On 17 August 2026, The a2 Milk Company released its FY26 annual results. Its annual report and FY27 outlook record China-label infant milk formula revenue down 14.0% to NZ$544.3 million. The decline was concentrated in the second half: revenue rose 6.5% in 1H26 and fell 33.0% in 2H26 after product shortfalls reduced availability at distributors and retailers.
The company says the contributing supply factors have been resolved and product availability has significantly improved. That changes the supply state, but it does not yet establish recovered demand. FY27 guidance assumes China-label sales will recover gradually, total infant milk formula sales will remain broadly similar to FY26, group revenue will grow at a mid-single-digit rate and EBITDA margin will be approximately 15%. Group FP&A therefore needs a staged recovery model rather than one replenishment assumption carried through revenue and margin.
What changed and what it means
Treating improved availability as recovered demand could front-load revenue, overstate favourable mix and distort the approximately 15% EBITDA-margin bridge.
- Decision affected
- Set the FY27 China-label base case and sensitivities only after separating availability, distributor stock, customer reacquisition, unit sell-through, label mix, price, freight, production and marketing assumptions.
- Evidence in brief
- The final FY26 report records China-label IMF revenue down 14.0% to NZ$544.3 million, says the contributing supply factors are resolved and availability has significantly improved, and guides to gradual China-label recovery in FY27.
- What remains unresolved
- Distributor inventory by label, past-user reacquisition, unit sell-through, price and mix attribution, and a quantified cost-to-margin bridge are not disclosed.
- Next verification
- Reconcile the 19 November recovery update to China-label offtake, reacquisition, channel mix, marketing and cost assumptions before changing the base forecast.
Key takeaways
- a2 Milk’s final FY26 result shows China-label revenue down 14.0% to NZ$544.3 million, with the decline concentrated in 2H26.
- Resolved supply factors and improved availability do not prove that distributors have normalised stock, former users have returned or unit offtake has recovered.
- China-label and English-label infant formula need separate forecast lines because English-label growth does not establish China-label recovery.
- The approximately 15% FY27 EBITDA-margin guide sits between reported and underlying FY26 margins, so FP&A still needs a driver-level bridge.
What a2 Milk’s FY27 outlook actually commits to
The FY27 outlook is a management forecast, not an observed recovery. It calls for first-half revenue broadly in line with 1H26, an EBITDA margin materially lower than 1H26, and group revenue and EBITDA materially weighted to 2H27.
| Disclosure | Current status | FP&A treatment |
|---|---|---|
| China-label availability | Contributing factors resolved; availability significantly improved | Update supply and service assumptions, not the demand baseline by itself |
| China-label sales | Expected to recover gradually through FY27 | Phase recovery by evidence milestone rather than back-solving a smooth curve |
| Total infant milk formula | Expected to be broadly similar to FY26 | Reconcile China-label recovery and English-label momentum without netting the two too early |
| Group revenue | Mid-single-digit growth; 1H27 broadly in line with 1H26 | Keep the base case materially weighted to the second half |
| EBITDA margin | Approximately 15%; 1H27 materially below 1H26 | Bridge mix, production, freight, input costs and marketing before accepting the target |
Availability is only the first China-label recovery gate
The prior state matters. In its 13 April trading and supply-chain update, a2 Milk described temporary China-label shortfalls caused by strong preceding demand, freight constraints, a Synlait production backlog, longer product-release testing and additional customs requirements. It expected those factors to materially affect fourth-quarter availability.
The 7 July preliminary update then said product flows had materially improved and stock levels were returning to target. It also confirmed that a large proportion of existing users had switched to alternatives and that the company was trying to encourage past users to return while recruiting new users.
| Recovery stage | Evidence available now | Model treatment |
|---|---|---|
| Product availability | Company-stated improvement | Base the supply line on service levels and confirmed production flows |
| Distributor and retailer stock | Stock returning to target was stated in July; quantities by label and channel are not disclosed | Use actual channel inventory where available; otherwise retain a watchlist assumption |
| Past-user reacquisition | Recovery activity is underway; outcomes are not disclosed | Keep return rates in sensitivity cases until cohort data supports the base case |
| New-user recruitment | Management action, not an observed unit result | Separate acquisition volume and marketing cost from returning-user volume |
| Units and offtake | No FY27 result has been reported | Raise the base case only after sell-through or depletion confirms recovery |
The same forecast control applies to the July 2026 U.S. container-import signal: more product moving through a supply chain can change receipt timing and inventory exposure without proving end-customer demand. For a2 Milk, the model should distinguish factory output, distributor receipts, shelf availability and consumer offtake.
