| Course | BU 443 Marketing Analytics |
|---|---|
| Unit | Unit 4 |
| Paper type | Customer and cohort analysis |
| Length | About 940 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 4
Spring Buyers Are Many and Holiday Buyers Rarely Return: A Customer and Cohort Analysis of a Composite Online Houseplant Retailer's 2025 First-Time Customers
Student Name
Business, Herzing University
BU 443: Marketing Analytics
Instructor Name
April 8, 2026
Spring Buyers Are Many and Holiday Buyers Rarely Return: A Customer and Cohort Analysis of a Composite Online Houseplant Retailer's 2025 First-Time Customers
Why Group Customers by When They Arrived
So far this course has compared Greenline Plant Co.'s channels; the Portland plant seller is a composite. This analysis divides customers instead. It groups every first-time buyer by the quarter of their first order in 2025, and a partial group for January and February 2026, then follows each group forward across the same windows: whether each customer placed a second order within 30, 90, 180 and 365 days of the first. Grouping by arrival time lets the owners see whether customers found at different times of year behave differently, which a single repeat purchase rate for the whole year would hide.
Fader (2020) argues that customers differ enormously in future value, and that a business learns more from how groups of customers acquired at different times behave than from averages across all of them. The data come from the store platform's order records, with customers identified by email address, as of March 31, 2026, and are part of the invented case.
The Cohorts
Each cell shows the share of the cohort that placed a second order within the window, with the number of customers behind it. A window marked open has not yet closed for every customer in the cohort, so no rate is reported.
| First order in | Customers | Within 30 days | Within 90 days | Within 180 days | Within 365 days |
|---|---|---|---|---|---|
| January to March 2025 | 5,120 | 4.1% (210) | 11.3% (579) | 17.6% (901) | 24.2% (1,239) |
| April to June 2025 | 8,940 | 3.6% (322) | 9.8% (876) | 15.1% (1,350) | Open |
| July to September 2025 | 4,380 | 4.4% (193) | 12.1% (530) | 19.0% (832) | Open |
| October to December 2025 | 6,610 | 2.9% (192) | 7.4% (489) | Open | Open |
| January and February 2026 | 3,020 | 3.9% (118) | Open | Open | Open |
What the Cohorts Show
Three patterns stand out. The spring cohort, April to June, was the largest by far, with 8,940 new customers, but it repeated less than the winter and summer cohorts at every window that has closed: 9.8 percent within 90 days, against 11.3 and 12.1 percent. The holiday cohort, October to December, repeated least of all, 7.4 percent within 90 days. And the summer cohort, the smallest of 2025, repeated best at every window.
Size and loyalty therefore run in opposite directions. The quarters that bring in the most customers bring in customers less likely to come back. The owners had judged spring as their best season because it brings the most first orders; measured by repeat behavior, it is not.
Possible Explanations, and What the Data Can and Cannot Say
Two explanations fit the holiday cohort. Many holiday orders are gifts, and the buyer is not the person who cares for the plant; the store's gift-message field was filled on 41 percent of holiday first orders against 9 percent in summer. And winter shipping damages more plants, since cold stress in transit is a known risk; replacement credits were issued on 6.8 percent of holiday first orders against 2.1 percent in summer. A buyer whose first plant arrived damaged is less likely to order again, although the data cannot separate that effect from the gift effect.
The spring cohort's weaker repeat rate has a different candidate explanation. A larger share of spring first orders came through paid social, 31 percent against 18 percent in summer, and the channel analysis found that paid social tends to introduce new customers. People reached by an ad in their feed may be more casual buyers. But spring buyers also buy seasonally, for balconies and new homes, so they may simply not need a second plant until the next spring. The 365-day window for the spring cohort closes in June 2026, and only then can the owners see whether spring buyers return a year later. Until then, the lower rate is a signal to watch, not a verdict.
Gupta et al. (2004) show that a customer's value depends heavily on retention, so a cohort that repeats at 9.8 percent rather than 12.1 percent is worth noticeably less per customer even if its first order is the same size.
What the Owners Could Do With This
The analysis suggests actions to test rather than conclusions to act on. For the holiday cohort, two responses follow from the two explanations. If gifts are the reason, a card in gift orders inviting the recipient to join the email list, with care tips for the plant they received, would try to turn recipients into customers. If shipping damage is the reason, extending the heat-pack policy to all orders shipped to cold states from November through February would address it directly, at a cost the owners can price. Running one response in some regions and not others would show which explanation matters more. For the spring cohort, the owners should wait for the 365-day window before changing acquisition spending, but they can begin measuring now whether spring customers who joined the email list repeat more often than those who did not. Each of these produces a comparison the next units can read.
Reading the Newest Cohort Honestly
The January and February 2026 cohort has only one closed window, 30 days, where it repeated at 3.9 percent, between the spring and summer figures. It would be easy to compare its early numbers with the full-year figures of older cohorts and conclude that 2026 customers are worse, but every young cohort looks weak before its later windows close. This analysis reports only its closed window and says so.
