Key Takeaways
- Transformance tests AR automation in four gates, in order: connection and data checks, user acceptance testing (UAT) on your own payments, a parallel run and a cutover rehearsal.
- UAT for cash application covers the payments that cause the work: lump sums, short pays, missing remittances, duplicate files, locked periods.
- A parallel run is safe when only one side can post: your team stays authoritative while Vero, Transformance’s AI agent, reads and proposes.
- Exit criteria are set per payment segment in week 0, not as one blended rate.
- The cutover runbook gives every step an owner, a backup, a verification and a rollback, and is rehearsed outside period close.
In This Article
- Key Takeaways
- How do you test AR automation before go-live?
- What is the difference between UAT, a parallel run and a cutover rehearsal?
- Which scenarios should UAT cover for cash application and collections?
- How do you run a parallel run without double posting?
- What are the exit criteria for the go/no-go decision?
- What data and environment do you need for the tests?
- What does a cutover runbook for AR automation contain?
- How does onboarding with Transformance handle testing?
How do you test AR automation before go-live?
Transformance tests AR automation in four gates: connection and data checks, user acceptance testing on your payments, a parallel run and a cutover rehearsal, each with a named signer.
The test gates: what each proves, the pass rule, who signs
Microsoft’s implementation guide for Dynamics 365 says “Do SIT before UAT” (Microsoft Learn, 2026) and calls UAT “the final type of test that you do before you deploy your solution to production” (Microsoft Learn, 2026). Each gate tests what the next relies on. The thresholds are yours: no neutral benchmark exists. For the calendar, see how long an AR automation rollout takes.
Set the gates up in six steps:
- Write pass rules per payment segment in week 0 (owner: controller).
- Name the posting approver and the exception owner (owner: finance sponsor).
- Prepare the answer key: your team’s matches for a past period (owner: cash application lead).
- Run UAT in staging; fix and retest, do not just rerun (owner: AR lead).
- Run the parallel run with one side unable to post; review weekly (owner: controller).
- Rehearse the cutover at least once, outside period close (owner: project lead).
Takeaway: a test without a pass rule written beforehand is a demo.
What is the difference between UAT, a parallel run and a cutover rehearsal?
UAT checks scripted scenarios against expected results; a parallel run compares live results of old and new process; a cutover rehearsal checks the switch, its timing and its rollback.
Microsoft’s guide describes UAT as “always a manual test, and it’s done by business users in a dedicated and integrated test environment”, with the purpose “to get sign-off and approval from the business stakeholders” (Microsoft Learn, 2026).
Gate 1 comes first. Microsoft’s checklist says to “Simulate external systems that are down” (Microsoft Learn, 2026); in AR terms, a late bank file, an unavailable ERP, a revoked mailbox. At Transformance, gate 1 starts with the standard seven-test integration plan: integration agents run it in a test environment and report, and a person signs each test (how AI agents build and test an ERP integration). Then open items and balances are tied out to the ERP.
Takeaway: UAT proves it can; the parallel run proves it does; the rehearsal proves you can switch.
Which scenarios should UAT cover for cash application and collections?
Cover the payments that break rules, not only the ones that match: twelve scenarios, each with the expected result written down before the test runs.
UAT catalogue for cash application: 12 payment scenarios and the expected result
Amounts are illustrative. Write each expected result as your ERP shows it (what AR automation reads from and writes to your ERP). Rule-based versus AI cash application shows where rules stop; matching a payment with no reference shows the evidence a proposal should carry.
Microsoft’s checklist adds “Don’t use a general security role or assign the System Admin role to all users” (Microsoft Learn, 2026). Give each tester their production role.
If AR collections is in scope, add five: a reminder due after a promise to pay, a disputed invoice, a customer on hold, a wrong contact, and the customer’s language. Each expected result follows your policy.
Takeaway: test the payments that cause the work, not only the ones that match.
How do you run a parallel run without double posting?
