Most companies think they’ve automated onboarding because new hires fill out a digital form instead of a paper one. I’ve been building Make.com automations for HR and operations teams for over 20 years, and I want to say this plainly: uploading your paperwork to a portal is not automation. It’s the same manual process wearing a nicer shirt. If you want the full breakdown of what real onboarding automation looks like, I already laid it out in Automating Employee Onboarding the Right Way. This post is the opinion version — why the digitized-paperwork approach fails, and what I’d build instead.

Here’s my thesis, stated flat: digitizing a form is not automating a process. Real onboarding automation means the system moves data between the tools you already use — HRIS, payroll, IT provisioning, benefits, e-signature — without a human retyping anything or chasing an email thread to find out if step three happened. If a person still has to open five tabs and copy information between them, you’ve added software. You haven’t removed work. The companies getting this right build automation first, using it to standardize the process end to end, and only then layer AI on top to handle the messy, unstructured parts — an offer letter phrased ten different ways, a resume format nobody agreed on, a question a new hire asks at 11 p.m. Skip the automation foundation and AI just processes chaos faster.

The Common Wrong Approach

Walk into most mid-size companies and ask how onboarding works, and you’ll hear some version of: “We use a portal.” New hires log in, fill out their W-4, upload their I-9 documents, e-sign the handbook acknowledgment. It feels modern. It looks like progress on a demo call. And it is a genuine improvement over a manila folder of paper forms mailed to a new employee’s house.

But watch what happens after the new hire hits submit. An HR coordinator gets an email notification. She opens the portal, downloads the PDF, and re-keys the data into the HRIS. Someone else pulls the same information into payroll. IT gets a separate ticket, manually filled out by HR, to provision a laptop and set up accounts. Benefits enrollment happens through yet another system that doesn’t talk to any of the first three. Every one of these steps still runs through a human transcribing information a computer already has.

I see this pattern across companies that will tell you, with complete confidence, that they’ve “automated onboarding.” They haven’t. They’ve replaced a paper form with a web form. The document changed. The workflow around it didn’t move an inch. For a deeper look at which tasks fall into this trap most frequently, I wrote up the specific list in 9 Employee Onboarding Tasks You Should Never Do Manually in 2026. And if paperwork specifically is your bottleneck, I broke down the fix step by step in How to Automate New Hire Paperwork.

Why It Fails

Digitized paperwork fails for one structural reason: it automates the input and leaves the handoffs manual. The form is the easy 10% of the process. The hard 90% is what happens to that data next — does it reach payroll correctly, does IT provision the right access on day one, does the manager know the hire is starting Monday, does compliance get the signed documents filed where an auditor can find them two years later.

None of that gets solved by a nicer form. It gets solved by connecting systems so data flows without a person in the middle re-typing it.

I watched this fail in slow motion with a client I’ll call David, an HR manager at a mid-market manufacturing company. Someone transcribed a new hire’s salary by hand from one system into another and fat-fingered it: $103K became $130K. Nobody caught it for months. The overpayment ran to $27K before it surfaced, and the employee — who had done nothing wrong, who had just been paid what the system told him he’d earn — quit over how the correction was handled. That is not a training failure. That is not a “our people need to be more careful” failure. That is what happens when your onboarding process runs on a human being as the integration layer between two systems that should be talking to each other directly.

The other failure mode is slower and quieter: adoption. Teams roll out a beautiful new onboarding portal, and six months later half the department is still emailing spreadsheets around because the portal doesn’t touch the tools people actually live in every day. A tool nobody adopts isn’t automation — it’s shelfware with a login screen. I go deeper on why this specific failure mode repeats itself in Manual vs. Automated Onboarding.

The Better Approach

Automation-first, then AI. That’s the whole model, and I don’t say it as a slogan — I say it because I’ve watched what happens when companies do it backwards. They buy an AI tool to “handle onboarding” before they’ve fixed the plumbing, and the AI ends up guessing its way through a broken process just as fast as a human did. Speed isn’t the problem. Structure is the problem.

