Angi's Long Pivot: From Quarterly Revenue to an AI-First Bet — With Receipts This Time

Angi Inc., the home-services marketplace now worth just ~$205 million after shedding ~97% of its value since 2013, is living a very particular kind of turnaround: one where the CEO openly admits the first-year strategy was built on bad data, guidance stays withdrawn, and the entire bull case rests on executing an AI agent roadmap that is — for the first time — starting to show small, concrete wins.

A quarter that frames the whole drawdown

Q2 revenue fell 11% year-over-year, and the CFO was blunt about the three contributors: the deliberate wind-down of network channels (down 34%), a step back from lower-quality paid marketing, and what Julie Hoarau described as a material shift in homeowner mix beginning mid-March. “Our revenue in Q2 was materially impacted by a shift in mix and traffic starting about 10 days into March as oil and gas prices rose sharply following global events.” — Julie Hoarau, Chief Financial Officer · 2026-08-05 Demand moved away from large-ticket roofing and HVAC work — where Angi has the most pro capacity — toward smaller jobs with less supply, an echo of the Middle East impacted theme that has rippled through this entire earnings season across airlines, insurers and energy names.

The more defining line item, though, is the one-time one. The sustained collapse in the company's market capitalization forced the required interim goodwill test, and Angi recorded a $235 million non-cash charge against its U.S. reporting unit. Operationally the charge changes nothing, but it crystallizes just how far the equity has fallen: operating income swung to −$234 million, and the shares themselves are down 33.7% over just the last 90 days, extending a drawdown that now measures −98% from the 2013 peak. Angi is a profitable-ish business wearing a distressed-balance-sheet haircut.

The May promise, with receipts

Three months ago Jeff Kip pulled guidance, framed the pivot to an "agentic" AI-native platform, and quietly floated an internal floor. The skeptical reading of that call was that it was an excuse for continued revenue decline. This quarter's answer is: maybe, but execution is now visible. The company reiterated it is on track for roughly $50 million of EBITDA minus CapEx, and claims July EBITDA margin already sits several hundred basis points above Q2. Contrast the current posture with the earlier hedge: “We'd be happy with the cash flow cushion and, give or take, the range of $50 million a year. That's adjusted EBITDA minus CapEx.” — Jeffrey Kip, Chief Executive Officer · 2026-05-06 Now the emphasis has subtly shifted toward the growth side of the ledger: “If we execute our strategy, we should be growing again sometime in 2027.” — Jeff Kip, Chief Executive Officer · 2026-08-05

What makes the new claims credible is that they come with concrete product anchors rather than aspiration. Management reports the homeowner experience agent now touches roughly half of homeowner traffic and converts at three times the rate of untouching traffic; the first Pro-side agent, the AI Front Desk, is live with dozens of pros and "dozens of appointments booked" at a baseline comparable to human call centers; and ChatGPT — the new channel that did not exist as a line item a year ago — is already spending profitably at a ~$3 million revenue run rate. The trajectory of that channel is told against the company's own history: “a year and a half ago, our Meta business was half that, and we're now approaching a profitable $100 million revenue run rate on Meta.” — Jeff Kip, Chief Executive Officer · 2026-08-05 This is the Angi Pro / Chief Revenue Officer thesis beginning to form.

Why couldn't we do this before? I'll take the blame. My first year in the job, we were looking at the wrong data with the wrong team, which meant the wrong execution.

That candor — an unusual admission for a CEO — is the emotional core of the story. But it also frames the risk: the strategy was only really re-derived “a little over a year ago," and the company is still burning through the consequences of prior missteps. Win rate — the metric Kip points to as the single most instructive signal — is up roughly 20% year-over-year, with pros now winning about one in six leads versus one in nine two summers ago, and the large-Pro segment is growing north of 20% even with less than a third of the targeted sales headcount in place. Churn reduction is the promised flywheel: a 25% improvement would add ~10 points to annual growth, a claim consistent with the internal "double win rate, halve churn" heuristic.

What has to break right

The balance-sheet discipline remains the quiet counterweight. Effective net cash sits at −$210 million, and the CFO named refinancing the 2028-due bonds as the top capital priority over the coming year — ahead of any new AI spend. CapEx rose 22% year-over-year to $15 million, a reminder that the re-platforming is real investment, not just rhetoric. Revenue is still guided down sequentially-modestly-improving rather than growing, and the company left the door open to another goodwill write-down when the annual test lands in October. The SEO tailwind is presumed dead: unbranded organic is down to roughly 5% of service-request volume and is not assumed to recover. Meanwhile the whole industry — from personal agent builders to every adjacent marketplace — is racing toward the same LLM layer Angi is betting on, so the competitive clock is ticking even as the company's own clock for returning to growth keeps sliding toward 2027.

In short, this is no longer a story about a broken lead-gen business. It is a story about a micro-cap using a $50 million cash floor to fund a genuinely ambitious AI-native rebuild, while an accounting charge and a net-debt balance sheet scream that the market has priced in very little patience. The 2027 growth promise is now backed by a live agent, a profitable new channel, and a candid CEO who finally admits what went wrong — which is precisely why the next two quarters of pro-capacity and win-rate data will matter far more than the revenue print.