Alfa Financial Software: Investors Sold the Headline Slowdown While Its Recurring Engine Accelerates

When Alfa Financial Software posted its 2025 results in March, the numbers looked like a win. The London-based company that builds the software banks and auto lenders use to run leases and equipment-finance books grew revenue 17% at constant currency to £126.7m, lifted operating profit 17% to a 31.6% margin, and converted 97% of that profit into free cash flow. The shares fell 7% that day anyway, and it stands near 166p now, down roughly 14% on the year. Meanwhile analysts still carry a target of about £2.67 a share — some 60% above the market price.

That gap is the whole question. Both sides can be right if the market is pricing the slowdown in the headline numbers while the company's most durable part keeps compounding. The detail worth untangling is whether the selloff has reset the valuation faster than the business has actually deteriorated, or whether analysts are clinging to a growth story the reported accounts no longer support.

The engine the headline misses

Alfa's revenue splits into three pieces. Subscription — the recurring, always-on licences customers pay for year after year — grew 16% in 2025 and 14% in the first half of 2026, to £24.1m, and has climbed to 37% of the mix. The other two arms are lumpier: Delivery (implementation work on new customers) and Software Engineering (bespoke development). Those jump around by project and are where much of the weakness sits.

The health of the recurring engine is best read from two metrics software investors watch closely. Annual recurring revenue, or ARR — the annualized value of the contracted subscription stream — rose 17% to £48.5m in H1 2026. And net revenue retention came in at 110%, meaning the customers it already has are paying it 10% more than a year earlier, before counting any new business. Total contract value jumped 17% to £247m, with the subscription portion of that backlog up 21%. In plain terms: on the part of the business that repeats, growth is not just holding up, it is accelerating.

Management frames the disconnect in a way that sums up the bull case. On the H1 call, Chief Executive Andrew Denton and CFO Duncan Magrath argued the market undervalues "revenue in the oven" — subscription fees already locked in for customers still working through implementation — and worried the company gets lumped in with per-seat software firms, even though Alfa licences by lending volume, not by seats. That last point matters because it is the direct rebuttal to the fear that AI will shrink its billings: if a customer books more loans, Alfa's fee rises regardless of headcount.

Why the stock sold off anyway

The bear case is not invented. Management itself guided in March that it does not expect to repeat 2025's 15% growth this year, and that software engineering revenue — historically a profitable stream — would fall roughly 35% in 2026 as implementations get simpler and customers stop commissioning bespoke work. That is a genuine slowdown, and it landed the share price.

First-half 2026 confirmed the shape of it. Total revenue grew just 5% at constant currency, as the software engineering line fell 17%. Reported operating profit fell 15% to £18.4m. Dig into that decline and the quality issue becomes clearer: £1.6m of it was one-off severance tied to 31 non-voluntary departures in product engineering, and another roughly £2m was the swing in currency hedging versus a year ago. Strip those out and operating profit actually rose 2%. That is the tension in a single number — the profit the market saw is being dragged by costs the company has already taken and by accounting noise, not by the core business weakening.

Subtract the transient and you get a company that churns out a 28–32% operating margin, holds net cash (£22.2m at June), targets 90–100% cash conversion, and pays an ordinary dividend plus a special dividend each year. At roughly 16 times trailing earnings with no debt, that is not a rich multiple. The most honest framing is that the market is pricing the lumpy, visible part of the business — the shrinking project work and the guided slowdown — while analysts focus on the recurring engine compounding underneath it.

What would prove one side right

The question is conversion timing, and it is resolvable within the next two quarters. Subscription is still only 37% of revenue, so the accelerating ARR has to overwhelm the drag from delivery and software engineering for the headline number to turn. Management says the H1 severance savings will more than pay back by year-end, lifting the full-year profit — the first test lands with the H2 results. The bigger proof point sits in a marquee US auto-finance customer that went live on a new-business pilot in H1 and is expected to migrate its existing finance book, which management describes as its largest cloud implementation to date. If that converts into contracted subscription revenue while ARR keeps compounding in the high teens, the 166p price looks like it has priced in a decline the engine does not show.

The weakness is real, and a heavier hand than mine would flag that a company guiding to 10–12% long-term growth is not a compounder — it is a high-margin niche leader selling at a fair, not broken, price. But the valuation has already absorbed the bad news, and the durable engine is pointing the other way. That combination — risk named, multiple compressed, recurring growth accelerating — is closer to a buy-with-something-to-prove than it is to a value trap. The share price 60% below the consensus target is not proof the market is wrong; it is the market placing a large, correctable bet that the acceleration in ARR will take longer to show up in reported profit than bulls expect. The next half-year decides it.