The grandpa Soap COMPANY

🏆 Winner of the American Marketing Association, Cincinnati 2017 Pinnacle award for Best Use of Insights to Drive Experience, Engagement, and/or Innovation.

This is an in-depth case study, so pour a mug of your favorite caffeinated or herbal beverage, and dig in. Ages of the personas have been obscured, and interests and purchase behaviors have not been discussed, to avoid revealing information proprietary to the client.

Opportunity / Problem

The Grandpa Soap Company is known for all-natural Pine Tar Soap, manufactured since 1878, and a line of soaps with natural and Earthy ingredients such as witch hazel, cornmeal, and baking soda. In 2016 they modernized their branding and product lineup, including new Earthy ingredients such as charcoal and rose clay.

An independent review conducted by The Grandpa Soap Company recommended certain personas to target for their modernized branding and products. These personas were developed without any supporting data. We had been running social advertising for the client for several years, and saw the opportunity to test the personas and potentially develop better ones, all supported by data.

The study also demonstrates that smaller brands can uncover valuable, actionable insights with a modest and carefully-designed effort.


  1. Develop personas supported by quantitative data. Personas will be used to guide the client’s marketing and advertising across all channels. Categories sought included… 

    1. BRAND CHAMPIONS – who’s most enthusiastic

    2. VOLUME LEADERS – lots of customers

    3. COST-EFFECTIVE – least expensive to advertise to

  2. Test the personas that had been recommended without supporting data to the client.


We utilize Facebook as the world’s largest focus group. Like any focus group, Facebook isn't perfect, but it does offer some outstanding advantages:

  • an enormous number of subjects;

  • unparalleled knowledge of their interests, demographics and purchase behaviors; and

  • the ability to test their responses without being influenced by the experiment.

Tested Personas

For this study, we constructed audiences on Facebook to match eight base personas, as well as variants based on income and interests, for a total of twenty-five personas. We ran Page Like ads to these audiences, and recorded the response rates and costs. In the process we brought followers to the client, but that was a side effect of the test.

Audience members could be any ethnicity.

•Karrah, Suzie and Konrad are the personas recommended to client without supporting data. In addition, we created Karrah’s Mom/Daughter, Suzie’s Aunt/Niece, and Konrad Sr/Jr, who differed only by age. We’ve obscuring whether they’re older or younger to keep client results confidential. Ditto for interests and purchase behaviors.

Charles, Juliet, Sierra, Thomas and Virginia were created for this study as likely consumers for client’s flagship product, with a variety of interests and purchase behaviors. Each has a different age, and they were split into three personas differing by income. ($50k is the US median household income.)

For visuals, we used two photos we knew were high-responding from previous advertising. The product was shown bare, without packaging or branding. It’s clear from the photos what the product is and its natural, Earthy character.

The test was a page like ad on Facebook. The visual is the bare product, no packaging or strong indication of branding. The client was in the process of modernizing branding, so we weren’t testing the brand – testing personas for the flagship pine tar soap product.

Control Personas

During the study it became clear that age was the biggest factor in response rates. The question arose whether age is a factor for Facebook responses, or for interest in soaps or interest in natural products, and thus might have nothing to do with the pine tar soap we were testing. To address this, we created a control group with a different natural soap brand owned by the same client, using four personas differing by age; bare photos of that product (which we knew were well-received from previous advertising); and general-purpose interests and purchase-behaviors for natural products.

When it became clear that the strongest variable was age – with a 2% response rate at one end and nearly 7% at the other end of the spectrum – we created a control group to test whether age was always a factor for Facebook page likes, or possibly always for soaps or natural products, or if it was really due to pine tar soap.

The client owns several natural soap brands and Kirk’s was a perfect control. We created four general personas differing only by age, and tested them with this page like ad. Again the image is the bare product, no packaging or strong indication of branding.

The control results were only slightly age dependent, indicating that age is a real factor for response to the product.



Suzie’s Aunt/Niece    
Virginia Upper-Mid income
Karrah’s Mom/Daughter

Brand champions are the personas with the highest response rates. These people are most interested in the product, most likely to tell everyone they know about it. They can be targeted for word-on-the-street buzz.

Response Rate (%)

% Likes Per Impression – Ordered By Age


Thomas Lower-Mid income
Virginia Upper-Mid income
Suzie’s Aunt/Niece

Cost-effective personas had the lowest advertising cost per response. This is a combination of bid prices for those people on Facebook’s platform, determined by how many other advertisers are targeting them, and their response rate to our page like ad. These people can be targeted for efficient ad spen

As with response rates, the response costs depend on the test – in this case, a page like ad with photo – and platform – in this case, Facebook. And as with any study, we assume that similar patterns will occur in other situations. If many companies are targeting people like Karrah, she will be expensive to advertise to.

Response Cost ($)

Dollars Per Thousand Likes – Ordered By Age

Juliet Upper-Mid income
Thomas Upper-Mid income
Juliet Lower-Mid income

Volume Leaders:



Volume leaders are the personas with the most people in the US. They can be targeted to generate sales volume. We didn't create a graph for this one; the data is in the table above.


Lower-Mid income consumers are often overlooked, yet their personas were among the top performers for this product in each category. The Lower-Mid income version of Thomas had the best response rate per ad spend. 


The strongest correlation for response rates, indicating brand champions, was not with income, interests or purchase behaviors – it was age.

There are reasons to doubt age – or any factor, when one first finds it – is a real factor in people’s interests. It can always be an artifact of the test situation, in this case page like ads on Facebook. For age, our experience is that older users tend to like lots of things for the excitement of being online, yet younger users tend to like lots for FOMO or to identify with it. Older users are more interested in prosaic stuff like soap, but younger users are more interested in natural products. Hence the control group with a different natural soap.

Ages of the personas have been obscured to avoid revealing information proprietary to the client. 

Tested Personas


One Age Extreme

Other Age Extreme


Response Rate

2.0% (average of Charles Lomid and Karrah)

6.4% (average of Viginia Upmid and Suzie’s Aunt/Niece)

+220% better response at one end of age spectrum;
we assume needs to be corrected for response to natural soaps

Control Personas


One Age Extreme

Other Age Extreme


Response Rate



+30% better response at best end; we assume this is the age response to natural soap in this test situation

Response Rates of Control Group

% Likes Per Impression – Ordered By Age – Compare to Test Personas Ordered by Age

Tested Personas, Adjusted for Age Response to the Test Situation

True Improvement

+190% = 220% - 30%

Thus the client can expect a 190% higher interest in this product when targeting one end of the age spectrum over the other.


All the personas previously recommended to the client were dramatically out-performed by their generational relatives Karrah’s Mom/Daughter, Suzie’s Aunt/Niece, and Konrad Sr/Jr, as well as most of the other personas we created for the study.












Karrah's Mom/

Susie's Aunt/

Konrad Sr/Jr












Based on this, adjusting the personas previously recommended to them results in a 146% higher interest in the product

Side Note

You might wonder, with all the mojo going to other platforms, why we depend Facebook for testing.


Even without access to IBM Watson-style big data, it’s possible to do quantitative research for marketing decisions. In this study and other work for clients, we basically use Facebook as the world’s largest focus group. The quantitative support that results can make a big difference in decision-making for marketing success.

Client is now able to target…

  • Brand Champion audiences to generate word of mouth buzz,

  • Volume Leader audiences to generate high sales volumes, and

  • Cost-Efficient audiences to maximize paid advertising.

  • One persona in particular, the Lower-Mid income version of Charles, had the lowest cost per response and an above-average brand champion rating. Charles Lower-Mid and similar audiences are the obvious sweet spot for marketing and advertising.