notes. The share of task time AI can handle today is 31 out of 100, and what that coverage figure counts is worth a read before comparing jobs.
What stays with people is the approach and the close. Walking the route, demonstrating a product in someone’s hands, judging whether to push or back off, calming an annoyed resident, handling cash safely in public. Our task split leaves 60% of task time in that group.
What the evidence actually shows
No study has put an AI system on a porch against a working vendor and measured who sold more. The evidence grade recorded for this job is D, which means there is no direct test against people in this occupation yet, so we publish no quality-parity number for it. The grades and what each one requires are explained on the quality-parity page.
What would settle it is narrow and testable: a field trial comparing conversion rates on matched routes, one worked by a person and one worked by phone, app or kiosk outreach, with the same product, price and territory. Until something like that exists, the useful numbers come from the labor market. The Bureau of Labor Statistics records median pay of $41,380 and employment of about 2,760 in this occupation, with projected employment falling 10.4% between 2025 and 2035 (BLS, 2025). That decline is mostly about how people buy now, not about a machine doing the knocking.
Good to know: shrinking employment and automatable tasks are two different measures, and this job shows pressure on both for different reasons.
When the picture could change
Most likely between 2035 and 2056 (8 in 10 of our scenarios). The method behind that window, including what the median and the spread mean, is set out on the replacement-year page.
Two things could pull it earlier. One is channel substitution: if app ordering, delivery and targeted ads keep absorbing the impulse sale, fewer routes get walked at all, which thins the job without automating a single doorstep. The other is cheap autonomous delivery. The robotics tier recorded here is mobile robots, and sidewalk machines that carry goods to a customer already exist in limited pilots.
Two things hold it back. Trust and permission come first: solicitation rules, building access, vending permits and food-handling licenses are written around a named human being. Second is the economics of a small, variable transaction. Hardware that can climb steps, carry stock, take cash and recover from a shut door is expensive to run, while the human cost here is low and flexible. The cost comparison printed above shows that gap.
How to stay needed in field sales
Lean into the parts of the day that are hardest to hand over. First, the live demonstration: let people touch, taste or try the product. Second, objection handling in person, where tone and timing decide the sale. Third, territory relationships, the repeat customers and local contacts who buy because they know you.
Two skills raise your floor. One is using the sales tools well, so lists, routes and follow-ups cost you minutes instead of hours. The other is payments and compliance, including permits and clean record-keeping, which makes you the person a brand or a market trusts with cash and a license.
If you want adjacent options, the closest work sits in the same family. Compare this job with Demonstrators and Product Promoters, which keeps the in-person demo and drops the cold approach, or Telemarketers, where scripted calling faces very different pressure. Steadier storefront work shows up under Retail Salespersons. The wider group is other sales and related workers, and demand patterns across the trade are on the retail sector page. You can also put any two of these side by side, or check where this job sits among jobs expected to shrink.