The Service Decision Tree

Junior faculty are often told to protect their time, but nobody provides instructions for how to do so. As an assistant professor at a public university, I have struggled to balance my course load, my research projects and the constant stream of invitations to serve on committees, task forces and affinity groups. I had a hard time saying no, especially when an invitation included a reference that serving would “look good on your CV” or “look good for tenure.” And the more I agreed to participate in service engagements, the more frequently colleagues invited me. I took this as a compliment, but after a few semesters of overcommitting, I realized I needed a more structured approach to assess service opportunities and reduce the guilt of saying no when necessary. I enlisted the help of an academic coach, and from our conversations came the Service Decision Tree.

The Service Decision Tree

The result of an extended design process, the decision tree consists of a series of three major steps, each with several secondary questions to consider.

The decision tree starts by asking questions of relevance and alignment: Does the invitation align with the service, teaching and research requirements for tenure? This step also asks if the service aligns with my professional values, a nebulous term that was nonetheless necessary.

Invitations that pass the first level are then evaluated based on a series of questions focused on logistical and social factors.

1. Is the invitation for a single event or an ongoing series of commitments?
   1. Single events are a smaller commitment. This question is a reminder of time availability and long-term consequences.
2. Do I have more than enough time to fit this commitment into my schedule?
   1. This might seem obvious, but it can be easy to underestimate the time a service engagement will require and how it will impact availability for other duties like teaching and research.
3. Opportunity cost: Would I have to give up anything else if I accept?
   1. This question works like a safety valve. If the service requires giving up something else that is essential, then it cannot be considered further.

4. Option to get out: Can I back out or adjust my commitment after the fact without major consequences?
   1. Plans often change. Having an option to cancel or back out gracefully helps.
5. Do I feel compelled because others are doing this (a.k.a. FOMO)?
   1. Sometimes social pressure motivates decisions more than rational assessment. This question probes at the bandwagon effect and seeks to minimize its role in accepting an invitation.
6. Do I feel compelled because nobody else is doing this?
   1. This question examines the other side of the coin of social motivation. Some people like to help, especially if nobody seems to be filling the need, but this is not a good reason to accept a service invitation and can easily result in overcommitment.
7. Is the invitation personalized or is it a general call?
   1. We all have unique abilities and interests to bring to our service. This question places special value on service invitations that are individually addressed and deprioritizes general calls made to everyone.

This second set of questions is primarily designed to draw out additional contextual information about a potential service engagement. If the ultimate point of a decision tree is to lead the user to a yes/no answer, this set of questions probes a wide range of possible reasons that might bear on that decision.

The third step in the service decision tree focuses on selecting an appropriate standard for making decisions. Whether we adopt an open-minded willingness to accept the invitation or take a more defensive stance that automatically declines service unless there are compelling benefits depends on the responses to questions in Parts 1 and 2. Although systematic, the decision tree still relies heavily on personal judgment—this final step helps users orient their judgment for each situation.

Initially, I thought it would make sense to categorize any service request into one of two groups: “conditional consideration” would be used for any request that seemed worthy of further consideration and “strict scrutiny” would be used for any request that appeared to be unacceptable or unappealing. I chose these legalistic labels not because I wanted a rigid system to classify cases but because their alliteration made them easy to remember, and most importantly, I found them a little funny. As my design process developed, incorporating humor helped maintain flexibility, especially when sharing with others for feedback.

At the wise suggestion of my academic coach, I shared an early draft of the decision tree with the chair and vice chair of my department as well as the dean of the school. Sharing the decision tree while I was still developing it served a few critical needs. First, it involved my supervisors in the design process as active commentators. Rather than waiting until the end and delivering it to them like a process server, I asked for their review and input while it was still in draft form so they could be part of its development.

Second, by sharing a copy of the decision tree with my chairs and dean early on, it eliminated any element of surprise that might come once I started using the tool to make decisions that invariably meant declining some of their invitations. The decision tree meant I would be accepting far fewer service requests, and I wanted them to understand my new approach was deliberate and determined using a transparent metric.

Decision trees are useful diagrams that systematize and visualize points of consideration and foreclosing choices that take place in a decision-making process. The usefulness of a decision tree depends on how well it balances specificity with flexibility. By being tailored specifically, the tree helps users make decisions more systematically and consistently across like scenarios. At the same time, a useful decision tree also needs to allow for some flexibility, as no two cases will be identical and the tool needs to allow users to arrive at contextually sensitive judgments in unanticipated scenarios.

Making Decisions With the Support of AI

The most unexpected part of developing and using the decision tree was its near simultaneous development alongside the rise of ubiquitous generative AI. Because the decision tree is designed to function systematically, AI became a highly useful tool and “thought partner.”

I created a chat in ChatGPT 4.0 dedicated to the service decision tree. After priming it with a PDF copy of the tree, I added as much relevant context as possible. I added PDF copies of my department’s tenure standards as well as the universitywide retention, tenure and promotion policy. I supplemented this with a copy of my most recent CV as well as narratives from my third-year review portfolio about my teaching, service and research. I wanted the AI to know as much as possible about me and my context so it could accurately analyze invitations using the decision tree. Fortunately, I didn’t have to write anything new. At this point I was just putting the materials into relation with each other to help the AI understand how to implement the decision tree.

To test the system, I pasted in a few actual emails from recent service invitations to see how the AI system would assess them. I quickly found out that static and objective criteria were easy for it to assess, but ascertaining and applying my values was a little more difficult. I used these tests as an opportunity to tell the AI some of my values (e.g., I value service where my input is respected and where the service leads to meaningful outcomes for students, the university and my broader discipline). This took time, but the fine-tuning resulted in detailed situational analysis that aligned with my perspective.

With all the content primed, I loaded in the text of an email invitation that had recently come into my inbox. I asked ChatGPT to consider the offer based on what it knew about me, using the decision tree. To my astonishment, it applied the standards methodically and generated an analysis report! It didn’t just say “do it” or “don’t do it.” It showed its work and how it assessed each step and resulted in a recommendation to decline.

Later, I asked the AI to go one step further: draft an email to decline the offer. Coming full circle, if my original problem was having a hard time saying no, somehow it felt a lot easier if I let AI do it for me—or at least get the process started. In most cases, the AI rejection email was mechanical or blunt, definitely not my style, but seeing the core message on the screen made it easier to revise for tone and then send without too much reservation.

For some, saying no to service invitations might not be that difficult, but for many junior faculty who are seeking to create positive impressions with colleagues, declining an invitation might be challenging. Offloading some of the decision-making process to a predesigned system like this decision tree can make it easier to say no, preserving time for commitments that matter the most.