Two weeks before a quarterly earnings call, the mood in a review room changes. Spreadsheets get printed. Someone brings the prior year's comparatives. And somewhere in that short window before the call, an income statement lands on the table for a newly consolidated digital subsidiary, and the first line anyone reads out loud is the operating loss.
I have watched that number stop a conversation cold. Heavy upfront marketing spend, booked in full in the period it was incurred, sitting against a fraction of the revenue it will eventually generate. On paper, it looks like a bleed. In the room, it sounds like a mistake.
Table of Contents
- Why EBITDA Lied About the Digital Subsidiary
- Reading the Revenue That Hasn't Landed Yet
- Marketing Spend as a Capital Decision
- Churn Compounds in Reverse
- What Engineering Costs Reveal in Cash Flow
- Forty-Five Minutes in the Investment Committee
Why EBITDA Lied About the Digital Subsidiary
The default instinct is to reach for the multiples that already work. If a group has spent years valuing distribution assets, manufacturing lines and securities operations on EBITDA, the same arithmetic gets pointed at the technology entity almost automatically.
That is exactly what happened during one of those pre-earnings preparation windows. Legacy EBITDA multiples were applied to a subscription-driven digital subsidiary, the resulting valuation came back looking dire, and the framework was rejected outright before it reached the committee. The subsidiary was not underperforming. The measuring instrument was calibrated for a different kind of business.
Here is the structural problem. A manufacturing business spends capital once, depreciates it over years, and earns against it. A software business spends its capital on customers, expenses that spend immediately, and earns against it for years afterward with almost no incremental cost. Both are investments. Only one of them gets the courtesy of a depreciation schedule.
Indonesia's listed technology cluster makes this concrete. When a group holds securities operations alongside digital distribution and connectivity subsidiaries, the consolidated income statement blends two economic species. PT M Cash Integrasi Tbk (MCAS) as a parent, with subsidiaries such as PT NFC Indonesia Tbk (NFCX), PT Distribusi Voucher Nusantara Tbk (DIVA) and PT Surya Teknologi Perkasa (STP) operating across very different revenue mechanics, cannot be read through a single lens without producing mispriced assets.
Shareholders lose money on both ends of that error. They sell growth assets that look expensive on trailing earnings, and they hold declining ones that look cheap on the same measure.
Reading the Revenue That Hasn't Landed Yet
Annual Recurring Revenue is the closest thing a subscription subsidiary has to a north star. It answers a question the income statement cannot: what is the contracted, repeatable revenue base as of today, independent of when accounting rules permit it to be recognised?
The mechanics turn on deferred revenue. A customer signs a twelve-month contract and pays the full amount upfront. Cash arrives in the operating section of the cash flow statement on day one. Revenue, under IFRS 15 guidelines on revenue from contracts, is released as the performance obligation is satisfied. The finance team ends up mapping the deferred revenue amortization schedule day by day, spreading that twelve-month payment across the contract year while the liability on the balance sheet unwinds in parallel.
Read only the income statement and you see a company earning a twelfth of its contracted value. Read the deferred revenue balance alongside it and you see the whole book.
Where ARR Stops Working
ARR analysis breaks down entirely when it is pointed at transaction-based digital platforms. A voucher distribution business or a payments rail does not have contracted recurring revenue in any meaningful sense; it has volume that can evaporate in a quarter. For those entities, Gross Merchandise Value and take-rate stability dictate the valuation model, and importing ARR language into that conversation produces a number nobody can defend.
The practical test is simple. If a customer has to make an active decision to transact again, you are looking at GMV. If a customer has to make an active decision to leave, you are looking at ARR.
Before You Quote ARR
Confirm the subsidiary's revenue model before applying the metric. Mixed entities — part subscription, part transactional — need both figures reported separately, never a blended headline number that flatters the recurring portion.
Marketing Spend as a Capital Decision
Reclassify sales and marketing in your own model. Not in the audited accounts, where it belongs exactly where the standards put it, but in the analytical model you carry into the valuation discussion. Money spent acquiring a customer who will pay for three years is capital deployment wearing the costume of an operating expense.
The arithmetic that follows is unglamorous and decisive. Customer Acquisition Cost is fully loaded marketing and sales spend for a period divided by customers acquired in that period. Lifetime Value is gross-margin-adjusted revenue per customer multiplied by expected lifespan. The ratio between them tells you whether the subsidiary is compounding capital or burning it with better graphics.
