Check begins by considering the purpose of the system from the customer’s point of view. In the loss adjusters case the purpose was clear: to settle claims. But how did the customer judge the process? By equity or fairness, and by time or how long it took. The loss adjusters team quickly recognized that although managers knew the key targets for making appointments, visiting, writing reports and paying cheques, they had no idea how long the whole process took end-to-end, from the customer’s point of view. All measures were focused on internal activities; they had no relation to purpose. When the team plotted the end-to-end claim-to-settlement times on a capability chart (a chart that shows what the organization is predictably achieving) they were startled to discover (as is always the case) that under current conditions a claim could equally well take 10 or 120 days, with an average of 60. In other words, the system was stable (a sort of good news), but (not so good) over a very wide range. This was a revelatory experience. Managers could see instantly that, given that they accepted the theory behind the capability chart, there was no point in arguing about why some cases took so long, or praising people who closed cases in 10 or 12 days. The variation was in the system. Where did it come from? Here was another revelation. When the team, now wholly committed to discovering where the real problem lay, dug deeper, it found that by the time the paperwork reached the branches, it was riddled with ‘dirt’—errors that would stop claims moving forward ‘cleanly’. Thus 80% of claims had wrong or missing postcodes, addresses or phone numbers, lacked some information or had been sent to the wrong branch. Most of the ‘dirt’ was