- Most ERP frustration comes from the manual steps around the ERP, not the ERP itself.
- Spreadsheets survive next to ERPs because they solve real problems: exceptions, speed and flexibility.
- Integrate before you replace. Build custom software only where your process is genuinely different. Add AI where the work is reading and sorting information.
- Start with one frequent, painful, measurable workflow.
Ask people in an industrial business what is wrong with their ERP and you will often hear a list of complaints about things the ERP does not actually do. The quote that has to be retyped. The approval that lives in someone's inbox. The monthly report nobody trusts until it has been through the master workbook.
That is the real subject of this article. ERP workflow automation means connecting your ERP to the people, systems and documents around it, so that approvals, data transfers and status updates happen by rule instead of by export, email and re-entry. The ERP stays where it is. The manual layer around it gets replaced.
10 Ways to Fix the Manual Workflows Around Your ERP
ERP systems are rarely the whole problem
An ERP is very good at one job: holding a consistent, controlled record of transactions. Orders, stock, purchase orders, invoices, ledgers. It is not designed to know that one customer needs a different approval route, or that a supplier sends confirmations as scanned PDFs, or that engineering changes need a human to check three other things first.
Surveys keep showing the same pattern. Rockwell Automation's 2026 study of 1,560 manufacturing decision-makers found that 93% have a manufacturing execution system in place, but only 23% report full integration across ERP, PLM, quality and operational technology systems. The software is there. The connections are not.
The hidden spreadsheet layer around enterprise applications
Almost every company with an ERP also has a second, unofficial system made of spreadsheets. It has an owner, usually one person. It has formulas nobody wants to touch. It is where the real answer gets worked out before it goes back into the official system.
Finance shows this clearly because it has been measured. A September 2026 survey of 215 finance leaders by Spendesk and CFO Connect found that for four out of five teams, planning and forecasting is still mainly done in spreadsheets, and 55% still track cash in them. The AFP 2026 FP&A Benchmarking Survey of 332 finance professionals found the average company still takes 8.7 weeks to produce a budget, unchanged from three years earlier. In the 2025 edition, 96% of FP&A professionals used spreadsheets for planning even though 71% also had a dedicated planning platform.
Operations teams have their own version: the estimating workbook, the production tracker, the supplier chase list. Different department, same layer.
Why manual workarounds develop
Nobody sets out to build a shadow system. It grows one sensible shortcut at a time.
The ERP is slow to change:
A small change request can mean a vendor quote, a release cycle and a regression test. A spreadsheet takes an afternoon.
Your process is not standard:
Customer-specific terms, product variants and approval rules rarely fit the default module.
Information arrives from outside:
Suppliers, customers and site teams send emails, PDFs and photos, not clean records.
Judgement is needed:
Some workflows require context and human approval, particularly when customer-specific rules or exceptions are involved.
Integration was out of scope:
Implementation projects deliver the core. The edges are left for later, and later becomes permanent.
A workaround that works is almost impossible to justify removing, because it appears to cost nothing. The cost is spread across hours, errors and delays that nobody tracks.
Typical industrial examples
These six turn up in manufacturing, engineering and field-service businesses again and again.
Quote to order:
An approved CRM quote is retyped into the ERP as a sales order, often with the pricing rules checked by memory.
Purchase approvals:
Requisitions are exported, emailed to approvers and re-entered once someone replies.
Engineering changes:
A revised drawing is released, but the bill of materials update waits for someone to notice.
Job and work-order status:
Technicians report by phone or WhatsApp; the office updates the ERP at the end of the day.
Supplier invoices:
PDF invoices arrive by email and are matched to purchase orders by eye.
Project cost reporting:
Each month, costs are exported, merged in a master workbook and reconciled before anyone can read them.
The operational cost of disconnected workflows
The most-cited number is from Parseur and QuestionPro's 2025 survey of 500 U.S. professionals. Respondents spent more than nine hours a week moving data between emails, PDFs, spreadsheets and systems, which the study valued at an average of $28,500 per employee per year. 56% reported burnout from repetitive data tasks. It is a self-reported, cross-industry figure, so read it as an indicator, not a forecast for your plant.
