We reviewed 25 publicly available AP job postings across five industries to identify which manual tasks still dominate AP roles — and which are realistic candidates for automation.
What 25 AP Job Postings Reveal About Where Teams Are Wasting Time
Why We Read 25 AP Job Postings So You Don't Have To
Most conversations about accounts payable automation start with industry surveys and vendor-supplied benchmarks. The numbers get cited, the slides get built, and somewhere in a conference room a finance leader nods along — then goes back to a team still manually keying invoice data at 4pm on a Tuesday.
We wanted a different starting point. One that was harder to spin.
So instead of reaching for benchmark reports, we pulled 25 AP job postings from publicly available listings on Indeed, LinkedIn, and ZipRecruiter during a single two-week collection window. We looked for postings from a range of company sizes and sectors — the sample included postings from manufacturing, professional services, healthcare, distribution, and retail, though not all sectors were equally represented. We selected postings by searching for the job titles "AP Clerk," "Accounts Payable Specialist," and "AP Coordinator," and took the first results returned in each title search without filtering by employer rating or company size.
We read them the same way a skeptical operations manager would: looking for what employers were actually asking people to do. Not software proficiency. Not "strong communication skills" boilerplate. The actual duty bullets — the day-to-day task lists that reflect what real AP departments have decided requires a human being, repeating it, indefinitely.
This approach has real limitations worth stating plainly. 25 postings is not a statistically representative sample, the collection method was informal, and job posting language doesn't always reflect how work is actually performed. This analysis won't hold up in an academic paper, and we're not presenting it as rigorous research. What it gives you is a ground-level signal — a snapshot of what employers in the labor market were writing down as the core of an AP role during that window. When the same tasks show up across multiple unrelated postings, that repetition is worth paying attention to, even if it can't be generalized with confidence.
It suggests those tasks are still being done by hand, across industries. And that's a reasonable place to start an automation conversation.
The Five Duties That Appeared Most Frequently
Across the 25 postings, five categories of work appeared with enough consistency to be worth examining as a group. The counts below reflect how many of the 25 postings included language that clearly matched each category. These counts represent our reading of the postings; a different analyst might code some borderline cases differently.
| Duty Category | Postings Where It Appeared (of 25) | How It Typically Appeared | |---|---|---| | Invoice data entry and coding | 23 | "Enter invoices," "code to GL account," "match to PO" | | Three-way match processing | 18 | Listed as a core duty, not an occasional task | | Vendor communication and follow-up | 20 | "Resolve discrepancies," "communicate payment status" | | Payment processing and scheduling | 19 | Preparing runs, processing checks, initiating ACH | | Month-end close support | 15 | Reconciling subledgers, preparing accruals, aging reports |
1. Invoice data entry and coding (23 of 25 postings)
Nearly every posting included some version of "enter invoices into accounting system," "code invoices to the correct GL account," or "match invoices to purchase orders." The phrasing varied. The task didn't. Someone is manually moving invoice information from one place — a PDF, an email, a paper document — into a system, and making a judgment call about how to categorize it.
2. Three-way match processing (18 of 25 postings)
Matching invoices against purchase orders and receiving documents appeared in 18 postings, often described as a core duty rather than an occasional task. This is time-consuming work that requires pulling information from multiple systems, and it grinds to a halt the moment any of the three documents don't align cleanly.
3. Vendor communication and follow-up (20 of 25 postings)
"Respond to vendor inquiries," "resolve invoice discrepancies," "communicate payment status to vendors" — this appeared in 20 of the 25 postings, often listed as a separate duty from data entry. That separation suggests it consumes enough time to warrant its own line item in the job description.
4. Payment processing and scheduling (19 of 25 postings)
Preparing payment runs, processing checks, initiating ACH payments, and matching disbursements to invoices were requirements in 19 postings. The mechanical execution of getting money out the door is still, in many organizations, a hands-on task.
