Customer Acquisition Analytics Success: This Week at &pizza!

May 01, 2020


The comprehensive customer list that I have been working on over the last few weeks is fully built, and I am excited to share that my work was used today in a board meeting! I am glad to know that the analytics I have been working on are making a difference to &pizza. It feels great to make a contribution!

As I continue to work on customer acquisition analytics, I think it is a good time to walk through what I mean by customer acquisition and how I am analyzing the success of promotions in this regard. Without further ado, let’s dive in!

Customer acquisition encompasses promotions and involves attracting new customers to &pizza. We often conceptualize this through what we call the customer acquisition funnel. It starts with getting a person who is unknown to &pizza to take advantage of a promotion and text &pizza. For the #HeroPies promotion that I worked on to give free pizza to hospital workers, this would involve a person texting “#Hero” to 200-03. In this case, &pizza gets some information about the customer, but the customer-business relationship is still in an early phase. The next step is to get the customer to create an account with &pizza. From there, &pizza continues marketing to them with the goal of getting them to carry out a transaction, before next focusing on making the person into a long-term, loyal customer.

A primary indicator of the value of a promotion is how many new customers it leads us to acquire. Therefore, we are aiming to better understand how effective our promotions are at successfully acquiring new customers (and at what price). Right now, this is my job: to design a data-driven way to measure the value of promotions. The customer acquisition funnel largely mimics the data-driven approach that I have created. Let’s talk about this data-driven approach.

To provide context, new data sets are generally created by performing what is called a “join” between two tables utilizing a field common to both of them. For example, if I were a bookstore owner and wanted to connect two tables, one of book inventory and one of book sales, I might perform a join on a field common to both tables, like book title (or, better in the eyes of data scientists, an artificial field without inherent meaning on its own like ISBN number).

The data-driven approach I designed continuously takes two data sets and performs a join on a field like customer ID or phone number. It starts with creating a comprehensive customer list populated with phone numbers by performing a join on customer ID between a spreadsheet of customer names and another spreadsheet of customer names and emails. This creates a singular, comprehensive customer list containing names, phone numbers, and emails. My boss said that this consolidated, cleaned-up, and comprehensive customer list filled with phone numbers is a huge resource that will help many others at the company as well. He is very happy with it, and I am, too!

Using this comprehensive customer list, I then performed a join using phone number with a database of customer text messages. To take advantage of a promotion, customers text a keyword to &pizza. For example, to get a promo code for a free #Hero pie, hospital workers texted the keyword “#Hero.” I created a list of customers who took advantage of the #Hero pie promotion using what is called a “where” clause. A where clause only returns rows that satisfy a specified criterion. Using the bookstore example again, the owner might run a query with a where clause requiring that the sale date be within the last business quarter. I used a where clause to select only customers whose texts included the keyword “#Hero.” This yielded a promotion-specific customer list that indicates the number of customers who we interacted with as a result of this promotion.

Using this promotion-specific customer list of participants in the #HeroPie promotion, I am currently in the process of creating a query that reveals what subset of #HeroPie participants proceeded to create an &pizza account. This will provide a key measure of whether our promotions are enlarging our customer base and how successful they are at doing so. Lastly, using yet another join, I will be able to see whether registered customers attracted by the #HeroPie promotion transacted and, if so, how much they spent. This will tell us the overall value of the promotion: how much the new promotion-recruited customers spend at &pizza. When all is said and done, I will make a dashboard out of all of these customer acquisition metrics for each promotion.

I am excited to be making great progress on this project and am pleased to know that this work is helping the company. I will keep you updated in the coming weeks. I appreciate your comments and questions so far and am always happy to answer any questions you may have. Please stay healthy and safe!



P.S. For this week’s Stat of the Week, I want to share a very well-done dashboard of coronavirus information. Created by The New York Times, this dashboard details coronavirus cases over time, by country, and by state, while also providing an overview of where things are trending for different places. Take a look if you are interested!

2 Replies to “Customer Acquisition Analytics Success: This Week at &pizza!”

  1. Thomas E. says:

    Tad, you have been doing some exciting work, especially in the past few weeks. While big companies like this want to give back to the community, they want to make it a win-win, also getting more customers and profit. Also, congratulations on having your work discussed at that board meeting – great to see that you continue to make a difference at your internship site.
    Question: How will you know whether a customer (that was already registered) is attracted by #HeroPies, another promotion, or just from hunger and remembering that &pizza is delicious? I am referencing the query that you are currently working on, mentioned in the second to last paragraph.

    1. Tad B. says:

      Thanks for the congrats, and fantastic question! The answer is that we have three key pieces of info that enable us to filter out truly new customers.

      First, we know what keyword the customer texted to take advantage of the promotion. This tells us what promotion the customer is taking advantage of. For the #HeroPies promotion, the keyword is #Hero, so we look for all text messages containing a “#Hero” (or something effectively similar).

      Second, we have a field that tells us the date when a person first texted &pizza. This marks the first time that their phone number appeared in our messaging system.

      Third, we have a field that indicates the date when a person has formally registered for an &pizza account. We then make sure that this registration date is after when the customer first texted &pizza.

      In the end, a new customer attracted by the #HeroPies promotion is one who texted the #Hero keyword and whose registration date occurs after the first time he or she texted &pizza. Hopefully this clears things up! Let me know if anything doesn’t make sense or if you have any other questions.

Leave a Reply