The watch desk
Six bearings from one meridian wall
Every service on this page is a way of holding a fixed line against a moving system. They are sold separately, because most teams need one of them first, but they are designed to fit together into a single continuous record kept by MeridianBD LLC at the Draper watchroom.
One
Bearing Check Desk
A Bearing Check Desk session is the plainest service we offer and the one most teams should buy first. One of our watchkeepers sits with the people who actually run the system, in a room or on a call, and walks the length of the architecture one segment at a time. We ask what each component is meant to do, who depends on it, what happens when it stops, and where its configuration currently lives. The questions are unglamorous and that is the point. Most drift shows up in the answer to a question nobody had been asked in years.
The output is a bearing sheet: a short document that records the stated purpose of each component, the observed state, and the delta between them. Each delta is given a bearing, meaning the direction the problem will travel if it is left alone, and an order in which it should be addressed. We deliberately rank by exposure rather than by effort, because the cheap fix is rarely the one that matters.
What a session covers
A typical session breaks into four passes. The first pass reads the intended design from whatever documentation exists, including the informal notes in a wiki that nobody has opened for two years. The second pass reads the live system through its configuration, logs and dashboards, checking every claim in the first pass against observable evidence. The third pass interviews the operators, because a document can only describe the system somebody thought they were building. The fourth pass writes the bearing sheet and reads it back to the client on the day, while the context is still in the room.
What you receive
You receive the bearing sheet, an annotated component list and a one page summary that a director can read without technical help. Nothing is held back for a later sale. If the sheet finds nothing serious, it says so, and a clean bearing check is a perfectly good result.
Two
Escalation Ladder Rounds
An escalation ladder is a fixed stair of five named levels: raised, sighted, weighted, routed and logged. Rounds are the recurring sessions in which every open issue is walked up or down that stair. The value of a fixed ladder is that movement becomes visible. An issue that has been raised for three weeks and has not been sighted is a fact the whole room can see, and it is much harder for a problem to quietly age out of attention when its rung is written on a board.
A round has a set order. We read the board aloud, one issue at a time. The owner of the issue states what changed since the previous round. The watchkeeper records the statement and assigns the next rung. If an issue moves backwards, which happens often and is not a failure, the reason is written beside it. The round ends when every issue on the board has a rung, an owner and a dated note. Rounds are short by design, usually forty five minutes, because a round that overruns stops being attended.
Cadence and membership
Most clients run weekly rounds during an active programme and fortnightly rounds once the system settles. Membership is kept deliberately small: the internal system owner, one operator, one person from the business side who can speak to consequences, and the watchkeeper. Larger rooms slow the ladder down and turn every round into a meeting. Where a client has more than one system, we run separate ladders rather than one crowded board.
Why it holds
The ladder is not a tool and there is no licence fee. It is a discipline written into the engagement. Because every rung has a name and every note is dated, a new manager can read the history of an issue in minutes. That readability is what keeps an escalation culture alive after the original team has moved on.
Three
Sighting Log Weeks
A sighting log week is a full working week in which the watchroom observes a live system without touching it. We watch dashboards, read logs, follow a handful of transactions end to end and write down what happens. The rule is absolute: no configuration changes, no restarts, no helpful fixes. The week exists to produce a record of behaviour as it truly is, unpolluted by our presence.
Each day of the week has a focus. The opening day establishes the normal picture, which means recording the ordinary rhythm of the system before anything unusual distracts us. The middle days follow specific paths through the system, one per day, chosen with the client during the bearing check. The closing day compares what we planned to observe against what we actually saw and writes the gaps into the log. By Friday the client holds a plain record of a real week, including the quiet oddities that a busy team has learned to ignore.
The log itself
The sighting log is written in time order with three columns of text beside each entry: what we saw, what it suggests, and what we would still need to confirm. We keep observation and interpretation separate on purpose, so that a later reader can disagree with our conclusion without losing the evidence underneath it. The log is delivered as a document the client can edit, quote and archive.
When it is worth it
A sighting log week is most useful when a team disagrees about what the system is doing. A patching window that seems to fail, an integration that seems slow only on certain days, a backup that everyone believes runs but nobody has watched complete: these questions are settled quickly by a week of disciplined observation. It is also the right preparation before an audit day, because an audit without a sighting log tends to redo the discovery work at audit prices.
