by Andrew Flake
Arbitration is a creature of contract. The principle is familiar enough to state in a breath: no party is bound to arbitrate a dispute it has not agreed to arbitrate. The harder questions live at the margins — and one of them is when, if ever, a party that never signed the arbitration agreement can be pulled to the arbitration table anyway. Last week, in Jackson v. Stevenson, No. S25G0922 (Ga. May 19, 2026), the Georgia Supreme Court, in a thoughtful opinion by Justice Land, clarified that question — and, along the way, signaled where it may be heading next.
A word on the background. Richard Jackson and Mark Stevenson, through various companies each of them owned, were members of a real estate development joint venture, with the Jackson entities holding seventy percent and the Stevenson entities thirty. Two operating agreements, each carrying an identical FAA arbitration clause, governed the venture. When the relationship soured, the Stevenson entities exercised a contractual “buy/sell” right to purchase the Jackson interests, and then brought arbitration alleging that the Jackson entities had stripped the venture of valuable assets ahead of closing.
The Stevenson entities then sought to join — and the arbitrator agreed to add — a Jackson-owned company called RICSHA, a non-signatory to the operating agreements, on the theory that RICSHA had colluded with the signatory respondents. After a six-day hearing, the arbitrator entered an award in favor of both parties, that netted out to roughly $3.75 million warded against the Jackson entities and RICSHA alike. The trial court confirmed and the Court of Appeals affirmed, each deferring to the arbitrator’s decision to bring RICSHA in. The Supreme Court reversed. Three points are worth drawing out.
First, the Court confirmed that equitable estoppel, as a means of binding a non-signatory, is largely a one-way street. Meaning that where a non-signatory who has been sued on a contract seeks to hold the signatory plaintiff to the arbitration clause in that same contract. The idea is that that the plaintiff should not, in fairness, be allowed to rely on the arbitration clause for one purpose, i.e., asserting its claim, and then avoid it for another, i.e., participating in arbitration.
What the Stevenson entities attempted was the mirror image: a signatory plaintiff using estoppel to force an unwilling non-signatory into arbitration. As the Court explained, that reverse application is far less established and is not supported by the same notions of fairness and equity. The reason, at bottom, is consent. Estoppel reaches a non-signatory only where the non-signatory has itself done something — relied on the contract, or sought its benefits — to make arbitration a fair consequence of its own conduct. RICSHA had done no such thing, and so there was nothing on which an estoppel could fasten.
Second, the decision is a clear reminder that the standard of review follows the answer to the “who decides” question. The United States Supreme Court taught in First Options of Chicago, Inc. v. Kaplan, 514 U.S. 938 (1995), that the deference a court owes an arbitrator’s arbitrability ruling depends on whether the parties agreed to send the arbitrability question itself to the arbitrator. That is, we have three kinds of disputes: (i) the merits; (ii) whether the parties agreed to arbitrate the merits; and (iii) who holds the primary power to decide that second question.
Where the parties clearly committed arbitrability to the arbitrator, a court reviews with considerable leeway. Where they did not, the court decides arbitrability independently and reviews de novo. And a non-signatory, almost by definition, has agreed to none of it. Because RICSHA never agreed to submit anything to the arbitrator — including whether it could be made to arbitrate at all — the courts below should have decided that question for themselves, accepting the arbitrator’s factual findings unless clearly erroneous but resolving the legal question of RICSHA’s amenability to arbitration de novo. By instead deferring to the arbitrator, as though RICSHA stood in the shoes of a party that had agreed to arbitrate, both the trial court and the Court of Appeals applied the wrong level of deference. Reviewing the matter properly, on a de novo basis, the Supreme Court found no basis to bind RICSHA — which meant the arbitrator had exceeded his powers under Section 10(a)(4) of the FAA in compelling it, and the award against RICSHA could not stand. The Supreme Court reversed as to RICSHA and vacated that portion of the arbitration award.
Third — and here is the part to watch — the Court took evident interest in the doctrine of “direct benefits” estoppel, examined it with care, and then expressly declined to decide whether it is part of Georgia law. Direct benefits estoppel asks not whether a non-signatory has been sued on a contract, but whether it has reached out and taken a benefit under the contract containing the arbitration clause. The line the cases draw is between a benefit that flows directly from the agreement itself and one that flows only indirectly, from the parties’ contractual relationship. The Seventh Circuit’s decision in Everett v. Paul Davis Restoration, Inc., 771 F.3d 380 (7th Cir. 2014), which the Court discussed, illustrates the distinction. There, a franchisee’s wife — a non-signatory — took over a ninety-five percent ownership interest and continued operating the franchise under a new name to evade a post-termination non-compete. She was compelled to arbitrate because the benefit she seized was the very thing the franchise agreement conferred: the right to own and operate the franchise.
Picture the arbitration agreement as a restaurant’s menu. Two diners make the reservation and place their orders; an unexpected friend joins them at the table. If the friend orders a meal of his own, the benefit he receives comes straight from the menu — and it makes little difference that his companions pick up the check, that he never signed it, or that he had nothing to do with booking the table. In ordering off the menu, he receives a direct benefit. Contrast the friend who simply sits, enjoys the company of the other two, and perhaps gleans some useful business conversation along the way. He has benefited from the evening, certainly — but indirectly. He never ordered. The Stevenson entities’ theory put RICSHA in the second chair, not the first: their complaint was that RICSHA interfered with the operating agreements to advantage itself, not that RICSHA invoked those agreements to claim a benefit under them. Interference of that kind may expose RICSHA to liability, but it is not the sort of direct benefit that — even under a doctrine Georgia has not yet adopted — would support compelling arbitration. The Court was careful to say only that, even assuming direct benefits estoppel were good law here, it would not reach RICSHA on these facts.
Still, that attention with which the Court delineated direct benefits estoppel from the ordinary equitable-estoppel scenario reads as more than addressing an argument raised by Stevenson. It tells us that on the right set of facts — a genuine “ordering off the menu” case — the Court may take the doctrine up squarely and decide whether it is part of Georgia’s law of arbitration. –ABF
[The opinion is Jackson v. Stevenson, No. S25G0922 (Ga. May 19, 2026). The cases discussed are Everett v. Paul Davis Restoration, Inc., 771 F.3d 380 (7th Cir. 2014), and First Options of Chicago, Inc. v. Kaplan, 514 U.S. 938 (1995).]