Two further limits apply. Customers are identified by email, so a person who uses a second email for a later order appears as a lost customer. And a group of a few thousand customers can move a percentage point by chance, so differences smaller than about one point between cohorts should not be treated as meaningful. Farris et al. (2010) recommend always reporting the base behind a rate for this reason. Greenline and every number here were invented for the course.
References
Fader, P. (2020). Customer centricity: Focus on the right customers for strategic advantage (2nd ed.). Wharton School Press.
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.
Gupta, S., Lehmann, D. R., & Stuart, J. A. (2004). Valuing customers. Journal of Marketing Research, 41(1), 7-18. https://doi.org/10.1509/jmkr.41.1.7.25084
What the BU 443 Unit 4 instructions ask for
For the fourth unit, BU 443 asks students to analyze customers in cohorts. Typical requirements include defining cohorts, usually by time of first purchase, following each cohort across the same number of periods, reporting rates with the counts behind them, interpreting differences between cohorts and recognizing that recent cohorts are incomplete. Many instructors ask how large a cohort must be to read and whether cohorts could be built on something other than time. Students should avoid comparing young cohorts with mature ones as if both were finished. Composite data are acceptable. A cohort table with interpretation in APA 7 is standard, and strong papers suggest how to test explanations rather than asserting them.
How this BU 443 Unit 4 example is built
The analysis explains why grouping customers by arrival time reveals what averages hide and describes the data source and date. A cohort table gives customers and repeat rates at four windows for five cohorts, with counts in every cell and open windows marked. A findings section identifies three patterns and notes that size and loyalty run in opposite directions. An explanations section weighs gift orders and shipping damage for the holiday cohort and paid social's share and seasonal buying for the spring cohort, citing research on customer value. A section proposes tests of possible responses. The close explains how the newest cohort is read and notes limits from email matching and chance variation.
Reading the BU 443 Unit 4 rubric
Cohort analysis rubrics usually score cohort definition, table construction, handling of incomplete cohorts, interpretation and limits, together with presentation. Definition earns full credit when the grouping rule and outcome are stated precisely. Table construction scores highest when every rate appears with its count. Incomplete cohorts are judged on whether open windows are left unfilled rather than reported as partial rates. Interpretation rewards explanations checked against other data, such as gift messages or replacement credits, while acknowledging what cannot be separated. The limits criterion credits discussion of identity matching and chance. Instructors also value proposals for testing explanations, since cohort patterns suggest causes without proving them.
BU 443 Unit 4 help: mistakes that cost points
Cohort analyses often lose points by reporting rates without counts, which hides how reliable each figure is. Another common problem is comparing the newest cohort's partial figures with older cohorts' complete ones and concluding that recent customers are worse. Students frequently build cohorts on overlapping definitions, so customers appear twice. Some explain every difference with a single story, such as channel quality, without checking alternatives like seasonality. Others treat differences of a fraction of a point as meaningful in groups of a few thousand. Finally, recommending large changes from one year of cohorts overstates the evidence; proposing tests is the stronger response. Presenting the cohort table before any interpretation lets readers form their own view first, which makes the later explanations more persuasive.
Get a BU 443 Unit 4 example written to your instructions
Attach the Unit 4 instructions and the BU 443 rubric, plus the dataset your section supplies. We write a custom example that defines the admitting event, puts counts inside the cells, holds follow-up equal across groups, splits the customers a second way and marks the newest cohorts incomplete. 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 2: Data collection and tracking audit
- BU 443 Unit 3: Channel performance 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 245 Unit 4: Persuasive message
- BU 302 Unit 4: Team dynamics write-up
- BU 321 Unit 4: Confidence interval report
- BU 380 Unit 4: Capability and resource audit
BU 443 Unit 4 questions, answered
How large does a cohort need to be before I can read it?
There is no threshold to quote, which is precisely why the counts belong in the cells. Say plainly which groups are too small to support a claim and treat their movement as noise rather than signal. Where several small groups sit beside each other, combining them and saying that you combined them beats reporting six cells nobody should act on.
Should cohorts be built on time or on something else?
Time first, because that is what the method is for, then a second split on whatever your case makes interesting. Acquisition source is the usual choice since it connects this section to the channel work. Follow your instructions where they fix the grouping, and say what each split is meant to reveal, since two grids with no stated purpose read as duplication.
What if the data only covers a short span?
Then say so in the opening and scale every conclusion to it. A short span still supports early comparisons between groups, and it rules out anything about long term retention. Naming that limit yourself earns credit; letting a reader discover that a claim about customer lifetime rests on a couple of periods is the version costing marks.
Where can I find a free BU 443 Unit 4 sample paper?
The complete BU 443 Unit 4 customer and cohort analysis is free above, including the cohort table, the explanations weighed and margin notes. For a different customer base, a matching cohort analysis can be built from your prompt without charge, generally inside 48 hours.
How should BU 443 Unit 4 handle cohorts that have not finished?
Report only the windows that have closed and mark the rest as open. The sample shows the 2026 cohort only at 30 days, since its later windows would otherwise look weak simply because they have not ended.