D613 Decision Intelligence, catalog number DTAN 6226, is the three CU course in the WGU Master of Science, Data Analytics that teaches the core principles of optimizing decision making by balancing technology, processes and people, including machine augmentation of human judgment. It is the course where analytics stops being about producing an answer and starts being about designing the arrangement that turns answers into action.
The unit of analysis is a decision
Students who have spent a degree analysing datasets often struggle with the first move in DTAN 6226, because the object of study changes. The thing being examined is not the data and not the model. It is a decision: who makes it, how often, on what information, under what time pressure, with what consequences for being wrong in each direction.
Getting that framing right is most of the course. A well specified decision has a decider, a trigger, a set of available options, an information basis, a timeframe and an asymmetry between the costs of the two kinds of error. A submission that describes an analytics capability without ever specifying the decision it serves has not started the assignment, however sophisticated the technology discussion is.
The second scored theme is the allocation of work between people and machines. Machine augmentation is in the catalog description for a reason, and the interesting question is never whether to automate but which part to automate. Machines are better at consistency, volume and recall; people are better at context, exceptions and accountability. A defensible design says which part goes where and why, and it says what happens at the boundary when the machine is uncertain.
Turning scored aspects into a section plan
Scoring detail sits in your Course of Study rather than the public catalog. Count the aspects and treat them as your headings. Each is judged on its own against a three point scale and each needs a 2 to pass, so a strong technology section cannot compensate for a thin account of the people involved.
This course covers three domains at once, and the balance between them is itself scored. If your draft is two thirds technology, it is out of balance regardless of quality, because the course is explicitly about the arrangement of all three.
The word budget, worked. Assume six scored aspects and about 1,900 words of narrative. Reserve 150 for an opening naming the decision and the decider, and 130 for a close. That leaves roughly 1,620, near 270 per aspect. Then move 60 words from each of two technology heavy aspects into the aspects covering people and governance, taking those to about 330 each. Almost every submission arrives over weighted towards tooling, and correcting that on paper before drafting is the cheapest structural fix available.
A structure that fits a decision intelligence task
Where your directions specify a structure, follow it. Where they do not, this ordering carries the balance the course expects.
| Section | What belongs in it | How it gets read |
|---|---|---|
| Decision specification | The decision, the decider, the trigger, the options and the frequency | Everything else is judged against this; vagueness here is fatal |
| Current decision process | How it is made today, on what information, and how well that works | Baseline for any claimed improvement |
| Error asymmetry | What each kind of mistake costs, and to whom | The most under written and most differentiating section |
| Information and analytics | What evidence is available, what it can and cannot support | Scored on fit to the decision rather than on sophistication |
| Human and machine allocation | Which part is automated, which is advisory, which stays human, and why | The centre of the course; needs explicit boundaries |
| Governance | Override rules, audit trail, accountability, monitoring for drift | Expected at graduate level and often absent |
| Implementation and adoption | How the new arrangement reaches the people who must use it | Where feasibility is judged |
Write the decision specification first and get it to one paragraph that a manager would recognise. Everything downstream inherits its clarity or its fog.
Evidence craft in a course about judgment
This course draws on behavioural research, organisational literature and your own observation. All three need handling with care.
- Cite behavioural claims properly. Statements about bias, overconfidence or anchoring are research findings, not general knowledge, and unsourced they read as folklore.
- Quantify the error asymmetry even roughly. A false positive costing an hour of staff time against a false negative costing a customer is a defensible framing.
- Describe the current decision process from observation where possible, and say when you are relying on report instead.
- Be specific about what the analytics can support. Correlation supporting a targeting decision is fine; the same evidence supporting a causal claim is not.
- Name the accountability holder. Automated decisions still have an owner, and saying who it is answers a governance aspect directly.
- Use APA for external sources and keep quotation short.
The strongest addition to a decision intelligence submission is a stated failure mode for the augmented arrangement itself: what happens when the model is confidently wrong and the human has stopped checking. Automation complacency is a documented phenomenon and naming it shows the design was thought through rather than assumed.
What separates Competent from a submission sent back
Aspects score independently, and returns here cluster on the human and governance sections rather than the analytical ones.
- The decision is specified precisely enough that someone else could make it.
- The cost of each kind of error is stated, at least approximately.
- The boundary between machine and human work is explicit, including what happens at low confidence.
- Override and escalation rules exist and name a role.
- Adoption is addressed as a real obstacle rather than assumed.
Performance assessment work at WGU can be revised and resubmitted with no grade penalty, so a return costs time rather than standing. Terms run six months at a flat rate, which makes queue time the expensive part of any rework.
Six mistakes that cost time in D613
- Describing a system instead of a decision. A dashboard is not a decision, and a course about decisions notices.
- Treating automation as the goal. The assessed skill is allocation between human and machine, which sometimes means automating less.
- Ignoring error asymmetry. Optimising a symmetric metric for an asymmetric decision produces a defensible model and a bad outcome.
- No override path. A design with no way for a human to intervene fails governance aspects and would fail in practice.
- Unsourced behavioural claims. Bias vocabulary used loosely is one of the easiest weaknesses for an evaluator to spot.
- Assuming adoption. People resist decision tools that make their judgment auditable, and a plan that ignores this is not a plan.
Drawing the line between human and machine
The allocation section is where D613 submissions are won, and there is a structured way to write it that reliably satisfies the aspects.
Break the decision into its stages: gathering information, generating options, evaluating options, choosing, executing and reviewing. Then assign each stage explicitly. Machines usually dominate gathering and evaluation at volume. Option generation is often mixed, because a model can only propose from what it has seen. Choosing may be automated for routine cases and reserved for humans in exceptional ones. Execution is frequently automated, and review is almost always human and almost always neglected.
For each stage you automate, state three things. The confidence condition under which the machine acts alone. The escalation path when that condition is not met. And the sampling regime by which a human checks a portion of automated decisions even when nothing appeared to go wrong, because an automated decision process with no sampling degrades silently.
Then write the accountability sentence: when this arrangement produces a bad outcome, who answers for it. The honest answer is always a person, and stating it directly resolves a governance aspect that many submissions leave implied. Together these take half a page and they demonstrate the balance of technology, process and people that the course is named for.
How support works on this course
Send the rubric from your Course of Study and the task directions. What comes back is aspect mapped: a decision specification a manager would recognise, an error asymmetry section with figures, an allocation section with confidence thresholds and escalation paths, governance with named roles, and an adoption plan. Plus a walkthrough so you can defend the design as your own reasoning.
D613 sits close to the process engineering course and the two reinforce each other. Terms are six months at a flat rate, so pairing related courses inside one term is the straightforward way to lower the cost per course.
Questions students ask about D613
Is D613 the same course as DTAN 6226?
Is D613 a technical course?
Can you complete the decision analysis for me?
Where D613 sits in WGU's programs
The July 2026 catalog places this code in 1 current WGU program. Open a program page for the complete standard path and term positions. The live Degree Plan remains authoritative after transfer credit, substitutions, and mentor planning.
The assessments, one by one
The public catalog does not publish this course's PA/OA identity or task count. WGU Tutors publishes at most one PA manual per course and only from a WGU-controlled public rubric. Until that source exists, PA help begins from the student's real Course of Study and OA support remains preparation only.