Your own number is easy to estimate: people involved × hours per week × loaded hourly cost × 48 weeks. Then add the costs that do not fit in a formula:
Delay:
Orders and approvals wait for the person who does the re-entry.
Errors:
Every retyped value is a chance for two systems to disagree.
Key-person risk:
The workbook only works because one colleague understands it.
Weak audit trail:
An approval in an email thread is hard to prove later.
Slow visibility:
Managers see last week's numbers, assembled by hand.
The example: same process, two ways
Six hand-offs, three people, no record of who changed what.
The approver sees the data, decides, and the ERP updates itself. The decision is logged.
The human decision stays. What disappears is the export, the reformatting, the email chase and the retyping.
When to integrate rather than replace
Replacing an ERP is expensive, disruptive and slow. It is the right call sometimes, but far less often than frustration suggests.
Integrate when
- The ERP holds clean master data
- It enforces the controls you need
- The vendor still supports it
- The pain is in hand-offs and approvals
- The ERP exposes an API, database or file interface
Consider replacing when
- It cannot support your core process
- It is out of vendor support
- Workarounds cost more than the licence
- Master data is unreliable at source
- You cannot extend or connect it at all
If the ERP is old but sound, wrapping it with a modern API often buys years. Our guide to REST, webhooks and FTP explains the options in business terms.
When custom software is justified
Custom software earns its place when three things are true: the process is specific to how you win or deliver work, no packaged tool fits without heavy compromise, and the volume repays the build. An estimating tool built around your own pricing rules is a good example. A generic task tracker is not.
Common candidates are quoting and estimating systems, job and project tracking, procurement workflows, and customer or supplier portals that feed straight into the ERP. If a configured SaaS tool or an ERP setting would do the job, use that instead. Our article on when to buy and when to build goes through the decision. If custom is the answer, see our custom application development and ERP solutions work.
Where AI can help
AI earns its place where the work is reading and sorting information, not moving structured data. In ERP-adjacent workflows that means reading supplier invoices and purchase orders, classifying incoming emails, extracting fields from PDFs and drawings, flagging unusual values and drafting summaries for approvers.
Two cautions. First, AI works best on top of connected systems, so do the integration first. Second, keep a person in charge of anything that changes money, prices or commitments: AI reads and recommends, a human approves, the system executes. Finance teams are adopting AI along these lines. The same Spendesk and CFO Connect survey found 67% now use at least one AI tool, up from 31% in 2024, mostly for analysis, reporting and reconciliation.
See where to start with AI automation, or our pages on AI document processing and adding AI to existing applications.
Not sure which workflow around your ERP to fix first?
Show us the systems, the spreadsheets and the emails. We will tell you honestly whether it needs an API, a workflow tool, custom software or AI.
How to identify a good first automation project
The best first project is not the biggest one. It is the one you can finish, measure and show. Score each candidate on five points:
Frequent:
It happens daily or weekly, not yearly.
Painful:
People complain about it unprompted.
Bounded:
It touches two or three systems, not ten.
Owned:
One named person can decide how it should work.
Measurable:
You can state the current hours, errors or turnaround time.
Quote-to-order handover, supplier invoice matching and purchase approvals usually score well. Processes that depend on a political decision about who owns the data do not, however valuable they look on paper.
Practical transformation roadmap
Map one process:
Pick one process and trace every system, file and email it touches, with the people involved.
Choose the lightest fix:
Integration, automation, custom software or AI, in that order of preference.
Build the smallest release:
One workflow, live with real users, with a named owner.
Measure:
Hours saved, errors, turnaround time, compared against the baseline you wrote down in step 1.
Extend:
Use the result to justify the next workflow, not a platform-wide programme.
Notice what is missing: a platform replacement, a multi-year programme, a new vendor. Each step is small enough to be judged on its own results.
Have a workflow around your ERP that still depends on spreadsheets or email?
Talk to Infomaze. Tell us which systems are involved and where the manual work happens.