Growth Playbook #1 - Historical Campaign Analysis

Introduction

Klaviyo is one of the most popular data sources in the DTC space. Klaviyo is an email marketing solution which is used to drive individuals back to the website with the hope that they make a purchase. Klaviyo has two core features, campaigns, and flows. Campaigns are one-off "email blasts" sent to one or more lists of subscribers. A flow is a sequence of automated actions, which can include emails or SMS messages that are triggered when a subscriber performs a specific action.

The opportunity

A typical DTC business will send dozens of email campaigns a year. The performance of each campaign can be seen within Klaviyo's native reports, as well as via Klaviyo's API. This provides us with an opportunity.

‍We can sort the brand's past Klaviyo campaigns by their relative performance to identify the high performing campaigns and any outliers. These campaigns can then be assessed by the brand to identify similarities. Any insights gained through this process can become hypotheses for A/B testing.

How would we identify the high performing Klaviyo campaigns?

The most effective way to identify high performing campaigns, and in particular outliers, would be by using a scatterplot data visualization similar to the one shown below.

Each circle in the screenshot above represents a previously sent Klaviyo campaign. The size of the circle represents the relative size of the campaign in terms of number of recipients. The Y and X axes would be dynamic but in order to normalize the data we would most likely use "Purchase Rate" (the percentage of the recipients who made a purchase), and "Revenue Per Unique Open" (total revenue divided by the unique count of individuals who opened the email).

We only want to identify high performing campaigns sent to a large number of recipients

It's important to note that we would filter out campaigns which were sent to a small number of recipients. Depending on the size and maturity of the brand, this could include all campaigns sent to 5k recipients or less. The reason for this important step is because there will be campaigns that were sent to a very healthy list of "high performing" past customers which will naturally be high performing. We won't learn much by including these campaigns in our list.

Creating the list and sharing it with the email team

Once we've selected the relevant options in the parameters of our spatterplot, we can isolate the high performing campaigns, export the list and share it with the team responsible for running the ads.

Analyzing the high performing campaign

For each campaign the following information should be gathered:

• The send month of the campaign ⟶ We will want to see if the time of the year has a significant impact on performance.
  
• The lists which were included in the campaign ⟶ A major factor to the performance of the campaign will be the audience who received it. We want to see if there is a specific list, or set of lists which appear multiple times across these campaigns.
  
• The subject used in each campaign ⟶ Any specific words or phrases which stand out?
  
• The copy, graphics and overall layout of the email ⟶ The look and feel of the email will have a direct impact on the click-through-rate which is a leading indicator to the revenue generated by the campaign.

The potential revenue generated from running this data playbook

Assumptions:

• Klaviyo is responsible for 20% of revenue generated by the average DTC brand.
• Campaigns are responsible for 50% of the revenue generated by Klaviyo, Flows would be responsible for the rest.
• In our example we are assuming the brand generates $50M a year.
• Therefore we can say that Klaviyo is responsible for $10M a year with campaigns being responsible for $5M.
  

Low probability: 4+ insights discovered which would lead to 15% improvement in campaign performance ⟶ Additional $750k a year.

Medium probability: 2 - 3 insights discovered which would lead to a 10% improvement in campaign performance ⟶ Additional $500k a year.

High probability: 1 insight discovered which would lead to a 3% improvement in campaign performance ⟶ Additional $150k a year.