Antora Majumdar

Those who frequent social media platforms like Instagram or YouTube know that beauty vloggers and lifestyle influencers command large followings, and those followers are always eager to hear their favorite personalities review the latest trends in beauty or watch them walk through their nighttime skincare regimen. The result on sales and market growth is evident as the beauty industry has only been growing globally, thanks in large part to millennials. Another industry that has been growing in popularity with millennials is subscription services, not just for streaming but things like meal prep, pets, books, and beauty. Newer companies like Birchbox and Ipsy have capitalized on beauty and subscription boxes, finding a market in customers wanting to try new beauty products every month. Even established brick and mortar beauty retailers like Sephora have started offering these subscription products. With so many competing offerings, I wanted to investigate what consumers might want the most from their beauty subscription box.

To explore different attributes of subscription boxes in the beauty category that might appeal to consumers I structured my conjoint analysis around four features: Type of beauty products in the box, monthly price per box, number of products in the box each month, and curation, or how products were chosen, either by the consumer themselves or an “expert.” Table 1 illustrates the four features and two levels that made 16 different product configurations presented for evaluation in this conjoint.  

So, why these particular features? Price helps determine a consumer’s willingness to pay, but it can also be a signal of product quality that may play a role in this type of subscription service. In addition to price, “Type,” “Products,” and “Curation” would be interesting to explore as they each correspond to a different value a consumer might prioritize in subscribing to a beauty box. Makeup and skincare markets have both demonstrated a global trend of growth in recent years, and I was curious to see whether one type of beauty box might be preferred over another, giving a clue as to what might do better in the market. I felt number of products could be an indication of what a subscription box is worth; do consumers see more products as better value and are thus willing to pay more? I chose levels of five and ten because five seems to be the standard number of products in existing beauty boxes, with 10 more rare and on the higher end. And finally, personalization has been a big topic in marketing as of late, especially with the amount of data and level of detail companies now have on consumers. In choosing “Curation” as a feature, I was curious to know whether customization was something consumers in this market cared about.

THE SUBJECT: For this conjoint I selected a respondent with the following characteristics: female, 23- years-old, currently a medical school student, no steady income but does part-time work, living in the Southwest United States, and in the millennial demographic. I felt that the respondent fit into the target market of many existing beauty brands and beauty subscription box companies. Additionally, I selected them because they are an avid user and consumer of beauty products, both cosmetics and skincare, have subscribed to beauty boxes in the past (both Birchbox and Ipsy), and follow beauty vloggers and influencers on social media. I also felt the respondent’s experience with beauty subscriptions would make it easier for them to visualize the product.

THE ANALYSIS: In order for the respondent to rate their preferences of the different feature and level combinations, I created a survey using Qualtrics, an online survey software. I instructed the respondent to rate their preference for the overall product on a scale from 0 – 100, with 0 corresponding to least preferred and 100 corresponding to most preferred. See Figure 1 below for an example of the individual question format.

I translated the respondent’s survey responses into an Excel table.

One thing I noticed while looking at the respondent’s results is that they had very high preference ratings for all of the products. The lowest preference rating was 48 for a subscription box with makeup products for $20/month, with 5 products in it, and an expert picking the products. Two options scored a preference of 100: a subscription box with makeup products for $10/month, with 10 products, where the customer can pick their own products. And a subscription box with skincare products for $10/month, with 10 products where the customer can pick their own products.

This might be a result of the respondent themselves, as they often purchase both makeup and skincare products and thus feel they know what products would be best for them. This likely also means that for a company looking to target consumers who are not as familiar with makeup and skincare items, these results might not be as accurate a measure. For companies wanting to target customers who already have a high degree of affinity for beauty products and beauty box subscriptions, this respondent’s scores would be a better fit. To translate the results into a for regression analysis I coded the levels I thought would be most attractive as “1” and the levels that would not be as appealing as “0.” The following is a matrix of the results.

I then ran a regression on this matrix for the regression output. Table 2 contains the coefficients, standard error, and t-statistic from the regression output.

The coefficients can be used to determine how much the respondent preferred each option within the different features of the beauty subscription box. To compare how the respondent valued these levels, I translated these results into a “Partworth.”