5. Month-end close support (15 of 25 postings)
Reconciling AP subledgers, preparing accruals, pulling aging reports, and supporting the close process appeared in 15 postings. The sample included postings at different seniority levels — AP Clerk, AP Specialist, and AP Coordinator — though the distribution was uneven (roughly 8 clerk-level, 11 specialist-level, and 6 coordinator-level postings). Month-end close language appeared across all three title categories, but it was more common in specialist and coordinator postings than in clerk-level ones.
What's notable about this list is not that these tasks exist. Of course they do. What's notable is that they're still being written into job descriptions as primary duties, not exceptions. These are not edge cases that require specialized human judgment. They are the routine infrastructure of AP departments — and they appear to still be largely manual in many organizations.
What Repetitive Job Requirements Tell You About Systemic Process Gaps
When you see the same task listed as a duty requirement across multiple unrelated employers, you're not necessarily looking at a quirk of one company's workflow. You may be looking at a widespread assumption that a certain type of work has to be done by a person, repeatedly, without significant automation support.
Job postings are, in this sense, a relatively honest document. Employers generally don't list duties they've already automated. Nobody writes "manually rekey invoice totals from vendor emails" into a job posting if a system is handling that. The duty list is a map of what the organization believes still needs human hands on it.
The pattern in these 25 postings suggests that many AP teams have adopted accounting software — ERP systems, payment platforms — but haven't meaningfully automated the inputs to those systems. The software exists. The integrations between documents and data fields often don't. So someone bridges that gap manually, every day, for every invoice.
This is a specific and potentially solvable category of process gap. It's not necessarily about lacking technology — most organizations already have the downstream systems. The gap tends to be in what happens before the data gets there: the capture, the classification, the validation, the matching. These steps are where manual labor concentrates, and they're what keep showing up in job postings as primary duties.
The Hidden Cost of Keeping These Tasks Human
The obvious cost is labor. If an AP specialist spends a meaningful portion of each day on tasks that don't require their judgment — data entry, status emails to vendors, routine payment scheduling — that's capacity that could be redirected to work that actually benefits from human attention: exception resolution, vendor relationship management, process improvement.
The less obvious cost is process latency. Manual workflows have natural delays built in. Invoices sit in email inboxes. Data entry happens in batches. Matching waits until someone gets to it. Those delays accumulate in ways that are easy to undercount — late payment fees, missed early payment discounts, strained vendor relationships, close processes that run longer than they should.
There's also the error surface to consider. Manual data entry creates opportunities for mistakes. Transposed figures, wrong GL codes, duplicate payments — these aren't signs of a bad employee, they're predictable outputs of any process that requires humans to move data between systems repeatedly. Automation doesn't eliminate errors entirely, but it does remove the categories of error that come specifically from repetitive manual transcription.
Finally, there's the talent allocation problem. People who are good at AP — who understand vendor relationships, can work through disputes, can spot pattern anomalies in an aging report — may not want to spend most of their day keying invoice numbers. That's worth considering as both an efficiency and a retention issue.
Which AP Tasks Are Actually Good Candidates for Automation
Not every task in an AP workflow is equally automatable. The candidates that tend to make the most operational sense share a few characteristics: they're high-volume, they follow a consistent structure, and the cost of occasional errors is manageable with human review.
Invoice capture and data extraction is often the clearest starting point. When invoices arrive in a consistent enough format — even allowing for variation — technology can extract the relevant fields without a human rekeying them.
GL coding for routine invoices is automatable for a portion of invoice volume. If your organization consistently codes utilities to the same account, or if a vendor always maps to the same category, a rules-based or pattern-learning system may handle the classification without human input on every transaction. The right proportion depends on your invoice mix.
Three-way match on clean transactions — where the PO, receipt, and invoice align within tolerance — is well-suited to automation. Human review should focus on exceptions, not on confirming that every matching transaction matched.