The right build order looks like this. First, connect your systems — HRIS, payroll, e-signature, IT provisioning, benefits — through something like Make.com so that when a new hire’s offer is accepted, that single event triggers everything downstream automatically: the personnel file gets created, payroll gets the correct salary and start date pulled directly from the source record (no re-typing, no transcription errors), IT gets a structured provisioning request with the right role and access level already specified, and the manager gets notified with a checklist scoped to that specific new hire.

Second, once that structure exists, AI earns its place handling the parts that don’t fit a rigid template — reading an offer letter that was drafted slightly differently than usual and pulling the right fields anyway, answering a new hire’s benefits question at 9 p.m. based on your actual plan documents, flagging when a start date and a background check clearance are about to collide. AI is excellent at unstructured judgment calls layered on top of clean data. It is bad at being the data pipeline itself. If your systems don’t talk to each other yet, read What Is Onboarding Automation? for the plain-language version of where to start.

Adoption-by-design is the other half of this, and it’s the half most vendors skip because it isn’t a feature they can put on a slide. The goal isn’t to hand people a new system to learn. The goal is to make the systems they already use — Slack, their inbox, their HRIS — start talking to each other invisibly, so the new hire’s manager doesn’t have to open a new app; the notification just shows up where they already work. Automation that requires training material to get adopted is automation that will die in eighteen months when the one person who understood it leaves.

The numbers back this up when you connect the whole chain instead of one link of it. TalentEdge rebuilt their onboarding process this way and posted $312K in annual savings — a 207% ROI — not because they bought a smarter form, but because they stopped paying people to be the connective tissue between systems. That’s the case study version of everything I’m arguing here, and you can read the full breakdown in the TalentEdge case study.

What I’d Do Differently

If I were walking into a company tomorrow that believed it had “automated onboarding” because it had a portal, here’s exactly what I’d change.

I’d start by mapping every handoff in the current process — not the form, the handoffs. Every place a piece of information moves from one system or one person to another. That map is almost always uglier than anyone in the room expects, and it’s the single most useful diagnostic I know for finding where the manual re-entry is hiding.

Then I’d pick the highest-risk handoff first — usually compensation data, because that’s where a transcription error turns into a $27K overpayment and a good employee walking out the door — and automate that connection before touching anything else. Prove the model on the highest-stakes link in the chain, not the easiest one.

From there, I’d build out the rest of the trigger chain in Make.com: offer acceptance fires payroll setup, IT provisioning, benefits enrollment, and manager notification, all from one event, all without a human retyping a single field. Only after that foundation is solid would I introduce AI — to handle the offer letters that don’t match the template, to field the after-hours questions, to catch the edge cases a rigid workflow can’t anticipate.

And I’d build every piece of it inside tools the team already opens every day. No new login. No training deck. If the automation is doing its job, most of the team should barely notice it’s there — they’ll just notice that onboarding stopped generating fire drills. That’s the actual standard. Not “we digitized the form.” Not even “we bought an AI tool.” The standard is: does a new hire start on day one with payroll correct, IT access ready, and a manager who already knows what to do, without anyone on your team re-typing a single piece of data to make it happen. If you’re not there yet, you haven’t automated onboarding. You’ve just given your paperwork a better font. For more on the questions I get asked most on this, see the Onboarding Automation FAQ.

Independent research backs the stakes here. SHRM has documented for years that structured onboarding programs drive measurably better new-hire retention and productivity than ad hoc processes, and Gartner has flagged process automation as a top lever HR leaders can pull to cut administrative overhead without adding headcount. Harvard Business Review has made a similar case about first impressions: the first ninety days shape whether a new hire stays two years or two months, and a process full of manual friction is the fastest way to blow that window. None of that research says “buy a portal.” It says fix the structure. That’s the whole argument.