Methodology
Every vendor is scored on the same 26 criteria. A criterion is marked met (1), partial (0.5) or not met (0). The score out of 5 is weighted coverage:
score = 1 + 4 × (weight met ÷ 55)
No weight is adjusted vendor by vendor, and nobody pays to appear, to move up, or to have a limitation removed.
What the score measures, and what it does not
It measures declared functional breadth against this grid. It does not measure how well any single function is executed, how good the support is, how fast the software runs, or whether customers are happy. A narrow product that is excellent inside its lane scores low on breadth by construction — that is not a judgement on it. It is why the ranking recalculates the moment you check the criteria you actually need: that ranking is the one to read, not the general one.
How a vendor gets coded
From what the vendor says about itself: product pages, documentation, published numbers, conference material, and — where it exists — what its own users say in public community threads. Where a vendor publishes nothing on a criterion, it scores zero on that criterion, because a buyer cannot verify a silence.
Products in the same family start from a shared baseline for that family, and each entry then records what it genuinely deviates on, with the reason attached to the criterion. Those per-vendor notes are visible on each scorecard — they are the audit trail of this grid.
Where the data comes from
Market Edge JD Edwards research for Aviron, September 2026 — 21 players mapped from vendor pages, JDELIST and Quest threads, and Oracle documentation.
Criteria and weights
| Criterion | Group | Weight |
|---|---|---|
| Runs inside JD Edwards Uses JDE's own technology, screens or tables. Not a separate application that happens to exchange data with it. | How it touches JDE | 3 |
| Creates the voucher in JDE The end product is a posted voucher in JDE with its batch and its match — not a recommendation for a clerk to key in. | How it touches JDE | 3 |
| Goes through paths JDE supports Orchestrator, AIS, business services, table interfaces, Batch Voucher Processor. No core modification. | How it touches JDE | 3 |
| No file transfer, no nightly batch No CSV over FTP, no Groovy script, no fifteen-minute sync. Batch windows are where AP automation quietly dies. | How it touches JDE | 3 |
| Survives ESUs and Tools Releases An upgrade is a test cycle, not a re-integration project. | How it touches JDE | 2 |
| Supports JD Edwards World Not just EnterpriseOne. World shops are the ones the automation market walked away from. | How it touches JDE | 2 |
| Publishes its IT-time cost A number for how much of your IT team's week the install takes. If a vendor will not say, the buyer assumes the worst. | How it touches JDE | 1 |
| Agents that execute, not advise It takes the action. A chatbot that answers questions about JDE is a different product. | What it actually does | 3 |
| AP end to end: capture to payment Document in, voucher out, payment released — not one slice of the middle. | What it actually does | 3 |
| Beyond AP: requisitions, receipts, statements Requisitions, receipts, contract and price validation, vendor statement reconciliation. | What it actually does | 2 |
| Order-to-cash on JDE The sell side. Almost nobody in the JDE ecosystem sells it. | What it actually does | 2 |
| Triages the AP mailbox Watches the shared inbox, classifies what lands in it, and files it against the right supplier. | What it actually does | 2 |
| Judgment on coding, AAI and tax Decides the AAI, the category code and the tax treatment. This is the part OCR with rules cannot do. | What it actually does | 3 |
| Resolves exceptions itself Tolerance variance, PO not found, suspected duplicate — handled, or escalated with a recommendation. | What it actually does | 2 |
| Handles freight and non-PO invoices The cases that fall out of every template: freight, utilities, complex non-PO invoices, tax variance. | What it actually does | 2 |
| Your shop's rules in plain English A readable, editable record of how this shop actually works — which supplier hits which AAI, how freight is coded, which approval goes where. | Knowledge that stays | 3 |
| Corrections become governed rules Autonomy climbs week after week instead of peaking on go-live day. | Knowledge that stays | 3 |
| Starts from your history, not a blank template Mines what your shop has already done, rather than asking you to configure a generic package. | Knowledge that stays | 2 |
| Audit trail on the voucher Documents, reasoning, approver and timestamp attached to the transaction — not in a separate log. | Control and proof | 2 |
| Scoped, revocable access Granted one system at a time by your administrator, narrow, revocable in a click. No global admin account. | Control and proof | 2 |
| SOC 2 Type II Externally audited, with a status you can check. | Control and proof | 1 |
| Hosting region you choose On-premise, your cloud, or a region you name. | Control and proof | 1 |
| Oracle Validated Integration The one badge this audience recognises on sight. | Control and proof | 1 |
| Named JD Edwards customers A named JDE shop on the record. Anonymised counts as half. | Control and proof | 2 |
| Published performance numbers A rate you can quote back to them, with a source. | Control and proof | 1 |
| Present where JDE people are Named, technical answers on JDELIST and Quest. In this market that is worth more than a website. | Control and proof | 1 |
Interest disclosure
STACK ERP is published by Aviron, which builds one of the 24 products compared here. Aviron gets no special treatment in the calculation: it is coded on the same grid, it sits at number 1 on breadth, and the 4 criteria it does not meet are printed on its own scorecard with the reason why — publishes its it-time cost, oracle validated integration, named jd edwards customers, present where jde people are.
If you find a rating that is out of date or wrong, that is worth more to this page than a compliment. Vendors change their pages weekly; this grid is dated September 2026.