Accounts Payable Invoice Automation Software (Forrester)

The Forrester Wave: Accounts Payable Invoice Automation Software, Q2 2026 is the second edition of an evaluation that grades a narrower thing than its counterpart at Gartner does, and the difference in scope is the most useful fact in either document. Forrester scored fifteen providers against twenty-six criteria spanning current offering, strategy and customer feedback. The Magic Quadrant for Accounts Payable Applications that Gartner published on June 18, 2026 scored twelve against its own axes. The two rosters overlap, the two Leader rungs do not, and the two report names describe two different sizes of the same problem.

That difference is worth stating before anything else, because a buyer searching for one of these documents will find the other.

The Forrester Wave: Accounts Payable Invoice Automation Software, Q2 2026, and the word the title adds

Forrester's title names the object of the work. Gartner's names the function. The word "Automation" in Forrester's version is doing specific work: it says the subject of the evaluation is a document moving through a pipeline, and the vendors are graded on how much of that movement they can take off a person's desk.

The scope that follows is deliberately tight. An invoice arrives, gets captured, gets coded to a general ledger account and a cost center, gets matched against a purchase order or a receipt, gets approved, gets scheduled, gets paid. Forrester's twenty-six criteria sit inside that sequence. Gartner's framework starts one level up, at the department, and asks what software runs the accounts payable function, which pulls in payment execution, supplier onboarding, working capital and the analytics layer above all of it.

Neither boundary is the correct one. They answer different questions. An organization that wants to stop keying invoices has a scorecard in Forrester's report. An organization restructuring the whole payables operation has one in Gartner's.

What the two documents share is timing. Both are second editions published in 2026. By the second edition the analysts have a before and an after to grade, which changes what the reports are able to say. A first edition can only establish who is present. A second can measure who moved.

Gartner scored the same ground in June under a wider name

The Magic Quadrant for Accounts Payable Applications published on June 18, 2026 covering twelve providers, credited to Miles Onafowora and David Condon. The first edition under that name published on March 19, 2025, which makes the June document the second pass.

Its Leader definition is the part worth reading closely. Gartner requires AI, machine learning, generative AI and agent capabilities built natively into the product, and then names what those capabilities have to do: autonomous capture and coding, dynamic exception handling, fraud and anomaly detection, and predictive cash and working capital optimization.

Notice what is absent. Speed is not in the definition. Neither is invoice volume, nor the touchless rate, which was the headline metric in this market for a decade. Gartner's bar describes a system that makes decisions and defends them, not one that moves paper faster.

Forrester reached the same conclusion from the other direction. Its framing for the Q2 2026 edition describes agentic AI shifting this category from back-office automation to a proactive finance control layer. Two firms, two names, two vendor counts, and one shared verdict about what the software is now for.

Both firms moved the bar from capture speed to how much of the decision a system owns

The metric that used to be quoted in every AP sales conversation was the percentage of invoices processed without a human touch. It was easy to measure and easy to inflate, because a system that clears only clean, matching, low-value invoices can post a high number while the difficult third of the volume stays exactly where it was.

Both 2026 reports replace that metric, and they replace it with the same idea in different vocabulary. Gartner names autonomous coding, dynamic exception handling and anomaly detection. Forrester elevated dispute resolution out of the exception queue and into a workflow differentiator in its own right, and moved supplier collaboration into the core of the product rather than leaving it at the edge.

The practical consequence for a buyer is that the honest comparison between vendors is no longer a percentage. It is the depth of the decision a system can carry on its own. Coding a line to the right ledger account is one decision. Resolving a quantity mismatch against a receipt is another. Agreeing a payment date with a supplier who is short of cash is a third, and it is the one where the phrase "finance control layer" starts to mean something.

A vendor that can clear clean invoices quickly and escalates everything else has not been graded well by either firm.

Fifteen providers were scored, and the public record names only some of them by tier

Forrester's fifteen, alphabetically: Basware, Coupa, Emburse, Esker, Ivalua, Medius, Newgen Software, Quadient, Ramp, Rossum, Serrala, Springtime Technologies, Tradeshift, xSuite and Zip.

