Question and scope
This protocol measures how often an active cross-team dependency has no single named person who can confirm status. A named contact is a specific employee, not a team, mailbox, or job title. Dependencies without that contact tend to stall because no one is accountable for answering the coordinator's question. The protocol produces a share, a count, and a short list of the missing names so the team can fix the register.
Scope is one coordination team and a defined window, such as every dependency open during the last two quarters. Include dependencies recorded in a register, ticket system, or recurring meeting notes, provided the record identifies the receiving and delivering teams. Exclude internal tasks that never cross a team boundary. State the systems and the window so another analyst can rebuild the population.
What counts as a named contact
A contact is named when the record identifies one person who is expected to answer status questions for that dependency. The name must be an individual, and it must be paired with a role clear enough that a new coordinator could contact them. A team alias, a shared inbox, or the phrase "the platform team" does not satisfy the definition. The test is whether the coordinator can ask a specific person for a status and expect an answer.
Distinguish the delivery contact from the receiving contact. Both sides need a name, because a dependency can stall when either side stops responding. Record both fields separately rather than accepting one name as coverage for the whole handoff. A record with a delivery name and no receiving name is only half complete, and the analysis should say so. A receiving contact who has never been told about the handoff is not coverage either, so note whether the name was confirmed.
Unit of analysis and exclusions
The unit is one dependency line with one deliverable, one delivering contact, one receiving contact, and one needed-by date. Split a bundled entry when each deliverable has its own owner and date. Keep one unit when several outputs share a single owner and a single commitment. Use the record as it stood at the cut-off, not a cleaned version rewritten during the audit.
Exclude duplicates created when a register and a ticket system both track the same dependency. Exclude test records and entries that were cancelled before any work started. Report the exclusion count and a reason for each category so the denominator is defendable. A small, well-explained sample is more useful than a large list with uncertain boundaries. Keep a short log of what was removed so the count can be reproduced.
Building the dependency list
Start from the register export and the open tickets that reference another team. Search meeting notes for phrases such as waiting on, blocked by, and needs input from, because many dependencies are agreed verbally before they reach a system. Add entries that appear in a recurring review but not in the register. The goal is a complete list of live handoffs, not a list of everything the team has ever discussed.
Give every unit a stable identifier before coding so the same dependency can be matched across sources. Record the source of the entry and the date it was first seen. Mark entries that were found only in meeting notes, since those are more likely to be informal and incomplete. Freeze the list before reading the contact fields to reduce selection bias.
Coding the contact fields
Code each side of the dependency as named, role only, team only, or absent. Named means one individual is identified, with enough detail that a new coordinator could reach them. Role only means a job title such as operations lead without a person attached. Team only means a group name or alias, while absent means the record gives no contact at all.
Record whether the named person is the actual decision maker or only a coordinator for the work. A name is weaker evidence when that person cannot confirm the date without asking someone else. Add a field for the fallback contact and code it with the same categories. The fallback matters because a single name still fails during leave or turnover.
Shared mailboxes and role accounts
Shared mailboxes and role accounts are common in operations, and they are not automatically a failure. A monitored queue can be a legitimate contact when the team has agreed on a response time and a named queue owner. Code the record as role only and note the named owner behind the queue. This distinction prevents the analysis from penalising a working support model.
Do not accept a distribution list that nobody owns. Ask who reads the list and who is allowed to answer for the dependency. If the answer is a group with no accountable person, code the record as team only. The purpose is not to ban queues but to find the ones where a question can disappear. Ask for the queue owner's name and the expected first-response time, then test both on a live question.
Sampling and verification
Use a census when the population is small enough to review in full, which is common for a single coordination team. Above roughly one hundred entries, draw a random sample and publish the method. Stratify by receiving team so a high-volume team does not dominate the result. Freeze the sample before contacting anyone about the missing names. Record the sampling frame and the date the export was taken so the population can be checked later.
Have a second reviewer recode at least ten entries and compare the results. Resolve disagreements by returning to the original record, not by discussing which answer sounds more reasonable. Record the resolution so a later reviewer can see how the definition was applied. If disagreement is high, rewrite the definitions and recode the full sample before reporting. Keep the original and the recoded values side by side in the working file.
Decision rules before coding
Decide before coding whether a dependency with a named delivery contact but no receiving contact counts as partially named or as missing a contact. Both choices are defensible, but the report must state which one was used. Decide whether a name that has left the company still counts, and whether the record was updated after the departure. A departed contact is evidence that the register is not maintained, which is useful information rather than a coding nuisance.
Decide how to treat a dependency that has not started. An entry with no contact may simply be too early for a handoff, and counting it as a failure could overstate the problem. Report not-started entries separately and show the named-contact share for active work as well as the full list. The split keeps the headline honest. Report that split in the same table as the main result so the reader sees both.
Analysis and reporting
Report the share of active dependencies with a named delivery contact, the share with a named receiving contact, and the share with both. Show raw counts beside every percentage, especially when the population is small. Add the count of entries with a fallback name and the count with no contact at all. A single headline number hides which side of the handoff is weak.
Break the result down by receiving team, delivering team, and dependency age. Suppress groups with fewer than five units so one unusual record does not look like a pattern. Pair the contact result with the time each dependency has been open, because a missing name matters more as an item ages. The dependency register is the place to fix the entries after the analysis.
Limits and sources of error
The sources cited here describe internal control and project management practice in general terms; they do not provide a benchmark for contact coverage in a specific register. A missing name in the record may mean the contact was never assigned, or that the coordinator simply did not write it down. The measure also depends on whether the register preserves the original entry or overwrites it. Treat a low share as a prompt to inspect the records, not as proof about how the teams actually work.
Informal dependencies are undercounted when they never reach a written record. Check a sample of meeting notes and messages to estimate the size of that gap, and report it as a limitation. If a large share of dependencies is invisible to the register, fix the intake before setting a target. A precise share built on an incomplete list is less useful than a range with a clear caveat.
Review cadence
Review the result at a fixed point and end with one change, one owner, and one check date. A practical first change is to require two named contacts, one on each side, before a dependency enters the register. Repeat the test after two full cycles before setting a target, because a single quarter may reflect a reorganisation or a system migration. Keep the definitions and the sample method with the result so the next reviewer can compare like periods.
Retire the measure if the register reliably names both contacts and the share stops changing. If the team keeps it, publish the delivery and receiving shares together, because a strong delivery name can hide an unowned receiving side. Use the weekly blocked-work review to confirm that named contacts are actually responding. The purpose is a register that answers questions, not a perfect count.
Sources
- U.S. Government Accountability Office, Standards for Internal Control in the Federal Government, reviewed November 11, 2026.
- Project Management Institute, PMBOK Guide, reviewed November 11, 2026.
- National Institute of Standards and Technology, SP 800-39: Managing Information Security Risk, reviewed November 11, 2026.