Four
Azimuth Audit Days
An audit day measures the angle between the process a team states and the process the team actually runs. The azimuth metaphor is exact here: the two lines are both real, they simply point in different directions, and the size of the angle is what matters. We audit one domain per day, such as identity and access, backup and recovery, plant telemetry, release change or incident response, because a single domain done properly beats five domains skimmed.
The day begins with a short briefing in which the client states, in writing, how the domain is meant to behave. We then sample evidence against each claim: an account list against the joiners and leavers log, a backup catalogue against the restore test calendar, a sensor calibration sheet against the maintenance record. Sampling is documented. We state how many items we checked, how we selected them and what we did not check, so that the finding can be judged on its own terms.
The audit report
Findings are ranked by exposure, with a short note on the consequence of each angle remaining open. Every finding carries the evidence we relied on and the smallest action that would close it. Where a finding is disputed, we record the dispute in the report rather than removing the finding. An audit is not a negotiation and its credibility depends on stating the disagreement openly.
Seasonal windows
Audit days are limited in number each season. A watchroom can only run so many audits before attention thins, and we would rather decline a booking than deliver an audit that was written in a hurry. Clients who book a sighting log week first usually get more from their audit day, because the observation week surfaces the questions that the audit then answers.
Five
Drift Correction Sessions
Drift correction is where the watchroom stops observing and starts correcting. A session takes a specific drift that a bearing check, audit or sighting week has already documented and closes it in the open. We insist on one order of work: the record is corrected first, then the configuration, then the documentation that describes the configuration. Doing it in any other order produces a system that is better than its paperwork claims, which is its own kind of hazard.
Sessions are held with the people who will live with the change. The watchkeeper proposes the edit, the internal owner accepts or amends it, and the operator who will carry the daily consequence has a veto. Where a correction needs a maintenance window, the window is scheduled in the session and the change is written into the log with a rollback note before anyone touches a terminal.
How a correction is closed
A correction is closed when three things are true. The system behaves as the corrected record says it should. The change is recorded against the original sighting, so the reader can trace the whole arc from drift to fix. And the documentation that a future team will rely on has been updated in the same week. If any of the three is missing, the correction stays open on the ladder and continues to be read at rounds.
Deliberate restraint
Not every drift should be corrected. Some are harmless artefacts of history, and forcing uniformity on them adds risk without adding safety. Part of the session work is deciding what to leave alone and writing that decision down, with the date and the person who made it. A drift that is knowingly accepted and recorded is no longer a hidden hazard.
Six
Handover Brief Nights
A handover brief night is how every engagement ends. We meet in the evening, when the operational day has slowed, and walk the whole record through with the team who will carry the system forward. The brief is not a summary of our invoices. It is a document written to be read cold by someone who was not in the room, months after the engagement closed, with no ability to call us for clarification.
The evening follows the same shape every time. We begin with the meridian, which is the design the system is now held to and the reasons behind each decision made along the way. We then walk the ladder as it stands on the final night, including the issues still open and the reasons they were left open. We read the sighting log, the audit findings and every correction. We finish with the questions the next team is most likely to ask and the answers we have written down for them.
What is delivered
On the night we deliver a folder containing the brief, the record, the corrected documentation index and a short letter naming the watchkeeper who signed it. The letter includes the watchroom phone line +15512998215 and the watchroom address, and it stays in the folder as a record of who to call if a question falls outside what the brief covers. After the night, the engagement is closed and the record belongs to the client without further obligation.
Why the last night matters
Most systems fail their second owner, not their first. The first team knows the quirks because they lived through the changes. The second team inherits a system and a folder, and whatever is in that folder is the whole of their inherited knowledge. Writing for that second team is the final act of the watchroom, and it is the one we take most seriously.
How the six services run in order
-
Enquiry and scoping
You describe the system and the question you are trying to answer. We reply within two business days with a short note on which service fits and what it would cover.
-
Bearing check
A half day at the desk establishes the stated design, the observed state and the order in which the deltas should be addressed.
-
Observation and audit
A sighting log week and an azimuth audit day build the evidence base, one in observation and one in measurement, so that findings can be defended.
-
Rounds on a fixed ladder
Escalation ladder rounds keep every open issue visible and owned while corrections are planned and scheduled with the people who will live with them.
-
Correction and handover
Drift correction sessions close the documented gaps, and a handover brief night passes the whole record to the team who will run the system next.
Ask the watch desk a direct question
Tell us which of the six services sounds closest to your problem and why. A short description is enough for us to say whether a watchroom is the right instrument at all.