On-time delivery

Late delivery is almost never a speed problem, and reading this number as one is the common mistake. Work slips where it QUEUES, and in an agency it queues in a small number of predictable places: a single reviewer everybody routes through, a client whose approvals arrive late, a kind of work whose scope is still open at the point the date was promised. The share tells you it is happening. It cannot tell you where the queue is, and the difference decides the response, because adding capacity to a team that is waiting on one approver changes nothing at all. On this workspace 4 projects missed a date. The worst is DTC Site Replatform at 43 days. The section above lists them together, which is where a shared reviewer, a shared client or a shared phase becomes visible.

Look at DTC Site Replatform, Website Redesign and Listings Integration and Q2 Rebrand and Packaging Refresh together and ask what they SHARE rather than how late each one ran. A shared reviewer, a shared client or a shared phase is the fix; three unrelated slips are a promising problem rather than an execution one.

This number moves for reasons that concentrate, so the useful unit is the pattern rather than the total, and the projects are named above so the pattern is readable rather than remembered. One thing to carry into it: of the projects on file, 0 count as on time on a status claim with no completion date behind them, and 5 are in flight and counted on the tracker's word. The share is softer than it looks, and the named list is the part that is not.

This closes no measured driver above, and does not claim to.

Commit dates against the review path rather than against the work, and set a level for this tile once there is enough history to set an honest one. Most missed dates were unrealistic when they were given, not mishandled afterwards.

Two structural pieces. The first is that a date gets committed after the approval path is known, so the promise carries the waiting time it will actually need rather than only the working time. The second is the target itself: this tile is currently read against nothing, which means a period can get materially worse without anything crossing a line and without anybody being wrong. Neither of those is a product setting. The first is how work gets sold and the second is a decision nobody has made yet.

This closes no measured driver above, and does not claim to.

Late delivery compounds in a specific order and only the last step is visible. A slipped project pushes the next one's start, so the team begins behind, and the recovery comes out of margin because those hours were sold once and worked twice. The client-facing part arrives last and arrives as a decision already taken: clients rarely complain about dates, they simply stop renewing. By the time this shows up in retention, the conversation that would have saved the account was available two quarters earlier.