In conjoint analysis, partworths are a quantitative way to gain insight into someone’s decision making process, by telling us how much each attribute can influence their choice. The results from this analysis tell us that the respondent likely values price as the most significant factor when making a decision about this kind of product. Please see Appendix D for a partworth table as well as the calculations for this measure. From the results, I found the partworths for each feature to be the following:

• Type of Product: 1%

• Price: 90%

• Number of Products: 33%

• Curation: -24%

At a glance, we can see that the respondent preferred makeup to skincare by one “util,” the price of $10/month over $20/month by 90 utils, 10 products over five products by 33 utils, and they actually valued the expert curation negatively, meaning they would prefer to pick their own products. When setting up the matrix, I guessed incorrectly that a respondent would prefer to have an expert pick out their products, which appears as a negative output when we look at the partworth here. See Appendix E for a graph comparing the relative partworths of each feature.

Comparing the partworths, I also noticed that there is a much more significant difference in the partworth of type of product and price, versus the number of products and the price. The actual difference in preference of partworths being .89 versus .57. This tells us that in relation to price, the respondent cares more about the number of products than they do the type of product. And they also care more about how the products are curated when compared to the type of product. I wanted to see how these differences measured up when it came to the type of product, which had a partworth of .01 meaning there was really not much of a preference for this feature. The following chart uses the absolute value of the difference in partworth, in order to compare the magnitude of the differences when it comes to the type of product in the box. We see that price matters much more, while number of products and curation are weighted

With this we can calculate the willingness to pay for each feature. The WTP results are as follows:

• WTP for more products: $3.63

• WTP for makeup as type of product: $0.11

• WTP for expert curation: -$2.64

The interesting result here is that there is a negative WTP for an expert to curate a box, meaning the respondent is not willing to pay more for expert curation, and in fact would expect to pay less for this attribute. When thinking about the features and levels in terms of trade-offs, we can calculate what the respondent would give up as part of the box for one feature for more of another feature.

For number of products, the trade-offs are as follows:

• Type: People would give 0.15 of a product for a box with contents of makeup over skincare

• Curation: People would give up 3.62 of products to not have an expert curate their box

• Price: People would give 13 products for a better (aka lower) price

Although the number of products is a continuous value, it is difficult to measure portions of one product in a meaningful way. If the metric used instead had been ounces of product, this would make more sense to determine how much product exactly a customer would be willing to give up for other features they liked more. For price, the trade-offs are as follows:

• Type: People would sacrifice $0.11, meaning they would pay $0.11 more for a makeup product versus a skincare product

• Curation: People would sacrifice -$2.64 to have an expert pick their products versus picking themselves, meaning they would pay more to be able to pick their own products.

• Number of Products: People would sacrifice $3.63, meaning they would pay more for a greater number of products in their box

I could not compute this trade-off value for the “Type” of product or “Curation” of the product because these variables are not continuous. There is no inherent numerical value or order associated with them. We can do a breakdown of how to predict what kind of subscription box a customer would buy if profits would be equal in all scenarios. The lower price of $10/month outweighs all other features, suggesting this respondent is most aware of price when deciding on the preference of a beauty box.

BUSINESS IMPLICATIONS: Thinking of this from a business perspective, if a subscription beauty box was targeting similar customers to this respondent and the company could not compete on lowest price, they would have to offer something that was significantly more compelling in order for the customers to buy from them. So, another way a company could increase subscriptions might be to add more products per box in a month. However, the problem with this tactic is that offering more products will cost the company more. To maximize profits, it could be beneficial to look towards customization.

Customization seems to be something consumers are demanding more frequently these days, but when it comes to beauty there may be some caveats in what exactly consumers are looking for. In this conjoint example, letting a customer pick their products versus having an expert pick products are both options that offer opportunity for a beauty box to be tailored. However, the results from analyzing the partworth of curation tell us that customers who align with the respondent would much rather choose products for themselves. Often beauty subscription box companies will have quizzes asking customers what they are looking for. If a company were to use this quiz to present potential customers with options they could then narrow down themselves, they could differentiate themselves from other similar subscription box offerings.

If a company was looking to capture customers who were not already interested in beauty subscription boxes or not as familiar with makeup and skincare, this respondent’s answers would not be as relevant to designing a product for that market. Even if a company was looking to target women interested in skincare over makeup, I would recommend conducting a conjoint analysis with respondents who fit their audience segment more closely, to find more accurate insights about which features would appeal to these potential buyers.