The review queue has one person's name on it

Consider a team arranged like this. A senior analyst owns an investigation into declining repeat use. Other analysts bring them model changes, cohort comparisons, and recommendation decks before anything reaches a stakeholder. They know the awkward history behind the definitions, can spot a change in grain, and ask the question that makes a recommendation hold together.

The team’s work keeps moving through their reviews. Their own investigation keeps slipping. At planning, both the investigation and the review work are still expected to fit, because only the investigation has an estimate beside it.

That is an illustrative arrangement, but the organizational choice is specific. The team has assigned someone responsibility for other people’s delivery without reducing what it expects them to deliver alone. The company depends on the review. The analyst’s visible commitments barely acknowledge it.

In Giving the models away, I describe how moving authorship outward leaves substantial review work behind. That work can concentrate again, this time around a capable individual rather than a central engineering queue. More people can write a change, but the same person still has to make it safe and explainable.

I would treat that concentration as a decision about the team’s capacity and development. Asking the analyst to become quicker at reviewing leaves both decisions unmade.

Find out what the signature is supposed to mean

The first question is what people expect the senior analyst to settle. A request for a quick check can contain several different jobs.

Work arriving for review	Question that needs an answer	Responsibility to make explicit
A model change	Does the implementation preserve the intended grain, definitions, and dependencies?	A qualified reviewer for the affected model and its consequences
An analysis	Does the evidence support the claim, with the relevant comparisons and uncertainty shown?	A reviewer able to assess the method and business context
A proposed action	Is this a decision the organization should make?	The person authorized to make that decision, informed by the analysis

One person can contribute to all three. Their approval of one does not automatically cover the others. A sound cohort comparison can support several reasonable actions, and an analyst who reviewed it may have no authority to choose among them.

Look at a recent sample of reviews and record what the senior analyst actually supplied. Some may have caught defects. Others may have reconstructed missing context, rewritten the argument, or resolved a definition that nobody had agreed. Those consume different kinds of time and need different fixes.

If the work routinely arrives before its author can explain the question, the review is also doing part of the analysis. Training, a clearer assignment, or access to the stakeholder may be required. Moving the same unfinished work to another reviewer simply moves the burden.

If several reviewers could assess the work but only one person’s approval is accepted, permission is the constraint. The manager needs to change the approval arrangement and tell the people who rely on it. A capable colleague cannot acquire that authority by volunteering for more comments.

Put review on the plan before assigning the next investigation

Review needs time when someone is available to do it. Calling it part of everyone’s job does not establish that availability.

Estimate the recurring load from actual work: time spent assessing it, following up, and helping authors resolve the substantial issues. Keep the measurement proportionate. A short sample with a few recognizable categories is enough to start a capacity conversation; it need not become a permanent timesheet for every helpful exchange.

Then make a choice in planning. Reduce the analyst’s delivery commitments, assign some reviews elsewhere, or reduce the work entering the team. Protected review periods and an agreed response expectation can make the arrangement predictable. They cannot create more hours than the plan contains.

Urgent work needs a route too. If a consequential decision cannot wait for the next review period, say who can interrupt the reviewer, why, and which planned work moves as a result. Without that last part, every exception becomes an invisible addition to the same person’s week.

For an analyst who does not control the plan, the useful evidence is a small record of these collisions: what arrived, what judgment it required, and which commitment it displaced. Take a proposed allocation to the manager. The manager then has to choose what the role contains. Repeatedly asking the analyst to negotiate each interruption alone leaves them carrying a staffing decision through personal goodwill.

The review work already has a cost. Leaving it off the plan only hides which commitment will pay for it.

Give the next reviewer a question they can answer

A rotation helps only when the people on it can perform the assigned review and have time to do so. Putting more names on a list without those conditions creates the appearance of coverage while questions still return privately to the original reviewer.

Start with a bounded area: routine changes to a model another analyst already maintains, or a familiar type of analysis with a well-understood comparison. Agree which consequences that person can assess and which require further expertise. A small change to a shared denominator can matter more than a large local presentation change. Route by consequence and familiarity, rather than by the number of lines changed.

The author should bring enough evidence for the review to begin: the question being answered, what changed, the checks already performed, the unresolved concern, and when the result is needed. The reviewer still investigates. They should not have to discover the assignment from a nearly finished deck.

For a change to a retention analysis, for example, the next reviewer might take responsibility for checking cohort eligibility and whether each cohort has had time to mature. The established reviewer can explain why those checks matter, including a concrete example where the conclusion changes. The next reviewer then leads the assessment and records their reasoning.

