Between late August and early September 2026 we collected 120 live job postings from B2B software companies in the US, UK and EU, all for implementation, onboarding, integration or data migration roles. We were not looking for candidates. We wanted to know, in the companies' own words, what these people are hired to do.
The answer is more uniform than the job titles suggest. Strip away the company names and a large share of the postings describe one task: take the data file a client exported from their own system, work out what is in it, map it into ours, validate it, and load it. Then do it again for the next client.
What the postings say
A few numbers from the set. 39 percent of postings explicitly mention data or field mapping as a duty. 29 percent name SFTP or FTP as a transfer method the candidate must handle. 22 percent name CSV, Excel or spreadsheets as the working material. 25 percent are framed around migrations or conversions from legacy systems, which in practice means the same file work with a bigger first batch. 15 percent mention validation. Only 7 percent mention EDI or standardized formats such as 834, which tells you how much of this world runs on plain files with no standard at all.
The title families were: implementation (33 postings), integration (32), data or solutions (19), onboarding (11), migration or conversion (9), and a long tail of hybrid titles. The pattern held across sectors: HR and benefits platforms, edtech, fintech, healthcare software, vertical SaaS for construction, dealerships, waste hauling and law firms.
In their words
The most striking part was the language. These are verbatim fragments from different postings, companies removed: "a roster file with the wrong school year, a missed character in an SFTP path." "Receive and review client-provided data files in a variety of formats (CSV, Excel), troubleshoot import errors and data mapping issues." "Extracting source data from external softwares and transforming it to fit our target data schema using tools such as Excel, Power Query, AI and others." "Unpicking a messy HRIS export." "Create the translation guides that map each firm's data into the master template." "Unprocessed or missing file review."
Read them together and a job description emerges that nobody would write on purpose: professional file wrangler. It is skilled work, it is client facing, and it does not scale, because every new client brings a new layout and every quarter brings a changed one. Our guide on customer data onboarding covers why this step is where onboarding timelines go to die.
Why companies hire for it instead of automating it
Three reasons come up when we talk to the teams behind these postings. First, the work looks like a people problem: a client sends a file, a person deals with it, so the fix is another person. Second, in-app importers were bought to solve it and did not, because the clients who cause the most work never log in to upload anything; they drop files on SFTP or email them. Third, the mapping feels bespoke every time, so nobody believes it can be systematized. We wrote about the mechanics of that in how to map customer CSV files automatically.
The counterargument is in the postings themselves. If 39 percent of these roles are mapping, and the mapping for a given client is stable once agreed, then most of the hours are spent re-doing a decision that was already made. That is the part software is good at. Learn the layout once per client, have a human approve it, then run every later file through the same mapping with validation in front and a clean delivery behind. The person in the role keeps the part that needs judgment: the first file, the exceptions, the client conversation.
What we are doing with this
We built FileFeed for exactly the task these postings describe: client files arrive over SFTP or email, get mapped to your schema once per client, validated on every run, and delivered as clean JSON to your API, with the run log to show what happened. If your team is about to open one of these roles, or already has one buried in spreadsheets, the fastest way to check whether the file work can come off their plate is to send us one real client file. It comes back mapped the same day. The mechanics of the pipeline are in our SFTP to webhook guide.
Related resources