| Course | MT362 Artificial Intelligence Applications for the Marketing Professional |
|---|---|
| Unit | Unit 3 |
| Paper type | Customer data inventory |
| Length | About 850 words, 6 pages double-spaced |
| Format | APA 7 student paper |
| School | Purdue University Global |
| Program | BS in Business Administration |
| Updated | October 2026 |
Free sample paper for MT362 Unit 3
Told, Observed, Bought: An MT362 Unit 3 Customer Data Inventory for a Pet Chain's Churn Model
[Student Name]
Purdue University Global
MT362: Artificial Intelligence Applications for the Marketing Professional
Unit 3 Customer Data Inventory
[Instructor Name]
[Date]
The chain, its records and its privacy policy versions are composite; the sources cited are real.
How the Table Is Organized
Collection method organizes the table, because the same kind of field can be fair to use or not depending on how it arrived. Data customers told the chain directly comes first, then records the chain generated by observing transactions, then data it bought or inferred. Each row has eight columns: the source, the fields it holds, the collection method, the privacy notice in force at collection, the date range, the internal owner, known quality problems and a verdict. The verdict column uses three values only. Use means the model may draw on the source. Hold means a named question must be answered first. Exclude means the source stays out.
Told by the Customer
Pet profiles, entered in the app since 2019: pet name, species, breed, birth date and weight. Notice: the 2019 privacy policy, which said profiles would be used to personalize the delivery program. Owner: digital product. Quality: weights are rarely updated after a pet's first year, so many adult dogs are recorded at puppy weight. Verdict: use, with weight treated as unreliable.
Exit-survey answers, collected at cancellation since 2022. Notice: the 2021 policy and the survey's own statement that answers would help improve the program. Owner: lifecycle marketing. Quality: about one third of cancelers skip it. Verdict: use.
Birth years, collected at Rewards sign-up until 2021, when the field was dropped. About 38 percent of current members gave one. Notice: the 2019 policy, which mentioned birthday offers. Owner: loyalty. Verdict: exclude from the model, hold for audit use only, since the bias review later in the term will need it.
Contact preferences and text consent, recorded since 2016. Notice: each consent screen as shown. Owner: lifecycle marketing. Verdict: use for eligibility only, never as a predictor.
Observed in Transactions
Point-of-sale transactions linked to Fernbrook Rewards since 2016. Notice: the policy versions of 2016, 2019 and 2023, each of which said purchase history would be used to run and improve programs. Owner: finance and analytics. Verdict: use.
Repeat Delivery order history since 2018, and, in a separate row, schedule edits since 2020. Notice: the program terms, which said order and schedule data would be used to manage deliveries. Owner: e-commerce. Verdict: use.
Returns and refunds since 2018. Owner: customer service. Verdict: use.
Email engagement since 2017. Quality: open rates have been unreliable since September 2021, when a major mail app began preloading images for its users, so opens are recorded that never happened. Verdict: use clicks only.
Customer service chat transcripts since 2021. Notice: collected to resolve complaints. Verdict: hold, pending the question of whether using complaint text to predict cancellation fits the purpose customers were given.
App location data since 2020. Notice: collected to find the nearest store. Verdict: hold, pending the same purpose question.
Bought or Held Elsewhere
A household data file bought from a broker in 2023, estimating household income and homeownership. Notice: none from the chain; the broker's terms cannot be located, and no one could say which campaign used the file. Verdict: hold, pending the terms.
Vaccine clinic records. These are held by the veterinary partner that runs the in-store clinics, under a separate agreement, not by the chain. Verdict: outside the inventory's reach. Listing them anyway prevents the common assumption that a retailer can draw on everything that happens inside its stores.
Model scores themselves will become a fourteenth source once the pilot runs, and the inventory reserves a row for them, owned by analytics, with the notice question answered in Unit 4.
The notice in force when data was collected, not the data's usefulness, decides whether a new purpose is fair.
Why the Notice Column Matters
Nissenbaum (2010) argues that privacy holds when information moves according to the expectations attached to the setting where a person first gave it, and breached when it moves to a context those norms do not cover. A customer who typed a pet's weight to get the right bag size shared it in a delivery context; using it to set delivery timing stays within that context. Using chat complaints to predict cancellation may not. Martin and Murphy (2017) review evidence that consumers react most strongly when data is used in ways they did not expect, which is why the inventory records the notice beside every source.
Signal the Model Can Do Without
Location data, chat transcripts and the broker file are marked hold, and the notes argue that the model need not wait for them. Order history, schedule edits, returns and pet profiles probably carry most of the signal about whether a subscriber is about to leave. Acquisti et al. (2015) describe how uncertain and context-dependent people's privacy preferences are, which argues for starting with the data customers would most readily expect a retailer to use.
