Introduction
At the 2026 AIR Forum in Washington, D.C., one of the standing-room discussion groups was titled "How Do I Say No? Discussing Strategies for Saying No to Stakeholders." Not AI, not analytics, not a new tool. The session that hit a nerve was about declining work.
That tracks with everything else we know about the field. The capacity research puts most IR teams at 40 to 60 percent of capacity spent on ad hoc requests, and the structural analysis shows why: request volume grows with every new system and every new accountability requirement, while team size stays flat. Nobody is coming to fix the ratio for you. The only lever entirely inside your control is which requests you accept.
Saying no well is a skill with components. It is a rubric you can show people, scripts for the situations that actually arrive, and a set of things to offer instead of the deliverable. This post is those three pieces.
Why yes is the default
- Every individual request looks reasonable. Nobody sends a request that sounds outrageous; they all sound like twenty minutes.
- The requester only sees their own ask. They cannot see the queue behind it, so "we are slammed" reads as an excuse rather than a measurement.
- IR offices are service offices by temperament and by org chart. The work that built your reputation is the work people bring to you.
- No is a one-time awkward conversation; yes is a quiet commitment you pay for over weeks. The brain picks the option that hurts later.
- The requests that arrive with authority (a dean, the cabinet, an accreditor) feel non-negotiable even when they are.
The real cost of a reflexive yes
It crowds out the work only you can do
Every hour spent on a one-off pull is an hour not spent on the reporting calendar, the peer benchmarking, or the analysis that anticipates the provost's next question. The intake-quality research found most requests need ten or more clarification rounds before work even starts. That friction is capacity you already spent.
It trains the campus to bypass prioritization
Every yes to an out-of-band request teaches one more person that the official channel is optional. Within a year, the intake form is a formality and the real queue lives in your inbox, untracked and unprioritizable.
It hides the capacity problem from the people who could fix it
If you absorb every spike, leadership never sees the spike. The data that would justify a hire, a tool, or a restructure is exactly the data you destroy every time you quietly absorb the load.
It degrades the quality of the yes
A team at capacity ships late, ships rushed, or ships wrong. The whitepaper found over 91 percent of institutions report challenges getting accurate answers to data questions; overload is a major reason. You cannot protect accuracy while accepting everything.
A triage rubric you can defend
| Criterion | The question to ask | High score looks like |
|---|---|---|
| Decision at stake | Is there a decision waiting on this, and who owns it? | A cabinet or board decision with a date |
| Deadline reality | What happens if this lands a week later? | A regulatory or accreditation date that cannot move |
| Effort to done | New analysis, or a variant of something we produce? | A cut of an existing report, hours not weeks |
| Repeatability | Will anyone ask this again next term? | Yes, and we can productize the answer |
| Requester reach | Is this one person's curiosity or a campus-wide need? | Serves multiple units or an external mandate |
Score each request one to five on each criterion and add them up. The number matters less than the fact that you scored it, because a rubric converts "I do not feel like it" into "this ranked below twelve other requests this month." That is the difference between a personal refusal and an institutional one.
Publish the rubric. Put it on the intake form, or link it from the confirmation email. Once requesters can score their own request before sending it, half the low-value asks disappear on their own, and the ones that arrive come pre-argued. If you want the intake layer to do this filtering for you, that is most of what Clema's intake does.
Four situations, four scripts
The dean's "quick pull"
Low effort claimed, high effort actual
The ask arrives framed as trivial, usually with "should be easy for you." Do not argue about whether it is easy. Answer with the queue, not the effort. Script: "Happy to. It lands behind four commitments with dates attached, so realistically it is three weeks out. If it needs to jump the line, tell me which of those four moves, or what date you need and I will tell you today whether it is possible." You said yes to the work and no to the timeline, and you made the tradeoff visible instead of absorbing it.
The accreditation panic
Real deadline, undefined scope
This one you cannot decline outright, but you can decline the unbounded version. Scope it before you accept it. Script: "Before I start: which standard and which element? Last cycle this section was 40 hours because the evidence cut across three systems. If you can get me the reviewer's exact ask, I can usually cut that in half." Accreditation requests expand to fill the anxiety around them; the script shrinks them back to what the reviewer actually asked.
The faculty research request
Worthwhile, but it is their project
A faculty member wants a custom extract for a study. This is collaborative work, not service work, and it should be scoped as such. Script: "This is doable as a collaboration: you own the research questions and IR reviews the definitions. Realistically it is a next-quarter project, not this month. If you need data sooner, here is the dashboard cut that covers the base population." You keep the relationship and the study, without becoming uncredited co-author of every grant on campus.
The cabinet one-off
Strategic ask, no repeated value
An executive wants a bespoke analysis for one meeting. The trap is that it becomes monthly. Script: "I can build this for the meeting. To make it repeatable rather than a one-off, I would build it as a standing view with a refresh date, which costs about the same. Want that version?" If they say no, you have learned the real shelf life of the request, which is information too.
Making no sustainable
Scripts rust without infrastructure. Three things keep a functioning no alive past the first month.
Track the queue where everyone can see it. Volume, turnaround, and aging by requester and unit. The moment "we are slammed" becomes a chart, it stops being an excuse and starts being an argument, and it is an argument you win. The numbers behind that argument are in the capacity benchmarks.
Route everything through intake. Not to be bureaucratic, but because a request that arrives by hallway or direct email carries no scope, no deadline, and no accountability. The intake form design research covers what a form has to capture to make triage possible at all.
Schedule the conversation about seasonal load. A no in October, during IPEDS peak season, needs different framing than a no in July. The IR year in seasons gives you the calendar to point at: "Not this week, we are in collection lock, and here is the two-week window that works."
One last thing. The goal of all this is not to say no more often. It is to make your yes mean something. A team that declines cleanly, explains briefly, and delivers what it accepts on the date it promised earns more authority than a team that says yes to everything and ships late. That authority is what eventually gets you the hire, the tool, or the restructure that fixes the ratio for real.
Make intake do the saying no for you
Clema captures requests, scores them against your rubric, answers the ones the data can already resolve, and escalates the rest with scope attached. The queue stays visible without you being the gatekeeper.
Book a Demo