Three Leaders have been announced publicly. Basware holds the highest score in all six of Forrester's strategy criteria, including vision, innovation and roadmap, and top marks in ten current-offering criteria including e-invoicing compliance, fraud and risk management, exception handling and process monitoring. This is the second consecutive edition in which Basware has held the top rung, which is the distinction the second-edition timing makes visible. Esker's placement rests on a customer-centric vision and on AI-mediated communication between internal teams and suppliers inside dispute workflows. Ivalua's rests on autonomous accounts payable capability and a supplier collaboration portal that carries event and status tracking with auditable workflows behind it.

Four Strong Performers have been announced: Emburse, Serrala, Tradeshift and xSuite. Serrala is included in the report for the first time. Quadient has described its placement as being among the leading providers in the evaluation without stating a tier, and this too is a first appearance.

The remaining vendors in the field have not stated a placement publicly. That is ordinary in this market and it says nothing about how they scored. A placement is a marketing asset, and companies announce the ones that help them. The full tier list has not been widely republished, so any ranking beyond the seven announcements above would be an inference rather than a fact.

Forrester graded the supplier relationship, Gartner graded the control environment

The clearest split between the two documents is in what each treats as the hard part.

Forrester's three themes for this edition are agentic AI, dispute resolution as a workflow rather than a queue, and supplier collaboration inside the core experience. Each of those points outward, at the exchange between the buyer's finance team and the supplier who sent the invoice.

Gartner's Leader bar points inward. Fraud and anomaly detection, autonomous coding, predictive cash and working capital optimization are all properties of the buyer's own control environment. They are about the record the buyer can produce and the accuracy of the forecast above it.

Read together, Forrester is grading the argument with the supplier and Gartner is grading the audit trail behind the payment. A product can be strong at one and thin at the other, and the reports will rank it differently depending on which question they asked.

That is also where the rosters diverge most sharply. Basware is the one vendor publicly on the Leader rung of both documents. Coupa is a named Gartner Leader and appears in Forrester's fifteen without an announced tier. Esker and Ivalua lead at Forrester, and whether either sits on Gartner's rung has not been stated publicly. A buyer who reads only one of the two gets half the profile and no signal that the other half exists.

What to ask before you automate an invoice desk

Ask for the exception rate and the exception mix together. A vendor that reports a high touchless rate over low-value, clean invoices and a high escalation rate over everything else has given two numbers that describe the same weakness. The exception mix is where the operating cost actually lives.

Ask which decisions the system owns in writing, and at what confidence threshold. Both firms now grade autonomy, and autonomy is a spectrum rather than a switch. The useful question is not whether the product uses AI. It is which coding, matching, dispute and scheduling decisions it will take without a human, and what it does when its confidence is low.

Ask how a dispute is resolved, not how it is detected. Forrester promoted dispute resolution to a differentiator because detection has become common and resolution has not. A queue with a status field is not the same product as a workflow that carries a supplier conversation through to a settlement.

Ask what the fraud and anomaly layer is actually reading. Gartner's Leader bar requires it, so every serious vendor will claim it. The difference is whether the model looks only at the invoice or at the invoice in the context of the supplier history, the purchase order, the receipt and the payment run.

Ask whether supplier onboarding is inside the product or bolted onto it. Forrester moved supplier collaboration into the core of the evaluation, which means the cost of a manual onboarding process now shows up in the scores rather than in the implementation budget.

Ask which of the two reports the vendor is quoting, and read that one. A vendor that cites a Leader placement without naming the analyst firm has usually picked the document where it scored better. With two 2026 reports evaluating overlapping vendor fields on the same ground, that choice is now available to almost everyone in the market.

Forrester does not endorse any vendor named here, and tier placement should not be read as a recommendation to buy.

The invoice is the last link in a longer chain. Source-to-Pay Suites covers the upstream half, from sourcing and contracting through the purchase order to the receipt, and the coding decisions an AP system makes are only as good as the purchase order data that reaches it. Both 2026 reports grade the downstream end of that chain, and neither of them grades the chain.

Supplier collaboration, which Forrester moved into the core of this evaluation, is an evaluated market of its own. Multienterprise Collaboration Networks covers the shared platforms where buyers and suppliers exchange documents, status and disputes, and several of the vendors scored in the Wave above are network operators as well as software vendors. That overlap is why the supplier-facing criteria carry weight in one report and barely appear in the other.