Delivery discount
The gap between the multiple the market pays for software and the one it pays for software that runs like a services business. In private M&A between 2024 and Q1 2026: around 3–3.8x revenue for SaaS, 2.3–2.6x for non-SaaS software and about 1.1x for IT services (Aventis Advisors). Where it’s used →
Absorbed custom work
Customer-specific commitments built into the core product and paid for from the R&D budget instead of being billed. The accounts don’t show it as services; a buyer’s diligence does. Where it’s used →
Product–services boundary
The line between what goes into the core product for every customer and what is built beside it for one customer, priced on its own contract. In most companies at this stage, nobody owns the decision. Where it’s used →
The six indicators
Six measures that show the discount from outside the codebase: ARR per head, R&D share of revenue, configuration absorption, support and success cost, commitment-to-release time, and services mix. Where it’s used →
Three-pile audit
An afternoon’s test. Sort every engineering ticket from the last 90 days into three piles: standard product, single-customer bespoke work, and break-fix or technical debt. If the second and third piles dominate, the delivery discount applies. Where it’s used →
Routing rule
Every customer request gets a destination and a price before it is promised. Only what many customers can buy enters the core; reusable work becomes an enabler; customer-only work goes to the edge. Where it’s used →
Core, Enablers, Edge
The three destinations of the routing rule. The Core: one product, built once and sold many times. Enablers: reusable templates on the public API, owned by the customer. The Edge: customer-specific work under its own priced statement of work. Where it’s used →
The Scaling Ladder
Four tiers by headcount and management layers, each with one binding constraint: Optimise (20–49 people), Categorise (50–99), Specialise (100–249) and Capitalise (250–500). Where headcount and layers disagree, the layers decide. Where it’s used →
Mandate
Decision rights over delivery, granted by board resolution, with one owner and a deadline for every decision that crosses the product–services boundary. The first step; without it the rest is advice. Where it’s used →
Phase 0
The first 100 days of a mandate, with named outputs: the listening report, the spend to redirect quantified, the map of people who hold the legacy knowledge, the signed baseline, and the first senior hires. Where it’s used →
The four handover logs
The records that stay with the company when we leave: the decision log, the baseline book, the benefits tracker and the boundary log. They make the improvement attributable when a buyer’s diligence checks it. Where it’s used →
Two-Week Read
Two weeks of conversations with the executive team, with no fee. The answer is whether there is a number worth going after, and roughly where it sits. Where it’s used →
Figure classes
Every figure we publish carries one class: realised (a cash or contractual outcome, verifiable against records), operational (a measured system improvement, not traced to the P&L), modelled (calculated against a stated baseline) or external (published by a named third party). The classes are never added together. Where it’s used →