Summary

  • Zscaler reported that no single customer accounted for at least 10% of revenue in fiscal Q3 2026, while channel partners produced US$718.853 million, or 85%, of the quarter’s US$850.475 million total.
  • Anonymous channel partner A represented 12% of net accounts receivable at 30 April 2026. That share sits on a US$730.461 million portfolio, but the filing does not disclose the partner’s exact balance, aging, identity or payment status.
  • Positive operating cash flow and free cash flow argue against treating the disclosure as evidence of distress. The useful question is whether end-customer demand, invoice ownership, partner remittance and final cash are monitored as separate states.

A customer count is not a collection map

Concentration analysis often starts and stops with revenue. If no customer contributes 10%, the business appears to have no single buyer capable of moving the income statement alone. That is useful. It is not a description of the route by which invoices become cash.

One intermediary can aggregate many independent buyers. A global systems integrator may resell subscriptions to unrelated enterprises. A telecommunications provider may bill customers in several markets. The underlying demand can be dispersed while the vendor’s unconditional receivable sits with the intermediary.

This is the distinction inside Zscaler’s filing. “Customer” is the unit in the revenue concentration statement. “Channel partner A” is the counterparty in the receivables table. The two units can overlap without answering the same question. The first measures who generated recognised revenue. The second shows where a portion of the unpaid invoice balance sat on one date.

The difference is not a loophole. It is how indirect enterprise software distribution works. It becomes a risk only when analysts carry reassurance from one denominator into another without preserving the commercial chain.

Eighty-five cents of each revenue dollar travelled through partners

Zscaler recognised US$850.475 million of fiscal-Q3 revenue. Channel partners supplied US$718.853 million, presented as 85% of the total; direct customers supplied US$131.622 million, or 15%. For the first nine months of the year, the channel share was 84%.

The annual record is similar. Fiscal-2025 revenue through channel partners was US$2.360224 billion, or 88% of the company total. This is not an incidental route wrapped around a direct-sales engine. The channel is the commercial operating system for most revenue.

Breadth exists inside that system. No single customer accounted for 10% or more of revenue in the periods presented. The top five channel partners and their affiliates together contributed 30% of nine-month revenue, up from 27% a year earlier. That aggregate does not reveal the five individual shares, and the filing does not say that the receivables partner is one of them.

The 30% figure also cannot be subtracted from 84% to manufacture a “long tail” of identical partners. Affiliates, direct relationships, geography, contract term and service role can differ. It says only that the largest five were meaningful collectively while each customer remained below the separate 10% line.

What the 12% actually measures

At 30 April, Zscaler carried US$730.461 million of net accounts receivable. Channel partner A represented 12% of that balance. The same current filing presents that partner at 12% against the 31 July 2025 balance.

Applying 12% to the April total produces US$87.655 million. That calculation is useful only as scale. The reported share is a whole percentage, so the exact partner balance is not available. Under an ordinary rounding interval, a displayed 12% could correspond to roughly US$84.0 million to US$91.3 million. The filing supplies neither endpoint.

More importantly, an account receivable is not a bad debt. Zscaler records the asset when its right to consideration becomes unconditional. The annual filing says invoice terms are typically 30 days but may extend to 90 days for some channel partners. A balance at quarter-end may be young, scheduled, disputed, late, partly collected after the reporting date or a mixture. The public table does not separate those states.

The 12% also is not a share of quarterly revenue. It is not 12% of cash, bookings, ARR, RPO or deferred revenue. It cannot reveal how many end customers stand behind the partner, whether they have already paid the intermediary, or whether the partner remits only after collection. Those are contractual facts that the sources do not publish.

Three dates show a clock, not a credit verdict

Zscaler’s net receivables were US$992.181 million at 31 July, US$551.820 million at 31 January and US$730.461 million at 30 April. The sequence resembles a year-end billing and collection cycle: a large July balance fell sharply by January, then rebuilt in the third quarter.

That is a description of balances, not proof of cause. Revenue timing, contract starts, invoice schedules, acquisitions, currencies, collection days and partner mix can all move the line. The company itself attributes the cumulative cash benefit from lower receivables primarily to billing and collection timing.