Keep China-label and English-label IMF on separate forecast lines
English-label sales increased 23.2% in FY26 while China-label sales declined 14.0%. The China and Other Asia segment still grew 11.2%, led by English-label infant formula and other nutritionals. It does not establish that China-label users have returned.
The products also had different disruption profiles. The final report says the availability impact on English-label a2 Platinum was limited and was concentrated mainly in a2 Genesis, while China-label shortfalls caused a large proportion of users to switch brands. Some users moved to a2 English-label products. A group-level infant-formula bridge could therefore hide substitution inside the portfolio.
FP&A should retain separate lines for label, channel and customer cohort. A returning China-label user, a former user who remains with English-label product and a newly recruited user create different unit, price, promotion and marketing assumptions.
Build the EBITDA bridge without calling 15% a clean recovery
Reported FY26 EBITDA margin was 14.4%. Excluding a2 Pōkeno operating losses and transformation costs, underlying EBITDA margin was 15.6%. The approximately 15% FY27 guide sits between those two measures. It is higher than the reported FY26 result but lower than the underlying comparison, so the word “recovery” needs a stated basis.
FY26 gross margin fell to 47.7%, down 3.4 percentage points. The company attributed the decline to a2 Pōkeno under-utilisation, a lower share of China-label sales, one-off supply-chain costs and higher milk and ingredient prices. It also invested NZ$325 million in marketing, focused on China growth and new-user recruitment. FY27 adds higher first-half marketing activity and planned Pōkeno improvement, but the company does not quantify each contribution to the approximately 15% EBITDA margin.
On Holding’s channel and margin bridge shows the same modelling constraint in another consumer-products forecast: a gross-margin improvement cannot pass automatically to EBITDA when channel costs and growth investment move separately. For a2 Milk, FP&A should assign owners to label mix, product margin, production utilisation, freight, ingredient cost, marketing and operating expense, then reconcile the residual rather than treating it as an unexplained recovery benefit.
Put each recovery assumption in the base case, sensitivity or watchlist
The base case should contain only the stages supported by current operating evidence. Improved availability can enter the supply plan, subject to service-level and production checks. Revenue recovery should remain phased and conditional because the company itself expects a gradual path and first-half revenue broadly in line with the prior year.
- Base case: confirmed production flows, verified distributor receipts, current pricing, committed marketing spend and the disclosed second-half weighting.
- Upside sensitivity: faster past-user return, higher new-user conversion, better China-label mix and earlier Pōkeno profitability.
- Downside sensitivity: slower reacquisition, heavier promotion, weaker offtake, higher input or freight costs and delayed production benefits.
- Watchlist: channel inventory by label, user cohorts, unit sell-through, price and promotion, customer acquisition cost and margin contribution by recovery stage.
This treatment prevents the same favourable assumption from appearing twice. A higher China-label mix can improve gross margin, but it should not also be added as a separate EBITDA benefit if the gross-margin line already contains it. Marketing that drives reacquisition belongs in the same scenario as the units it is expected to generate.
What FP&A should verify at the 19 November update
a2 Milk says it will update investors on the infant-formula recovery plan at its annual meeting on 19 November 2026. The next forecast review should ask for evidence that closes each stage of the bridge:
- China-label service levels and in-market availability by major channel.
- Distributor and retailer stock against target, including ageing and weeks of cover.
- Past-user return rates and new-user recruitment by cohort.
- Unit offtake, price, promotion and label or channel mix against the FY27 base case.
- Freight, production, ingredient and marketing costs tied to the recovery plan.
- A reconciled gross-margin and EBITDA-margin bridge for 1H27 and the second-half weighting.
Availability is a necessary first condition for recovery, not the final proof. The FY27 model becomes more reliable when every step from product flow to EBITDA has its own evidence date, owner and activation trigger.
