# Speed _Fast companies attract different customers and different margins._ --- You run a cloud payroll product. Implementation takes four months. Your team has been working on cutting it to three, and the project is going well. Fewer handoffs, tighter scoping, a better onboarding flow. Good operational work. Then you look at the pipeline. --- Three deals lost this quarter, all to the same competitor. Their product scored lower in two of the three evaluations. But their implementation takes three weeks. The first was a 300-person company whose payroll provider had just been acquired. HMRC submissions were due in six weeks. They couldn't wait four months. The second had just completed a management buyout and needed to separate payroll from the parent company before the next pay run. The third had hit 80 employees and outgrown their accountant doing payroll manually. Their finance director couldn't justify a four-month project to her board. None of them chose on price. None of them chose on features. They chose on time, because the cost of waiting exceeded the cost of the premium. These aren't FTSE 100 companies with dedicated payroll teams and 18-month procurement cycles. They're the messy middle: growing fast, making decisions in weeks, needing payroll running by next month. Cutting your implementation from four months to three wouldn't have changed the outcome. Three months to three weeks is a different market. --- Four months of implementation. How much of that is work? Scoping runs to two meetings, roughly four hours of conversation, spread across three weeks because calendars don't align. Employee data migration is a day's work, but the customer's HR manager takes three weeks to clean up their export from the old system. Pension provider integration needs about four hours, but the ticket sits in the professional services queue for twelve days. Running a parallel payroll test is half a day, booked six weeks out because the finance director is travelling. About three weeks of work spread across four months. In manufacturing and logistics, products add value roughly 0.05 to 5% of the time they spend in the system. Software implementations follow the same pattern: the work is fast, the gaps between work are slow. --- Your competitor's product has comparable complexity, a similar-sized team. They eliminated the gaps. The customer uploads employee data through a self-serve import that validates and maps fields in an hour. Pre-built pension provider connections replace custom integration work. Tax configuration happens during signup. A parallel payroll test runs inside the onboarding flow instead of being scheduled as a separate phase weeks later. Three weeks of work is still three weeks of work. Four months of calendar time collapsed to three weeks because the waiting disappeared. --- At four months, every configuration request is a professional services engagement. A non-standard pay structure or pension scheme variation means scoping, scheduling, delivery, review. The overhead makes small requests uneconomical, so you standardise and tell customers to fit their processes to your defaults. At three weeks, configuration happens inside the onboarding flow. A pay structure that would have been a two-week professional services project becomes a same-day change. The marginal cost of variety drops because the overhead that made it expensive was the waiting, not the work. The slow competitor can't reach either group: the messy-middle companies that need to be operational in weeks, or the ones that need non-standard configurations without a professional services engagement. This forces architectural choices, because the product has to be built for self-service from the start, which constrains what you can offer at launch. --- George Stalk identified this pattern in manufacturing thirty years ago. His research in [[Competing Against Time]] found that companies built around time compression grew three times faster than their industry averages and earned roughly twice the profit. Time is money. ---