Keep your current process authoritative, let the new system read and propose only, and compare every proposal with what your team did; nothing posts without human sign-off.
- Scope. One entity, its top payers, plus hard cases from the UAT catalogue.
- Window. At least one full month-end cycle, so period-end behavior is seen. This is guidance; no neutral standard exists.
- Method. Fixed before day one: customer identified, invoices selected, difference handling, what stays unapplied.
- Review. Weekly, with the controller.
At Transformance, the parallel run works this way. During testing, Transformance’s access to your ERP is read-only; write access starts at rollout. Vero, Transformance’s AI agent, investigates each bank line from Customer 360 and proposes a match with its evidence. Your team applies cash as before.
Every difference gets a class and, as Microsoft advises for UAT items, “an owner and a completion date” (Microsoft Learn, 2026):
- Configuration defect: a rule or tolerance set wrong. Fix and retest.
- Data defect: a missing payer link or wrong bank account (master data cleanup before cash application).
- Judgement call: two answers are defensible. Record which one your policy wants.
Illustrative: a parent company pays $48,600.00 for three invoices of a subsidiary. Your team applied all three; the proposal finds two, because the master data does not link the payer to the third. Class: data defect. Owner: AR lead.
Takeaway: a parallel run is only safe when one side cannot post.
What are the exit criteria for the go/no-go decision?
Set exit criteria per payment segment before testing starts, and go live only when each is met or its gap has a named cause, owner and date.
Use four segments, because a blended rate hides the hard ones: structured remittance, email or PDF remittance, no remittance, and short pays with multi-invoice payments.
Parallel run: what to compare and example exit rules (set your own)
To define a rate consistently, see how to measure the cash application auto-match rate.
Microsoft asks teams to “Complete all test cycles with exit criteria met and signed off by the business stakeholders” (Microsoft Learn, 2026), and calls the go-live readiness review “a quality gate for deployment” (Microsoft Learn, 2026). The controller or AR lead and the IT owner sign; security where your policy requires it.
Takeaway: agree the exit criteria in week 0; changing them after the results arrive turns a test into a rubber stamp.
What data and environment do you need for the tests?
Use your own transactions in a production-like staging environment: a few months of payments and open items, hard cases included, with personal data handled under your data-protection rules.
- Bank lines with your team’s matches: the answer key.
- Open items as of each test date.
- The customer master, with payer links and bank accounts.
- Sample remittances in every format you receive.
The full file list is in what data an AR automation vendor needs to start.
Microsoft’s guide says “UAT requires customer data, including migrated data” (Microsoft Learn, 2026).
With Transformance, data is hosted in your region, VPC deployment in your own cloud is available, and no customer data is used to train models. Testers sign in through your SSO/SAML. Use only the personal data the test needs.
Takeaway: test on last quarter’s real payments, with the answer key your team already produced.
What does a cutover runbook for AR automation contain?
A runbook lists each cutover task in order with an owner and backup, a verification step, a sign-off and a rollback, and you rehearse it before the real day.
Cutover runbook for AR automation (example)
Microsoft’s cutover guidance asks for an owner and backup owner, verification and sign-off steps and “A rollback plan in case something goes wrong”, advises a mock cutover “several times” and says “Don’t do the cutover when you have to close your books or report your results” (Microsoft Learn, 2026).
Rollback in AR terms: withdraw write access, which returns Transformance to the read-only access of testing, and process payments as before. Postings already approved are reversed through the ERP’s own route, and the audit trail records what the system did and who approved it (audit trails and reversals for AI agents).
Hypercare follows, “an elevated level of support right after go-live” in Microsoft’s words (Microsoft Learn, 2026): in AR, a daily review of exceptions and unapplied cash.
Takeaway: the cutover is boring when you have rehearsed it.
How does onboarding with Transformance handle testing?
Testing is part of one shared onboarding project: scenarios agreed by both sides, tests on your own transactions in staging, a parallel run, an acceptance report and sign-offs before go-live.
