Example Managing-Up Document

Also known as "How to Work With Me." A worked example of a managing-up document, written by your instructor. You will write your own — for your audit senior, your manager, your stakeholder, or your direct report — using this as the model.

Use this as a working reference for how to communicate with me (the instructor), Sean McCaman, and how to create higher quality deliverables.

If You Only Do Five Things

Whether you use an LLM or not, the checklist below is the core. If you are using an LLM, give it the assignment prompt, rubric, and the relevant parts of this document, and ask it to review your draft against the checklist.

For every deliverable:

  1. Lead with the point. State the conclusion before the supporting detail.1
  2. Make every number traceable. Source, calculation, or assumption should be findable.
  3. Pick consistent terms and stick to them. Pick one term per concept and use it everywhere.
  4. Match the format to the deliverable type. A chart, slide, report, model, and code project are not the same. Use the right checklist below.
  5. Proofread before submitting. Typos and inconsistency have a meaningful impact on reader trust.6

What This Is

This document explains how I tend to read, review, and give feedback on student work. It is not a replacement for the course syllabus (Canvas login required), assignment instructions, rubric, or course policies. Use those first. Use this as a practical guide for making your work clearer, easier to evaluate, and more useful to your audience.

The deeper skill this guide is practicing is constraint definition: being clear about requirements and expectations. Analytics used to be "how well do you know Tableau or Excel." It is increasingly "who can write the best prompt with the best context." The tool fluency still matters; the leverage now comes from the constraint-definition layer above the tool. Clearer constraints produce better work, whether the producer is an LLM, a teammate, or yourself.

These are my preferences as one instructor working in financial technology. My perspective skews toward fintech roles, data engineering, and finance/accounting automation; other industries, classes, and reviewers will weight things differently. Treat the standards in this document as practical guidance for working with me, not as universal rules.

Note on professional context: most production AI use in accounting now happens on enterprise-controlled platforms — in-house LLMs at firms like KPMG and Deloitte, sanctioned agentic workflows, Microsoft Copilot deployments. The principles in this document are tool-agnostic; the specific platform you use professionally will vary by employer, but the skill of defining good constraints transfers.

If you are using an LLM, reference this document — or just the relevant excerpts — as a standing guide. Give the LLM the current assignment instructions, rubric, and the specific deliverable you want reviewed. Good context is targeted context.

Who I Am

What I Value

Clarity is kindness. When you make your work easy to follow, you are respecting the time, attention, and decision-making responsibility of your reader.

What Good Work Looks Like to Me

What Usually Draws Feedback

I am usually reacting to something that makes the work harder to trust or harder to understand. These are fixable communication or analytical issues.

How to Get Moving Fast

How to Build Your Own Managing-Up Document

The same pattern adapts to any working relationship: your audit senior, your manager, a stakeholder you support, a direct report you manage, a peer team you collaborate with. The relationships differ; the document structure does not. You define who they are, what they value, what you can offer, and how you want to be evaluated.

Good source material includes the job description, performance criteria, prior feedback you have received, formal style guides of your firm or industry (e.g., Big Four work-paper conventions, GAAP presentation rules, internal style guides), and any public artifacts (LinkedIn profiles, internal bios, published deliverables) that show what your audience prioritizes.

Use a prompt like this:

Read the attached job description, performance criteria, prior feedback (if any), and any public artifacts about my [audit senior / manager / stakeholder / direct report]. Build a concise managing-up document for this relationship that explains:

1. What this person appears to value in the work I produce.
2. What strong work likely looks like to them.
3. What mistakes are likely to draw critical feedback.
4. What checklist I should run on my own deliverables before sending.
5. What deliverable-specific checks apply to writing, slides, charts, spreadsheets, models, code, or other outputs I will produce.

Write it in a calm, professional tone. Do not invent private information. Separate evidence from interpretation.

Once you have a draft, refine it against the real interactions you have. The document is most useful when it captures what you have actually observed, not what you initially guessed. Treat it as a living artifact — review and update it every few months.

This same pattern can also adapt to other classes (instructor as audience, syllabus as job description, rubric as performance criteria). The structure carries.

Skill-Ready Checklist

"Skill-ready" means the checklist is structured so a person or an LLM can apply it directly without further translation. Use it before submitting a chart, table, document, slide deck, spreadsheet, model, notebook, or code-based deliverable. A reviewer or LLM can mark each relevant item as [OK], [WARN], or [FIX], then give brief evidence and a suggested improvement.

Applies to Every Deliverable

Charts and Tables

Slides and Presentations

Written Analysis and Reports

Spreadsheets, Models, and Data Workbooks

Code-Based Deliverables

Final Principle

AI is a powerful collaborator, and learning to use it well is part of this course. There is no expectation that you start out expert. The students who get the most out of these tools are the ones who stay curious, keep their own thinking in the lead, and check the work before trusting it.

An LLM can help you reach a finished product faster, but the final submission still carries your name. Whether you send an email, post a Slack message, submit in Canvas, or deliver code, you are responsible for the quality, accuracy, and judgment behind it. Wherever you are with these tools, you are in the right place.

Clear thinking shows up as clear output. If your analysis is solid but the chart, table, document, slide deck, spreadsheet, or code hides it, much of the value is lost on the audience.

Fair Use

You are welcome to copy, adapt, and reuse this document for your own classes, teams, or projects. Please include attribution when you do.

Suggested attribution:

Adapted from "Example Managing-Up Document" by Sean McCaman, University of Utah.

Definitions

A few terms used above, in case any of them are new. They are listed alphabetically. Click any linked term in the body of this document to jump back here.

References

Sources backing claims made in this document.

Last Updated

This document was last updated 5/11/2026 and is intended for use with the University of Utah course ACCTG 5150 - 090.

  1. On readers: Weinreich et al. (2008) found people consume ~20% of words on a typical page and abandon nearly half of new pages within 12 seconds. On LLMs: Liu et al. (2024) document a U-shaped attention curve where models reliably attend to information at the start and end of context but miss content buried in the middle. ↩↩

  2. The heuristics-and-biases literature was launched by Tversky & Kahneman (1974). For systematic coverage of all six biases listed, see Gilovich, Griffin, & Kahneman (2002). ↩

  3. Saket, Endert, & Demiralp (2019) found chart effectiveness varies significantly across analytical tasks, supporting the principle of matching chart type to the question being answered. ↩

  4. Kosslyn et al. (2012) found that arbitrary typographic mixing and unsignalled visual changes are widespread slide-design failures, violating principles of discriminability and informative change. ↩

  5. Panko (1998) summarizes field audits showing approximately 88% of operational spreadsheets contain errors. Embedded constants inside formulas are a commonly flagged finding in EUC audit guidance. ↩

  6. Witchel et al. (2020) found spelling errors produced an 8.86-point trustworthiness penalty (out of 100) in randomized reading experiments; Cox, Cox, & Cox (2017) replicated this finding for online reviewer credibility. ↩