Open in interactive viewer → charts, metric popovers & call review

i3 Verticals Learned to Sell AI. It Still Hasn't Learned to Grow.

A GovTech micro-cap admits FY26 missed, guides FY27 to ~4% growth, and quietly swaps its M&A story for a share-count story — while the AI pitch finally gets concrete, down to the cost of a hamburger.
IIIV · Earnings Call · 2026-11-18

There is a version of this call where i3 Verticals is an AI turnaround. There is another where it is a 1.9x-levered, $500 million market-cap company buying back a quarter of its own float because nothing else is compounding. Both are on the tape, and management presented them in that order.

A Confession, Then a Pivot

Greg Daily did not bury the lede. “As we close fiscal '26, I want to start by acknowledging that our results for the year did not meet our expectations. We entered the year expecting stronger growth, and we are disappointed we did not deliver on those expectations.” — Gregory Daily, Chairman and CEO · 2026-11-18 The quarter itself was fine at the profit line and thin at the top: revenue up 1% to $55.4 million, annual recurring revenue up 7% to $177 million, SaaS up 35%, adjusted EBITDA up 5.5% to $15.2 million with margin up over 100 basis points to 27.4%.

The number that matters is buried in the middle. “As of the fourth quarter of fiscal 2026, net dollar retention was 100% — the percentage reflects the loss of a large utilities customer, an outsourced management company, which was acquired and then consolidated its operations off of our CIS.” — Geoffrey Smith, Chief Financial Officer · 2026-11-18 That is a deceleration from the 104% they cited a year ago, and it is the third consecutive quarter this wound has been open. A software company at 100% net retention is not growing through its base; it is growing only by adding logos, which is the most expensive kind of growth there is.

AI Moves From Headwind to Hero

Here is the genuine change on this call. Nine months ago, i3's president Rick Stanford was explaining why AI would arrive slowly in government software — agencies needing frameworks, laws, security protocols, inter-jurisdictional mess. “It's going to be a good bit of time away from this concept of proliferation of AI within GovTech being a real working asset.” — Rick Stanford, President · 2026-02-06 That was the bear case, delivered by management.

On Tuesday, Rick spent five minutes and a hamburger on the bull case. The education team rewrote its K-12 inventory product end-to-end with AI help and shipped it this quarter, adding Menu intelligence — a system that takes point-of-sale meal counts, recipe records and vendor pricing and tells a food service director which compliant menu items cost least and what to substitute when the price of ground beef jumps from $0.80 to $1.20 mid-year. He also flagged an AI document processing platform now in production that extracts, classifies and redacts document compliance paperwork — deeds, tax forms — with land records deployments planned across several states, and new applications under evaluation in court records and insurance verification.

Most products on the market today are good at recording what happened and documenting compliance. The gap is between reporting and deciding. ... This menu intelligence enhancement will allow us to potentially further monetize our products within the education market.

Rick Stanford, President · 2026-11-18

Note the shape of the pitch versus the market's. The global keyword tape this quarter belongs to agents and compute — Personal AI agents, Grok bots, and, from the reporters on the same calendar, Palo Alto's open-weight models and Nvidia's “strong demand for compute.” i3's AI vocabulary is menus, deeds and public administration workflows. That is not a wave it is riding; it is a moat it is claiming. “Customers don't ask for AI by itself. Instead, they want it embedded within our software they already use every day.” — Rick Stanford, President · 2026-11-18 Fair. Also unmonetized in the guide.

What Stopped Being Said

The instructive part of this transcript is the silence. For four quarters, the load-bearing promise was West Virginia — a statewide courts deal that was supposed to be the recurring-revenue ramp. It now gets one clause: “The growth ramp on that revenue stream is really more of a late '27 into 2028 item.” — Geoffrey Smith, Chief Financial Officer · 2026-11-18 Same for the Resolve product, the citation-payment engine that was “the right product at the right time” in May: two straight “tough growth quarters,” with recovery rates tied to whether people have spare cash to pay old traffic tickets.

And the M&A drumbeat has stopped. In November 2025 Greg promised “I think you'll see some activity sooner than later... I think there'll be a couple of meaningful ones we get done in '26.” — Rick Stanford, Unknown · 2025-11-18 Twelve months later: “I don't feel like we're out there pressing to buy stuff like we used to... I see probably stock buyback being a little bit more of a priority than M&A.” — Gregory Daily, Chairman and CEO · 2026-11-18 The pipeline is “mature” and has been for thirteen years, by his own telling. What is new is the admission that the price isn't there.

Meanwhile the guide itself softens the growth language. FY27 revenue of $220–235 million is roughly 4% at the midpoint — mid-single digits, with high single digits demoted to “over time.” License revenue is expected down and front-loaded into Q1 and Q2; nonrecurring is flat. The one thing they did fix: the elevated commercial-card interchange that cost them over $1 million a quarter is resolved, though the payment revenues take rate is still below pre-2025 levels, and the attach rate opportunity across the Customer Information System base remains mostly theoretical.

The Arithmetic That Is Working

So the stock is left with margin and a shrinking denominator. FY27 EBITDA guidance of $61.25–68.75 million on ~$227.5 million of revenue is a ~28.6% margin — well above the 50-to-100 basis point annual expansion they have historically guided, and even above the “100 to 200 basis point” range Geoff Smith sketched in August. He was explicit that this is achievable: “Corporate expenses as a percentage of revenue fell to 7.1% for Q4 2026 from 7.7%.” — Geoffrey Smith, Chief Financial Officer · 2026-11-18

The share count is the real earnings engine. Repurchases totaled $48 million in the latest quarter alone, after $51 million the quarter before. That is how a company guiding to ~4% revenue growth and $1.23–1.41 of pro forma EPS gets there. The leverage math boxes it in sensibly: debt of $118.5 million, total leverage at 1.92x, a freshly refinanced $350 million facility with a 4x leverage covenant and a spread 50 basis points tighter, against a self-imposed 3x ceiling.

What the market is paying for that is genuinely cheap — and getting cheaper. Price-to-revenue sits at 2.6x, down 30% year over year. Price-to-free-cash-flow is 17.9x, down from 27.7x. The cash is real: Free cash flow less stock-based comp reached $8 million, and effective net cash is negative $109 million.

The tape already knows half of this story. The stock fell 34% over ten weeks this summer, then rallied 45% over six weeks into the print — a net nothing over 90 days, still 42% below its February 2020 peak of $36.27. Investors bid the margin-and-buyback story before they heard it. What they did not get on Tuesday was a reason to raise the growth ceiling: Total revenue remains far below its $94 million 2019 peak after years of portfolio pruning, and the remaining business grew 1%.

That is the tension in one line. i3 has finally built a credible, specific, domain expertise-flavored AI story — and simultaneously conceded that the growth it promised for FY26 and pushed to FY27 is now a 2028 conversation. A cheap stock buying back its own equity is a perfectly rational response to that. It is not, however, a growth story, and the call did not pretend otherwise. It just put the hamburger in front of you first.