The threshold table adds another wrinkle. The January filing placed its Channel partner A line below 10% at January and at 12% at July. The April filing shows 12% at April and July. Because the counterparty is anonymous and percentages below 10 are suppressed, readers cannot reconstruct an exact continuous series. The observation that the disclosed line moved across the threshold is valid; a precise partner balance path is not.

This matters because a stable displayed 12% at two endpoints need not mean a stable dollar exposure. Twelve per cent of the July portfolio is about US$119.1 million by simple multiplication, while 12% of April is about US$87.7 million. Both are rounded illustrations. Concentration can remain visually unchanged while the gross balance moves.

Cash generation is the necessary counterweight

Zscaler did not report a collection crisis. Fiscal-Q3 operating cash flow was US$198.016 million and free cash flow was US$135.954 million. For the nine months, operating cash flow reached US$850.369 million.

Receivables contributed US$280.042 million of cumulative operating cash inflow over those nine months, primarily because of billing and collection timing. Through January, the cumulative contribution had been US$461.531 million. Subtracting the two comparable cash-flow lines implies that the third quarter alone used roughly US$181.489 million as receivables rebuilt. That is a whole-portfolio movement, not a partner-A movement.

The quarter therefore contains both a positive cash result and a receivables build. There is no contradiction. Subscription businesses can collect heavily after fiscal year-end and bill again during the next quarter. The analytical mistake would be to call every build deterioration or every decline superior collection without an opening-to-closing invoice ledger.

Strong cash is evidence against an immediate bearish claim. It is not proof that every channel node is interchangeable. A cash total can absorb one counterparty’s slower remittance while another pays faster. Portfolio health and node substitutability need separate proof.

ARR and RPO sit on different clocks

Zscaler reported ARR of US$3.525 billion and remaining performance obligations of US$6.4593 billion. Both demonstrate contractual scale, but neither completes the collection map.

ARR is a non-GAAP run-rate measure. Zscaler assumes contracts expiring in the next 12 months renew on existing terms, apart from specified Red Canary contracts. RPO is transaction price allocated to contracted performance still to be delivered; 46% was expected within 12 months and 91% within three years.

An ARR dollar can exist before an invoice. An RPO dollar can be recognised long after a partner first records an order. Deferred revenue reflects advance consideration. A receivable begins when the right to consideration is unconditional. Cash arrives when settlement completes. Subtracting any pair of these quantities would not produce a funding gap or a partner exposure.

The market needs the joins, not a blended number: end customer, sales partner, contract, invoice, payment term, collection state, remittance state, recognised revenue and cash receipt.

The bridge can protect commercial confidentiality

Zscaler need not name a reseller or expose customer invoices. A useful disclosure could group direct and channel receivables into aging bands, show the share covered by 30-, 60- and 90-day terms, and reconcile opening receivables to billings, collections, write-offs, acquisitions, currency and closing balance.

Partner concentration could remain anonymous. Bands for the largest node, next four and remainder would make it possible to compare revenue share, receivable share and overdue share without revealing identity. A separate state could show whether end-customer payment extinguishes the partner’s obligation, is merely evidence for remittance, or is irrelevant under the contract.

This would also improve operational decisions. Treasury could distinguish a predictable fiscal billing build from a partner-specific delay. Sales could see whether growth is deepening dependence on a few settlement rails. Security operations could identify where a commercial disruption might complicate renewals or support even if the service itself remains available.

The judgment

Zscaler’s disclosures support two statements at once. Its recognised revenue was not dependent on one 10% customer, and one anonymous channel partner held 12% of net receivables at the April snapshot. The first is evidence of demand breadth. The second is evidence that unpaid consideration can be more concentrated than revenue.

Neither statement proves trouble. The company generated cash, the receivables portfolio was below its July level and the partner’s exact balance and aging are unknown. Naming a risk does not mean declaring a failure.

The disciplined conclusion is that customer diversity and collection diversity are separate properties. Zscaler already publishes the endpoints. The missing evidence is the settlement bridge between them.

Sources