In Transformance’s onboarding hub, both sides work in one project:
- Scenarios the automation must pass are agreed and visible to both sides.
- Real data: tests run on your transactions in staging; then the new process runs alongside yours.
- Validation cycles are tracked; issues are triaged and fixes confirmed.
- An acceptance report records what was tested, what passed and what was accepted.
- Readiness gates, a go/no-go decision and formal sign-offs, then a cutover runbook.
- A standard hypercare period follows go-live, with support tickets and KPIs tracked from day one.
Every integration passes the same seven-test plan, run by Transformance’s integration agents and signed by a person. A person approves everything that reaches you or production, within thresholds you set (approval thresholds for AI agents), and every action is in the audit trail.
For timing, see how long an AR automation rollout takes.
Takeaway: the gates on this page are the gates in the project, signed by your people.
Who Transformance is for
- ERPs: in production with customers on SAP S/4HANA (on-premises, private or public cloud) and SAP ECC, NetSuite (including OneWorld subsidiaries), Microsoft Dynamics 365 Finance & Operations and Business Central, and Odoo; Oracle and Infor supported; other ERPs through our integrations hub, each tested against your own data in a test environment.
- Size: mid-market to large enterprise, 250 to 2,000+ employees, including groups with several entities and shared service centers.
- Payment volume: from under 1,000 to over 100,000 payments a month.
- Industries: manufacturing, consumer goods and food and beverage, logistics, e-commerce, agriculture, energy and professional services.
- Scope: cash application alone, or the full order-to-cash (cash application, deductions, AR collections, cash forecasting).
Conclusion: write the pass rules first, then test
AR automation is ready for go-live when four gates have passed against rules written before the first result, each signed by a named person. Book a Call to walk through the test plan for your payment mix.
Key terms
- Cash application: matching incoming payments to open invoices and posting the result.
- Payment reconciliation: agreeing each payment with the invoices it pays.
- Remittance advice: the customer’s note of which invoices a payment covers.
- Short pay: a payment below the invoice amount, often with a deduction.
- On-account cash: cash posted to the customer’s account, not to an invoice.
- Audit trail: who did what, when and on which input.
Related guides
- Which ERP systems does Transformance integrate with?
- How long does AR automation implementation take?
- How do AI agents build and test an ERP integration?
- What data does an AR automation vendor need to start?
- What master data cleanup is needed before automating cash application?
- Audit trails and reversals for AI agents
Frequently Asked Questions
How do you test AR automation before go-live?
In four gates: connection and data checks, user acceptance testing on your own payments, a parallel run, and a cutover rehearsal, each with a pass rule and a named signer.
How long should a parallel run last?
Long enough to include a full month-end cycle. No neutral standard exists, so set it from your payment cycle and your week-0 exit criteria.
Can a parallel run create duplicate postings?
Not if one side cannot post. Your process stays authoritative while Vero reads and proposes; during testing, Transformance’s ERP access is read-only, and nothing posts without human sign-off.
Which ERPs and editions can you test against?
Your own. Transformance runs in production with customers on SAP S/4HANA, SAP ECC, NetSuite, Dynamics 365 Finance & Operations and Business Central and Odoo, and supports Oracle and Infor.
Does this work for a group with many entities and 100,000 payments a month?
Yes. Test one entity first, then add entities. Transformance serves groups with several entities, from under 1,000 to over 100,000 payments a month.
How long until we can start testing and when do we go live?
The ERP connection typically takes 2-3 days, and your IT team may be involved. A proof of value takes 4-5 weeks on one entity, and a full rollout takes up to 4-8 weeks (Transformance customer data, 2026).
What does testing cost?
It depends on scope and the number of entities. Book a Call and we size it with you.
SAP, SAP S/4HANA, Microsoft, Dynamics 365, Business Central, Oracle, NetSuite, Odoo and Infor are trademarks of their respective owners (SAP SE, Microsoft Corporation, Oracle Corporation, Odoo S.A., Infor). Transformance is not affiliated with or endorsed by them.




