Data Analysis?

Buffer introduced a minimum vacation policy a year ago, in which we encouraged team members to take a certain number of vacation days, at minimum. If an individual was “falling behind”, i.e. not taking as much vacation as he or she should, our team would kindly remind the individual that vacation is beneficial to the entire team’s productivity and happiness. In this analysis we’ll try to measure the effect that this policy has had on the team in 2017. [Read More]

TL;DR Changes to the app and landing pages have led to a large (30%) and significant increase in Awesome trial starts. There are indications of a positive effect on Awesome MRR, but the probability of a true effect is inconclusive. The free plan experiment was a major confounding factor that did increase Awesome MRR. Background Back in September, Buffer experienced a period of relatively low MRR growth. [Read More]

Towards the end of 2017, Buffer sent out a survey to gauge the state of social media going into 2018. We had over 1700 responses, which we downloaded as a csv file and stored as an R data object. The questions and responses can be viewed here. The survey consisted of 30 questions about how users use social media. # get response data responses <- readRDS('state-of-social.rds') The data is quite messy and untidy. [Read More]

In order to gain a better understanding of how Buffer’s customers use the product within the constraint of the current plan limits, we can explore the data related to users’ active profiles. More specifically, we can use this data to answer the question “How many Awesome users with up to 3 Active profiles connected have 2 or 3 of the same profile types?” This is a very specific question. It will help us learn about usage and inform future decisions related to Buffer’s pricing structure. [Read More]

Buffer is well known for its 10 core values that guide the company forward. Continuously thinking and acting on these values has benefited the company a great deal over the years and has made an important contibution to our financial and emotional health. Every year, we ask team members to grade Buffer as an organization on how well it lives up to the values. In the survey, the responder is prompted to choose a value on a five-point scale for each of the ten values, with one corresponding to a value that needs a lot more work and five corresponding to a value that Buffer is strongly living. [Read More]

I often get questions like “How many profiles do Business customers have?” or “How often to Business cusotmers log in?”. These are useful questions to ask when designing and developing a product to fit the needs of a specific user segment, but a singular answer isn’t always sufficient. You could answer these questions with a mean or median, but these can be misleading if the metric in question isn’t normally distributed. [Read More]

It’s fairly well known that, for a subscription business model, subscriptions that are billed annually are more valuable than subsciptions billed monthly. That’s why many SaaS companies use the slightly misleading tactic of displaying a monthly price with an asterisk that indicates that the amount is the monthly amount of a subsciption billed annually. The prospective customer has to either multiply the amount by twelve to see what he or she will actually be billed, or lose out on this “deal” and pay more to be billed monthly. [Read More]

In September 2016 the growth team conducted an experiment on the price of Awesome plans. Approximately half of new users that signed up for Buffer were exposed to a variation of Buffer in which the price of the pro-monthly plan was raised from ten dollars to twenty dollars. The price of the pro-annual plan was raised from 102 dollars to 204 dollars. Increasing the prices of the pro-monthly and pro-annual plans failed to increase the amount of revenue generated by customers that subscribed to the plans. [Read More]

Every week Buffer hosts an hour long discussion on twitter called Bufferchat, in which participants talk about social media, productivity, and occasionally self-care. The topics change from week to week, and there is often a guest host that moderates. People join the chat from around the world. Many industries and company types are represented. This week, I thought it would be fun to collect the tweets and do some basic analysis on them. [Read More]

In recent times we’ve seen MRR growth that is roughly linear. This wasn’t always the case - Buffer has had periods of higher growth and periods of low growth. We weren’t always growing MRR by 25-30K each month. In this analysis we’ll try to identify points of inflection at which growth in MRR changed significantly. We’ll use methods in the changepoint package to identify points at which there is a shift in the amount of MRR growth. [Read More]