Evaluating customer cohorts over a 24- to 36-month horizon is where this stops being theory. Shorter windows flatter every business, because the churn has not had time to arrive. Longer windows on a young subsidiary produce cohorts too thin to interpret, and a cohort table built on fewer than three or four renewal cycles carries limited signal no matter how confidently it is presented.
Red Flags in the Footnotes
Portfolio managers earn their keep in the footnotes. The reallocation work involves pulling operating expenses apart line by line and hunting for marketing spend that has been quietly reclassified somewhere less visible.
- Brand campaigns booked under general and administrative rather than sales and marketing, which mechanically lowers reported CAC.
- Partner incentives, referral fees or channel rebates netted against revenue instead of shown as acquisition cost.
- Sales headcount charged to product or engineering because those staff "support implementation".
- CAC calculated on gross new customers while churned accounts quietly reduce the denominator elsewhere.
None of these are necessarily improper. All of them change the ratio you are about to underwrite.
Churn Compounds in Reverse
Recurring revenue is a compounding machine, and churn is the mechanism that runs the machine backwards. A subsidiary adding impressive new logos every quarter while losing an equivalent base underneath is running very fast on a treadmill, and the income statement will not tell you which one you are looking at.
Two numbers separate the cases.
Gross Retention measures how much of last year's revenue base you still hold, ignoring any expansion. It caps at 100% and it is the honest measure of whether customers want the product. Net Dollar Retention adds upsells, cross-sells and price increases from that same existing base. It can exceed 100%, and when it does, the subsidiary grows even if it never signs another new customer.
Cohort tables are where the truth sits. Reading them properly means isolating upsell revenue from existing customers so that gross and net retention can be seen separately, then tracking cohort revenue expansion over a 12- to 18-month lifecycle. What you are watching for is a specific pattern: aggressive new sales volume layered on top of cohorts that decay quickly. Total revenue climbs. The underlying asset erodes.
A subsidiary with 85% gross retention and 118% net dollar retention has a real product with an expansion motion. A subsidiary with 60% gross retention and 105% net dollar retention has a leaky bucket and a very good enterprise upsell team. The headline growth rate can be identical.
What Engineering Costs Reveal in Cash Flow
Software development is the one place where a technology subsidiary gets to choose, within limits, how profitable it looks this year.
The accounting logic is defensible in principle. Development of new features that create future economic benefit can be capitalised and amortised; routine maintenance, bug fixes and research must be expensed as incurred. Longer development cycles, the ones running past three or four months, are typically where the argument for capitalisation gets made most forcefully, because the deliverable is substantial and the useful life is arguable.
The audit work involves isolating capitalized new feature development from expensed routine maintenance within the cash flow statement, then testing whether the split holds up against what the engineering roadmap actually says.
Signals of Aggressive Capitalisation
- Capitalised development as a share of total R&D rising sharply without a matching change in product strategy.
- Amortisation periods stretching well beyond the realistic shelf life of a software feature.
- Reported profitability improving while operating cash flow stays flat, which usually means costs moved rather than disappeared.
- Maintenance-heavy quarters that somehow produce the same capitalisation ratio as build-heavy ones.
Aggressive capitalisation buys short-term earnings and pays for them with technical debt. The bill arrives later, as a platform rebuild nobody budgeted for.
Diligence Sweep Before The Call
Tech Subsidiary Unit Economics Evaluation Checklist
- Isolate upfront marketing spend from routine operating expenses.
- Map deferred revenue schedules across the 12-month contract lifecycle.
- Calculate LTV:CAC ratios using 24- to 36-month cohort data.
- Separate Gross Retention from Net Dollar Retention in every cohort table.
- Reconcile capitalised development against the published engineering roadmap.
- Confirm the revenue model before choosing between ARR and GMV.
Forty-Five Minutes in the Investment Committee
The deliberation ran roughly forty-five minutes. The portfolio manager stayed on her feet the whole time, marker in hand, one whiteboard between her and a committee that had opened by circling the subsidiary's negative short-term EBITDA in red.
She did not argue with the number. She wrote a second one beside it: Net Dollar Retention holding around 120% across consecutive cohorts. Then the gross retention figure underneath, so nobody could accuse her of hiding churn behind upsells. Then the deferred revenue balance, then the LTV:CAC ratio on the 24-month cohorts, then the capitalisation ratio with the roadmap reference written next to it.
Someone asked why the marketing line was so heavy. She drew an arrow from that line to the cohort table and let the committee follow it themselves.
The valuation held. Not because the loss was explained away, but because the room had been handed a different instrument and shown how to read it. When the meeting broke up, the whiteboard still had that retention figure sitting in the top right corner, uncapped and underlined, next to an EBITDA number nobody mentioned again.
