| Course | BU 443 Marketing Analytics |
|---|---|
| Unit | Unit 2 |
| Paper type | Data collection and tracking audit |
| Length | About 850 words, 6 pages |
| Format | APA 7 student paper |
| School | Herzing University |
| Program | BS in Marketing |
| Updated | September 2026 |
Free sample paper for BU 443 Unit 2
Five Spellings of One Ad Platform and a Purchase That Counts Twice on Reload: A Data Collection and Tracking Audit for a Composite Online Houseplant Retailer
Student Name
Business, Herzing University
BU 443: Marketing Analytics
Instructor Name
January 28, 2026
Five Spellings of One Ad Platform and a Purchase That Counts Twice on Reload: A Data Collection and Tracking Audit for a Composite Online Houseplant Retailer
What This Audit Asks
Greenline Plant Co., the composite Portland online houseplant retailer used in this course, defined its measures in the previous unit. This audit asks whether the systems producing those measures record what the business thinks they record. It inventories the events the website captures, tests the campaign labels that tell analytics where visitors came from, examines how visitor consent shapes what is recorded, and ends with a register of what this dataset cannot see at all. The audit covers February 2026 and was carried out with the analytics tool's debugging view, a sample of real orders matched against recorded events, and an export of every campaign label received.
Kohavi et al. (2020) recommend checking data quality before trusting any analysis, since errors in how events are logged can produce effects larger than the ones analysts are looking for. Every figure below belongs to the composite case.
The Events Being Captured
Nine events are configured. Each was triggered by hand in a test session and checked against what the analytics tool recorded, and the purchase event was checked against 200 real orders.
| Event | Where it should fire | What the test found | Status |
|---|---|---|---|
| page_view | Every page | Fires correctly | Working |
| view_item | Product pages | Fires correctly with plant name and price | Working |
| add_to_cart | Every add-to-cart button | Missing on the quick-add buttons on collection pages | Faulty |
| begin_checkout | Start of checkout | Fires correctly | Working |
| add_shipping_info | Shipping step | Fires, but without the chosen shipping speed | Incomplete |
| add_payment_info | Payment step | Fires correctly | Working |
| purchase | Order confirmation page | Fires again if the confirmation page is reloaded; 96 of 3,410 February purchase events repeated an order number | Faulty |
| sign_up | Email sign-up form | Fires correctly | Working |
| plant_quiz_complete | End of the plant-finder quiz | Fires on the last question, even if the visitor abandons the results page | Misnamed |
Two faults change reported figures. The duplicate purchase events inflate recorded purchases by about 2.8 percent, and they cluster among mobile shoppers who reload the confirmation page to check their order. The missing add-to-cart event on quick-add buttons understates carts; the store's own records show that about 31 percent of carts in February started from a collection page, so the analytics tool's cart-to-purchase rate looks better than it is. Both are fixable in the tag configuration: the purchase event should be sent only once per order number, and the quick-add buttons need the same trigger as product pages.
The Campaign Labels
Analytics learns where a visitor came from through labels attached to links, most importantly the source, the medium and the campaign name. An export of every label value received in February shows how inconsistent they have become.
| Label | Values found for the same thing | Sessions affected | Effect in reports |
|---|---|---|---|
| Source for the social ad platform | facebook, fb, Facebook, ig, meta | 11,840 | Split across five rows; some grouped as organic social |
| Medium for paid ads | cpc, paid, paid_social, social | 14,210 | Paid social partly counted as unpaid social |
| Campaign on email links | Missing on 23% of links | 3,960 | Email traffic reported without campaign detail |
| Plant blogger links | No labels at all | 2,730 | Counted as generic referral traffic |
The consequence is that the paid social channel, which the owners are deciding whether to expand, is understated in analytics by whatever share of its traffic falls under the unpaid medium values. The fix is a written labeling standard, lowercase throughout, with one permitted value for each platform and medium, a link builder that the marketing assistant and the blogger partners must use, and a monthly export check.
Consent and What It Removes
Since 2025 Greenline has shown every visitor a consent banner, and visitors who decline analytics cookies are not recorded by the analytics tool at all. In February, about 17 percent declined. The store still records their orders, which is one reason the two systems disagree about sessions and conversion. Consent belongs in a tracking audit because it decides who is in the data. Nouwens et al. (2020) showed that the design of consent pop-ups strongly influences how many people agree to tracking, which means a redesign of Greenline's banner could change its analytics figures without any change in customer behavior. The audit records the consent rate each month so that shifts in it are not mistaken for shifts in the business.
The Order of Fixes
Not every problem can be fixed at once, so the audit ranks them by how much they distort decisions the owners are about to make. The campaign labels come first, because the owners plan to decide in April whether to raise paid social spending, and that decision rests on a channel comparison the labels currently spoil. The duplicate purchase event comes second, since it inflates every conversion figure slightly but evenly across channels. The missing cart event comes third; it affects the analysis of where shoppers leave checkout, which is planned for later in the year. The quiz event is renamed last. After each fix, the date is recorded in an annotation inside the analytics tool, so that anyone reading a trend line later can see where a jump reflects a repair rather than a change in customer behavior.
What This Dataset Cannot See
The audit ends with a register of blind spots, so that later analysis does not treat the data as complete. Visitors who decline consent, about one in six, are invisible to analytics. Journeys that cross devices appear as separate users. Orders taken by phone, about 40 a month, and sales at the greenhouse's weekend pop-up in Portland never touch the website. People who saw an ad but did not click leave no trace in analytics. Plants that die after the 30-day guarantee and are never reported show up nowhere, although they shape whether a customer returns. And the recipients of gift orders, a growing share in spring, are unknown to the data entirely. Farris et al. (2010) note that the most dangerous gap in marketing data is the one nobody has written down, and this register exists so that each of these is named before the channel and cohort analyses begin. Greenline and all figures here are composites for this course.
