Cash Conversion That Isn't: Nordwind Software's Quality of Earnings
Nordwind Software reports FY revenue of $480m (prior year $400m), net income of $36m and operating cash flow of $92m. Capex was $18m, including $12m of capitalised development. Management headlines 'adjusted EBITDA of $132m', excluding $31m of stock-based compensation, $12m of restructuring and $4m of acquisition costs. Two items changed during the year. The useful life of capitalised software was extended from five to eight years, cutting amortisation by $8m. Days sales outstanding rose to 78 from 62, and deferred revenue grew $60m versus $35m the year before. A sell-side note values the shares at 12x EV/adjusted EBITDA and calls the $74m of free cash flow 'best-in-class conversion'. The CFO says the DSO increase reflects a shift to larger enterprise customers on 60-day terms and that deferred revenue is 'just billings timing'. Nordwind has no debt and 60m shares at $26.
1. Recompute free cash flow on a normalised basis and the cash conversion ratio. Identify each driver of the difference from the reported figure.
Read the worked answer
Reported FCF = 92 − 18 = $74m; FCF/net income = 2.06x. That is the number being celebrated. Two working-capital items explain most of it.
Receivables. At the prior 62 days, receivables on $480m of revenue would be 480 × 62/365 = $81.5m. Actual is 480 × 78/365 = $102.6m. The extra ~$21m is cash borrowed from customers, not earned.
Deferred revenue. Growth of $60m versus $35m. Some is structural: a business growing 20% would normally add 35 × 1.2 = $42m, so the excess is $18m. On a strict view the entire $25m step-up is timing.
Normalised FCF ≈ 74 − 21 − 18 = $35m (bottom-up case: $28m if none of the $25m step-up is sustainable). Against reported net income the ratio falls to about 0.97x; against net income normalised for the amortisation change (36 − 8 = $28m) it is about 1.25x.
Careful with the $8m amortisation change: it is non-cash, so it does not affect operating cash flow - do not add it back to cash flow. It only inflates earnings, so it belongs in the profit adjustment, not the cash adjustment. The practical conclusion: the 2.06x is a balance-sheet artefact. A software business with a genuine 2x conversion would show stable DSO and unbilled/deferred balances tracking revenue.
2. Assess the adjusted EBITDA figure and the useful-life change. What is a defensible EBITDA for valuation?
Read the worked answer
Adjusted EBITDA of $132m is a 27.5% margin. Strip out what is real:
- SBC of $31m is compensation. Excluding it while using a diluted share count - which already reflects the shares issued - double counts the benefit. SBC-adjusted EBITDA = $101m, a 21.0% margin.
- Restructuring of $12m is a judgement call: recurring for three years means it is operating cost; a genuine one-off can be added back. Allow half, $6m.
- Acquisition costs of $4m are a legitimate add-back for a one-off transaction.
Defensible EBITDA: roughly $101m to $107m, against a headline of $132m - 19% to 24% lower.
The useful-life extension from five to eight years adds $8m to EBIT and to net income, which is 22% of reported net income. It is not a cash item and it does not affect EBITDA (amortisation is already excluded), but it is a strong signal about how the numbers are being managed: eight years for software implies much slower obsolescence than most SaaS peers assume, and the change conveniently lands in a year with weak conversion. The honest presentation is to show EBIT and EPS under the old policy alongside the reported figures, and to ask what the actual replacement cycle of the capitalised product is.
3. The shares trade at $26 on 12x EV/adjusted EBITDA. Which metric should anchor the valuation, and what three diligence questions would you put to management?
Read the worked answer
Anchor on normalised free cash flow and treat EV/EBITDA only as a cross-check with SBC expensed.
At 12x: 12 × 132 = $1,584m EV, or $26.4 per share - essentially the market price. On $101m the same multiple gives $1,212m, or $20.2 per share. On normalised FCF of about $35m, the FCF yield at a $1,560m market capitalisation is 2.2%, not the 4.7% implied by $74m. The multiple itself was presumably set on a peer group with clean EBITDA, so applying it to a flattered figure compounds the error.
Diligence questions:
- DSO and receivables quality. What are the contractual payment terms in the new enterprise contracts, and what does the ageing beyond 90 days show? If the mix shift is genuine, DSO should stay at 78 and stabilise; if it is a year-end stretch, it reverses in Q1 and takes cash with it.
- Useful-life change. Why now, and what evidence supports eight years - actual retention curves, version-release cadence, or an independent benchmark against the five years used previously?
- Deferred revenue and billings. Is the $60m increase invoicing timing (annual prepayments) or genuine multi-year commitments? Ask for current remaining performance obligation and the billings trend for the next four quarters.
Also check whether $12m of capitalised development is being expensed by peers; if not, Nordwind's EBITDA and EBIT are structurally flattered and the amortisation will arrive later. Conclusion: the company may be fine, but the headline conversion and the valuation are being supported by reversible balance-sheet movements, so the risk is asymmetric.