References
Acquisti, A., Brandimarte, L., & Loewenstein, G. (2015). Privacy and human behavior in the age of information. Science, 347(6221), 509-514. https://doi.org/10.1126/science.aaa1465
Martin, K. D., & Murphy, P. E. (2017). The role of data privacy in marketing. Journal of the Academy of Marketing Science, 45(2), 135-155. https://doi.org/10.1007/s11747-016-0495-4
Nissenbaum, H. (2010). Privacy in context: Technology, policy, and the integrity of social life. Stanford University Press.
Reading the MT 362 Unit 3 instructions
Unit 3 asks students to inventory the customer data a proposed AI use might draw on, before deciding what the model may use. Each source should be described by its fields, how it was collected, the notice or terms in force at the time, the date range, who owns it internally and its known quality problems. Instructors expect a verdict for each source, with reasons. Data the business does not actually hold, but that people might assume it does, deserves a row marked as outside reach. Unknowns should be recorded honestly as questions to resolve. Notes on the hardest rows usually follow the table, and many sections ask which data the model could do without.
How the MT 362 Unit 3 example is put together
An opening section explains the eight columns and the three verdicts. Told data follows, from pet profiles with stale weights to birth years held for audit only. Observed data covers transactions, Repeat Delivery orders and schedule edits as separate rows, returns, email clicks, chat transcripts and app location, the last two held pending a purpose question. Bought and partner-held data closes the table, including the vaccine records the chain does not hold. A reserved row anticipates model scores. A highlighted line states that the notice at collection decides whether a new use is fair, and a section applies Nissenbaum and Martin and Murphy. The notes end with signal the model can do without.
Reading the MT362 Unit 3 grading rubric
Data inventories are typically graded on completeness, accuracy about collection and notice, attention to data quality and the soundness of each verdict. Organizing rows by how data arrived shows the student understands why the same field can be fair or unfair to use. Recording the notice in force at collection is the feature graders look for most, since the privacy review depends on it. Quality problems stated plainly show care. Rubrics reward verdicts tied to named questions rather than vague caution. Including data held by partners prevents a common assumption. Identifying what the model can do without shows business judgment. Privacy research cited in APA 7 supports the reasoning in the notes.
MT362 Unit 3 help from the desk
Inventories often list databases by system name, such as the CRM or the warehouse, without saying how the data inside them was collected. That hides the distinction the privacy review needs. Another pattern marks every source usable because it is available. Quality problems are rarely mentioned, though stale or inflated fields can mislead a model. Some inventories omit data held by partners, inviting later confusion about what the business can use. Before you finish, read each row and ask what the customer was told when the data was collected. If you cannot answer, the verdict should be hold, with that question written beside it.
Get an MT362 Unit 3 example written to your instructions
Whatever business your Unit 3 prompt assigns, list the kinds of customer records it seems to hold, even roughly, and add the prompt and rubric. A custom inventory sorted by how each record was gathered, with a verdict for every row, comes back in 24-48h, and the first one is on us. The paper above is an original model document written by our desk, not a submitted student paper and not an official Purdue University Global document.
More MT362 papers and BS in Business Administration samples
- MT362 Unit 1: AI Retention Offers Post
- MT362 Unit 2: AI Use Case Assessment
- MT362 Unit 4: Privacy and Consent Review
- MT362 Unit 5: Personalization Strategy Memo
- MT362 Unit 6: Seminar Reflection
- MT362 Unit 7: Bias Review
- MT362 Unit 8: Pilot Test Design
- MT362 Unit 9: Vendor Comparison Matrix
- MT362 Unit 10: AI Adoption Proposal
- MT361 Unit 3: Listing Agreement Clause Review
- MT359 Unit 3: Get, Who, To, By Creative Brief
- MT358 Unit 3: Six-Week Social Content Plan
- MT357 Unit 3: Digital Channel Selection Rationale
MT362 Unit 3 questions, answered
How many data sources should the inventory include?
Enough to cover every place the proposed use might draw from, which for a mid-sized retailer is often ten to fifteen. Graders care less about the count than about completeness for the use case and the detail in each row. A short inventory that records collection method, notice and quality for each source usually outscores a long list of system names.
What if I do not know how the data was collected?
Say so in the row and mark it as a question to resolve. An honest unknown is useful, since it tells the later privacy review exactly where to look. Where the case gives hints, such as a loyalty sign-up form or a purchased list, describe the likely method and label it as inferred rather than confirmed.
Should the inventory include data the business does not own?
Include it when people might assume the business holds it, and mark who actually does. Partner records, platform data the business can only view, and vendor-held files are common examples. Listing them with a clear owner prevents a later section of the proposal from relying on information the business has no right or means to use.
Where can I find a free MT362 Unit 3 sample paper?
The complete MT362 Unit 3 customer data inventory is on this page, sorting a pet chain's thirteen data sources by how they were collected.
Why are birth years held for audit in the MT362 Unit 3 sample?
They stay out of the model, but the later bias review needs them to test whether the model treats older subscribers differently.