Payment run preparation and the mechanical steps of scheduling and executing payments within approved parameters can follow automated rules once upstream approvals are complete.
Vendor status communications — the routine "your invoice is in process" or "payment scheduled for Friday" responses — can often be handled systematically, reducing inbound vendor calls and freeing AP staff from email queues.
When you audit your current AP workflow, the question to ask about each manual step is: does this task require human judgment, or does it require a human to do something a well-configured system could do just as reliably?
Which AP Tasks Should Stay Human — and Why
There's a real risk in automation conversations of overclaiming what technology can handle. Some AP work is genuinely judgment-intensive, and getting it wrong matters.
Exception handling and dispute resolution is the clearest example. When an invoice doesn't match the PO, when a vendor disputes a deduction, when a payment has been applied incorrectly — these situations require context, relationship awareness, and often negotiation. A system can flag the exception. A human has to resolve it. Automation that blurs this line creates new problems faster than it solves old ones.
Vendor relationship management lives in AP even if it doesn't always get named that way. How your team handles a vendor call, how they communicate about a delayed payment, whether they flag a pattern of overbilling — these interactions have downstream consequences that don't reduce to a rule.
Policy judgment calls — whether a particular invoice meets your organization's purchasing policy, whether an exception to standard payment terms is warranted, whether a new vendor has completed proper onboarding — require someone who understands the organization's intent, not just its documented rules.
Fraud detection and anomaly assessment benefits from automation as a first screen, but the final call on whether something looks genuinely wrong should involve a human with enough context to evaluate it. Pattern recognition tools can surface suspicious transactions; they can't replace the judgment of someone who knows your vendor base and your typical spend patterns.
The honest framing: automation handles volume and structure well. Humans handle variation, ambiguity, and relationship well. The best AP workflows are designed around that distinction, not around the assumption that one replaces the other.
How to Audit Your Own AP Workflow Before Talking to Any Vendor
Before you evaluate any tool, before you take a vendor demo, before you decide what to automate — you need a clear picture of what your AP team is actually doing and where the time goes. Otherwise you're making purchasing decisions against a hypothetical workflow instead of your real one.
A useful workflow audit doesn't require specialized software. It requires structured observation and honest documentation.
Step 1: List every task your AP team actually performs in a typical cycle — not the policy-compliant ideal version, but what actually happens. Talk to the people doing the work. Ask them what takes the most time, what they do the same way every time, what slows them down when it slows them down.
Step 2: Sort tasks along two dimensions. How rule-bound is the task — does it always follow the same logic, or does it require case-by-case judgment? And how frequently does it occur — is this a daily volume driver or an occasional exception? High-frequency, rule-bound tasks are your automation candidates. Low-frequency, high-judgment tasks are not.
Step 3: Trace a representative sample of invoices through your process from receipt to payment. Document every touchpoint — every system it enters, every person who handles it, every place it waits. Count the handoffs. Count the systems. Note where the process pauses and why.
That documentation gives you something real to bring to any vendor conversation — specific volumes, specific handoff points, specific failure modes. It also gives you a baseline you can measure against after any process change.
What to Do With This Information Next
Reading 25 job postings won't tell you exactly what to automate in your AP department. Your invoice volume, your vendor mix, your ERP setup, your team's current capacity — those specifics matter more than any general pattern.
What the job postings do suggest is where to look first. If the same tasks keep appearing in hiring documents across industries, those tasks are likely still manual in many organizations, possibly including yours. And if they're manual, they're worth examining honestly: is that because they require human judgment, or because nobody has yet made the case for changing them?
The difference between those two answers is significant. The first is a workflow that's working as designed. The second is a workflow that's waiting for someone to make a decision.
The practical next step is not a vendor evaluation. It's the internal audit — the honest documentation of what your team actually does, where the time concentrates, and which tasks you'd be relieved to never hire for again. That's the starting point worth spending time on. Everything after it gets easier when you've done that work first.