# Customer purchasing mix memo

Richie Lin · Portfolio sample · September 2026

Public data analysis · Portfolio companion

Prepared from the same 440 record historical UCI Wholesale Customers dataset used in the independent portfolio project. This companion compares recorded channel labels. It does not reproduce K means clusters.

## The brief

Do the two recorded channels have different category spending mixes?

## Worked example

Recorded annual spending in monetary units (m.u.). Category share is a ratio of aggregate spending

| Recorded channel | Customers | Total spending | Milk + grocery + detergents/paper |
| --- | --- | --- | --- |
| Horeca · Hotel / restaurant / café | 298 | 7,999,569 m.u. | 2,444,918 m.u. · 30.6% |
| Retail | 142 | 6,619,931 m.u. | 4,871,858 m.u. · 73.6% |
| All records | 440 | 14,619,500 m.u. | 7,316,776 m.u. |

## Recommendation

Use the mix difference to frame a follow up research question.

Milk, grocery, and detergents/paper account for 73.6% of recorded Retail spending and 30.6% of Horeca spending. A useful next question is whether category relevant examples improve message relevance within each channel.

## Show the method

Group records by Channel, sum the six spending categories, and divide the three selected category totals by each channel’s total. The displayed shares describe aggregate spending. They are not the average customer’s percentage.


## Separate observation from explanation

The records show a purchasing pattern. They do not establish customer motivations, profit margins, current market size, or the effect of a campaign. A few large customers can influence aggregate shares.


## Recommend the next study

Review within channel variation and gather customer feedback before proposing a message test. For a current business, first validate the pattern using current, permissioned customer data.

- Define the intended customer action.
- Compare category relevant messaging with a standard message within each channel.
- Use a planned sample and a consistent response window.
- Review commercial value and customer relevance before scaling.

## Learning note

Purchasing categories describe recorded behavior. They can suggest a research question, but they do not explain motivation or predict campaign impact.

## Files and sources

- Channel summary CSV: customer-insight-memo.csv
- Verification script: verify-samples.py
- Original source CSV: wholesale-customers.csv
- Cardoso (2013), UCI Wholesale Customers · CC BY 4.0: https://doi.org/10.24432/C5030X
- Creative Commons attribution license: https://creativecommons.org/licenses/by/4.0/

Related experience: https://richielin.com/pages/customer-segmentation

View online: https://richielin.com/pages/samples/customer-insight-memo