ZOO Digital: Losing Half Its Biggest Customer and Making More Money Anyway
The trade nobody at ZOO wanted to make — and did
ZOO Digital's interim FY27 statement contains a number that should be terrifying and instead argues the bull case. Its largest customer went from 55% of revenue to 26%, a roughly $6m absolute hit — around a 50% cut from that one account. Every other customer combined grew 72%.
“our largest customer, which in H1 FY '27 was 55% of the business, so a significant part of the business reduced to 26%. There's 2 reasons behind that. One was all the other customers grew at 72%.” — Robert Pursell, Group CFO · 2026-11-10The arithmetic: revenue up 5% year-on-year and 18% on the second half of last year, with group gross margin of 43% versus 45%. But the mix inside that flattish gross profit is the whole story. Media services, the technical formatting work that made up the lost volume, fell $4m in revenue and $4m in gross profit — deliberately, because that team is being held for the return of the account's output. Meanwhile the localization side lifted gross margin from 27% to 40%. Reported EBITDA went from $2.0m to $3.5m (the CFO stumbled over the digit mid-sentence), cash EBITDA from $0.6m to $1.9m, and the bottom line swung from a $1.3m loss to just under $0.8m of profit.
That is a rare shape: a company that lost half of its biggest account and still improved profitability, because administrative costs came down and because what remained was higher-value work. Capitalised development rose around 30%; gross cash of $3.4m; payables days deliberately unwound from the stretched levels of 18 months ago.
What ZOO stopped saying — and what it quietly contradicted
The most revealing material is in the prior call, from late July. Asked whether the largest customer would fall in absolute terms or just as a percentage, the CFO was explicit:
I would say it would be mainly from the increase in other customers... the relationship is still very strong, as strong as it's ever been with them.
Three and a half months later the answer is the opposite: it fell in absolute terms, by $6m. The relationship is still described as intact and the cause attributed to that customer's own original-content retrenchment — but the guidance was wrong in direction, not merely in degree. That matters for how much weight to put on the forward-looking reassurance that the account "could recover."
Second, the dubbing tell. In July, management offered dubbing order volume as the concrete leading indicator that original content was coming back:
“in our order book at the moment, we're seeing much more content that requires dubbing services. And therefore, that for us is a clear indicator that there is a return of original content.” — Stuart Green, CEO · 2026-07-30In November, dubbing is still a promise rather than a print: growth is expected to show "a little bit in H2 and beyond," with the largest customer's dubbing-led recovery deferred. The original content keyword that sat high in the prior quarter's language has dropped out of the top of this call's vocabulary entirely. When a company's own stated indicator stops being cited, that is the signal.
Third, the language of speed normalised. Last quarter Fast Track was the top-ranked theme with the loudest emphasis; this quarter the phrase persists but as one service name among many, replaced by accelerated delivery and the more interesting admission that speed is now the default rather than a premium product. The substance grew — 38% of first-half revenue came from projects won on turnaround times the industry's standard two-to-six-week cadence can't match — even as the rhetoric cooled. That divergence is usually a maturity marker, not a warning.
Speed as the moat, AI as the alibi
The strategic core of the call is an argument that the market has finally moved toward ZOO rather than ZOO having to move toward the market. Subtitling that takes peers two weeks can be turned in hours; dubbing that takes four to six weeks can be turned in twenty-four hours, on a follow-the-sun operating model plus fifteen years of proprietary workflow automation. Historically that capability was invisible because buyers would not source fast work from a single vendor when everyone else quoted six weeks.
The defence against generative AI is the most intellectually interesting part, and it is a genuinely new framing for this company: adaptive quality. Management's claim is that machine translation wins on transactional, factual, structured language — manuals, websites, documents — and loses on emotionally rich, culturally specific, subtext-heavy screen dialogue where the output is judged by audience response, not linguistic accuracy.
“AI actually performs best when the language is dealing with is transactional, it's highly factual, it's highly structured... Entertainment language, which is exactly what we deal with, it's the spoken word, is often the opposite of those things.” — Stuart Green, CEO and Co-Founder · 2026-11-10They also concede the other side: for cheaply acquired creator content where viewership is limited, customers will accept lower quality and AI takes a bigger role. That is an honest segmentation rather than a denial, and it lands against a market tape that has been ruthless on anything AI can plausibly disintermediate. Investors can accept a services business whose moat is quality-at-speed; they are far less forgiving of one whose moat is a claim that AI can't do it.
The visibility discount is the whole valuation
Here is the tension that keeps the shares where they are. The company is growing, profitable and cash-generative, and yet the analyst consensus $43.5m full-year revenue implies roughly $19.9m in the second half — below the first half. Management did not dispute it, and explained why in terms that are both credible and unhelpful:
“Q3 is trading well, and it's trading in line with H1, but we just don't have that visibility in terms of -- and it's the nature of the business for Q4.” — Robert Pursell, Group CFO · 2026-11-10The cause is the strategy itself. Big accelerated delivery projects worth $1m–$2m each land with one and a half months' notice, which is fantastic for gross margin and terrible for forecasting. The offsetting datapoint is capacity: the CFO put sustainable turnover at $50m–$55m on the current cost base, roughly a quarter above consensus, with "plenty of leverage" before any new hires. If that is real, the incremental drop-through on the next $10m of revenue is very high.
Against that, the shareholder questions are the same ones as three months ago, only worse: shares down about 80% (the CEO previously called the valuation "terrible" at roughly 0.3x sales), no dividend, and a policy that "remains set" on reinvestment. There is no buyback, no dividend signal, and no guidance upgrade — just a cautious analyst number. The addressable market story, tied to streaming services broadening beyond film and episodic into sport, news and live formats, is a genuine addressable market expansion thesis, but the market has heard it before and wants proof in H2 revenue, not in commentary.
The verdict
What genuinely changed this half: the customer concentration risk collapsed, gross margin inflected, and the cash-conversion profile turned. What did not change: the second-half visibility problem, the absence of any capital return, and the largest customer's spending. ZOO's own investor asked whether the Paramount Plus-style consolidation of Paramount Skydance and Warner Bros helps; management's answer — both are already customers, merging them should mean more share of a bigger wallet — is plausible and unprovable for several quarters.
This is a company that has been right about its technology for years and wrong about its timing for three. The first half is the first hard evidence that the timing has caught up. But with a ~$20m market capitalisation, an 80% drawdown, and guidance that deliberately under-promises a second half it cannot see, the report earns the right to be re-examined next time rather than re-rated today.