References
Farris, P. W., Bendle, N. T., Pfeifer, P. E., & Reibstein, D. J. (2010). Marketing metrics: The definitive guide to measuring marketing performance (2nd ed.). Pearson Education.
Kohavi, R., Tang, D., & Xu, Y. (2020). Trustworthy online controlled experiments: A practical guide to A/B testing. Cambridge University Press.
Nouwens, M., Liccardi, I., Veale, M., Karger, D., & Kagal, L. (2020). Dark patterns after the GDPR: Scraping consent pop-ups and demonstrating their influence. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems (pp. 1-13). Association for Computing Machinery. https://doi.org/10.1145/3313831.3376321
The Herzing BU 443 Unit 2 instructions
In Unit 2, BU 443 asks students to audit how marketing data are collected. Common requirements include an inventory of tracked events, a check of whether each fires correctly, a review of campaign labeling for consistency, attention to consent and privacy settings that decide who is recorded, and a list of what the data cannot capture. Many instructors expect students to test rather than assume, for example by comparing recorded events with actual orders. Consent is increasingly treated as part of data quality rather than a separate legal topic. Composite systems are acceptable when described clearly. An inventory table, a label table and a gap register, explained in APA 7, are typical deliverables.
How the BU 443 Unit 2 example is put together
The audit opens by stating its period and methods: a debugging view, 200 orders matched against events, and a full label export. An event table records nine events, where each should fire and what testing found, followed by a paragraph sizing the two faults that change reported figures. A label table shows four kinds of inconsistency, the sessions affected and the reporting effect, then proposes a labeling standard. A consent section explains how declined tracking shapes the data and why banner design matters. A section ranks fixes by the decisions they affect and records fix dates in the analytics tool. The audit closes with a register of blind spots.
BU 443 Unit 2 rubric: what full points look like
Tracking audit rubrics usually reward a complete event inventory, evidence of testing, analysis of labeling consistency, treatment of consent, a clear gap register and practical fixes, plus writing and APA. The inventory earns credit when every configured event is listed with its intended trigger. Testing scores highest when faults are found by comparing records with reality and sized by their effect on reported numbers. Labeling analysis rewards showing actual inconsistent values and their consequences. Consent earns points when treated as a factor in who is measured. The gap register is judged on completeness and specificity. Instructors also credit fixes ranked by business impact, not by technical convenience.
Common BU 443 Unit 2 mistakes, and how to avoid them
Tracking audits often lose points by listing configured events without testing whether they fire. Another common problem is noting that labels are inconsistent without showing the actual values or how many sessions are affected. Students frequently leave consent out, treating it as a legal matter, even though it determines who appears in the data. Some gap registers say only that data may be incomplete, which helps nobody. Others propose fixes without priorities or dates. Many audits forget to record when fixes happen, so later trend lines show jumps nobody can explain. Finally, auditing an employer's live systems for a class paper can expose customer data; a composite case avoids that. Keeping a dated log of every tracking change also lets later analysts explain sudden jumps without guessing at causes.
Get a BU 443 Unit 2 example written to your instructions
Forward the Unit 2 instructions with your BU 443 rubric and the case or dataset the assignment supplies. We write a custom example that inventories what fires and from where, examines the naming as a scheme, puts consent inside collection and closes on a register of unanswerable questions. First custom sample free, returned in 24 to 48 hours.
More BU 443 papers and marketing samples
- BU 443 Unit 1: Metrics definition exercise
- BU 443 Unit 3: Channel performance analysis
- BU 443 Unit 4: Customer and cohort analysis
- BU 443 Unit 5: Campaign evaluation
- BU 443 Unit 6: Test design and reading
- BU 443 Unit 7: Dashboard or visualization
- BU 443 Unit 8: Analytics recommendation
- BU 380 Unit 2: Mission and values stress test
- BU 421 Unit 2: Process flow documentation
- BU 423 Unit 2: Customer problem validation record
- BU 427 Unit 2: Change model comparison
BU 443 Unit 2 questions, answered
What do I audit if the case gives me no system to look at?
Whatever the case does describe, plus a documented set of assumptions. Most assignments supply a scenario with reports, sources and a few known problems, which is enough to inventory. Where a detail goes unstated, write the assumption into the audit and label it, since an audit resting on stated assumptions stays checkable and one resting on silent ones does not.
Does consent really belong in a tracking audit?
It decides what data exists, so yes. People who decline are not a rounding difference; they are missing from the counts everything downstream is built on, and a report never mentioning them presents a partial dataset as a complete one. Keep the treatment factual, describe the behavior your case specifies, and check any rule you cite against current sources rather than summarizing from memory.
How detailed should the gap register be?
Detailed enough that somebody could decide whether to act on each line. A gap stated as a question the data cannot answer, with a sentence on what would have to be collected instead, is usable. A list of field names is not. Keep it honest rather than exhaustive, since three gaps that genuinely limit the analysis matter more than a dozen theoretical ones.
Where can I find a free BU 443 Unit 2 sample paper?
A full BU 443 Unit 2 data collection and tracking audit is free above, including the event tests, the label table, the gap register and margin notes. For your own case or systems, forward the prompt and a custom audit is prepared free, typically within 48 hours.
Does consent belong in a BU 443 Unit 2 audit?
Yes. Consent settings decide who is recorded at all. In the sample, about one visitor in six declines analytics tracking, which is why analytics and the store disagree about sessions and conversion.