During this learning period, additional review will cost more. Make its end condition explicit: the new reviewer can explain material risks, recognize a limit, and support an approval within the agreed scope. Once that is true, routine work in that scope should stop requiring the original person’s final signature. Keeping that signature forever would retain the queue while adding another step ahead of it.

Some areas will still require the specialist. A small team may not yet have another person able to assess a complex shared model. Keep that limit visible, plan around it, and decide whether to fund development or additional capacity. Independence is something to establish with evidence.

Leave a review another person can learn from

The useful residue of a review is the reasoning that changed the work. An unexplained approval gives the next reviewer little to reuse. A rewritten query can fix today’s result while leaving the author unable to explain why the original was wrong.

For a material issue, record the claim being challenged, the evidence, and what would resolve it. If a join duplicates accounts, identify a case that shows the duplication and ask the author to verify the intended grain. If the analysis treats association as cause, identify the alternative explanation the recommendation has to survive. A few concrete examples can teach more than a growing list of general reminders.

Also distinguish a reason to block release from an optional improvement. Otherwise authors have to infer whether every preference is a requirement, and the most exacting reviewer quietly sets the capacity of the whole team.

Google’s code-review standard separates changes needed for code health from nonessential polish and educational comments. It treats knowledge sharing as part of review while asking reviewers to make optional feedback recognizable. I would apply that distinction to analytical work too, with the additional question of whether the evidence supports the decision. That extension is my recommendation; the guidance is about code review.

Repeated comments can reveal work that should move out of the review queue. A stable rule about an identifier may belong in an automated check. An unsettled metric definition needs a named decision owner. A recurring mistake in cohort construction may justify a teaching session and a worked example. The senior analyst’s contribution is partly recognizing which response fits.

Stop routing every disagreement back to the original reviewer

Sharing responsibility allows different judgments to appear. If every disagreement is settled by asking the original reviewer what they would have done, the new authority lasts only until it is tested.

Require reviewers to explain their concerns against the agreed standard and the consequence of being wrong. When two approaches are acceptable, allow the author and assigned reviewer to make the choice. When the disagreement concerns business meaning or the risk of acting, route it to the appropriate owner. Seniority alone does not settle that question.

Use selected joint reviews to examine decisions, particularly new failure modes and changes with broad effects. Look for risks that were missed and places where the standard was unclear. Sampling can expose gaps; it cannot establish that all the unsampled work was correct. Higher-consequence work may continue to require specialist review before release.

This arrangement also needs support when the original reviewer is away. Exercise the agreed coverage during ordinary work, while help is available. A person’s leave should not be the first time the team discovers which approvals nobody else is allowed to give.

The related problem in When every decision still needs you is a manager retaining context and authority. Here, the analyst may manage nobody. The manager has to establish the boundaries, defend the reviewer’s time, and back decisions made within the new arrangement.

Recognize the judgment that no longer needs a signature

A senior individual contributor needs an account of their work that survives beyond being described as helpful. Counting comments or approvals is a weak substitute: it can reward a growing dependency and say little about whether the review improved anything.

I would look for a small set of inspectable contributions. A recommendation became better supported because a reviewer exposed an assumption. A recurring failure became a check that now catches it earlier. Another analyst can review a defined class of work and explain the difficult cases. The examples should show the reasoning and its use, without attaching invented financial value to a mistake that might have happened.

Read that evidence alongside waiting time, rework, defects found after release, and the health of the reviewer’s own commitments. Fewer reviews by the senior analyst can be progress if qualified colleagues are carrying them. It can also mean that people have stopped asking. The surrounding evidence is what distinguishes the two.

Agree how review, teaching, and maintaining standards count in the role before the performance conversation. They need room in the goals as well as mention in the feedback. The analyst should also retain work that develops their own judgment. A permanent assignment to repair everyone else’s nearly finished analysis can consume the opportunities needed for their next step.

This takes capacity before it returns any. Pairing slows an early review, another reviewer will make choices the specialist would not, and some work will need correction. The original reviewer also gives up the visibility of being the person every consequential piece passes through. Those costs need to be named and supported, rather than presented as a personal failure to delegate.

A senior analyst should be able to grow by making judgment available to others, without remaining the signature every result needs.

For the team at the start, the useful change is an explicit review commitment, credible coverage, and a delivery plan that fits both. The investigation can then compete for time honestly. The analyst’s contribution includes the work they complete, the decisions their review improves, and the judgment that other people can now exercise without waiting for them.