SQL Tutorial for Marketers

Editor's Notes

• #3: Matt title suggestion – “Know everything about your customers: SQL For Marketers and Growth Hackers”
• #4: Let’s get through the boring stuff quickly! SQL is a database language designed to allow humans to query, manipulate and communicate with your database (what holds every bit of information about your product, customers and how they use your application). This database language – SQL – is incredibly powerful. It allows anyone who knows it to pull exact, specific data right from your database: the same place all other analytics tools are getting data from. Quickly explain what’s actually happening when you write an SQL statementMost internet startups and companies use MySQL, an open source database that SQL works with. A small group of other companies are using NoSQL databases, but we won’t cover those in this course.There are some other kinds of databases – Redis, mongo, memcache and the like – that we won’t get to in this course. Just know that SQL is by FAR the most common, and the most useful. For others, syntax is similar but not exactly the same
• #5: Oh, and you’ll make more moneySet the stage with common problems marketers ask. Say “can you answer these basic questions right now?”Script around common problems marketers want to answer. Who are my most valuable customers? Where did they come from? What’s my churn/LTV? Is it getting better or getting worse?
• #6: You have a question about your users. How much money have customers from the latest ad campaign spent on your product?Without SQL….You bug your development team for days, possibly weeks, to run a simple query that isn’t related to anything they’re working on, and (in their minds) isn’t that importantYou hack together an answer using a dashboard, admin panel and other tools like Google Analytics, Kissmetrics, and the like. It takes a long time and involves a lot of manual data entry. Plus, these kinds of integrations require HOURS of development timeTalk about how I had to wait weeks for devs to run a simple query that took them 10 minutes to see our top customers. Give example of improper filtering I did in Kissmetrics that led to incorrect assumptions and campaigns. BE SPECIFIC
• #7: With SQL, you have a question about your customers, campaigns or churn and you can answer it. No asking developers or product managers. Simply dig in, answer questions yourself and accept raises and praises as they come in from your boss and coworkers.Lots of new marketing managers and product marketing hires have “SQL Knowledge a huge plus”. This is a huge job hack, and immediately makes you 2x as effective
• #10: Set the stage with common problems marketers ask. Say “can you answer these basic questions right now?”Script around common problems marketers want to answer. Who are my most valuable customers? Where did they come from? What’s my churn/LTV? Is it getting better or getting worse?
• #11: Set the stage with common problems marketers ask. Say “can you answer these basic questions right now?”Script around common problems marketers want to answer. Who are my most valuable customers? Where did they come from? What’s my churn/LTV? Is it getting better or getting worse?
• #12: Set the stage with common problems marketers ask. Say “can you answer these basic questions right now?”Script around common problems marketers want to answer. Who are my most valuable customers? Where did they come from? What’s my churn/LTV? Is it getting better or getting worse?
• #13: Set the stage with common problems marketers ask. Say “can you answer these basic questions right now?”Script around common problems marketers want to answer. Who are my most valuable customers? Where did they come from? What’s my churn/LTV? Is it getting better or getting worse?
• #14: Set the stage with common problems marketers ask. Say “can you answer these basic questions right now?”Script around common problems marketers want to answer. Who are my most valuable customers? Where did they come from? What’s my churn/LTV? Is it getting better or getting worse?
• #15: Imagine not having to mess with Google Analytics, Excel dumps or bothering a developer to answer these key questions.Knowing this information gives you superpowers. It allows you to – in a matter of minutes – get answers to hard questions that you, as a marketer, and your company must know.Then, you can use your answers to inform and decide on your next brilliant marketing strategy.
• #16: I ran growth at SaaS company Airbrake before our acquisition by Rackspace. Like all companies, we wanted to improve our churn but first needed to learn more about why our customers were cancelling their accounts. Was it because they hated the product, couldn’t figure it out, or were struggling at one specific step? We also wanted to know the type of customers that were cancelling. Were they newer customers? Older customers? Was our churn getting better or worse?Airbrake is an error-tracking tool for developers, so customers who get value from the product use us to capture their errors. I had a hypothesis – were customers who cancelled not going through the process of setting up the tool so that they captured their errors? I decided to find out. I first ran the above query to see how many errors in production each user had generated. Then, compared the number of errors from cancelled accounts to the number of errors from our best customers (data I got using another SQL query!) to look at the difference. It was huge. It turns out, generating an error as an Airbrake user is a key metric – customers that generate errors are much, much more likely to remain happy customers than those that don’t. So, we fixed some things in our onboarding process and cut our churn by more than 40%. This was a huge win for a few hours of data analysis, and one day of implementing a few changes to get people generating errors!Don’t worry, all of these pre-built queries will be given to you later in the course for you to modify and run on your own. Just a quick note as we start, we’ll be using MySQL syntax for this course
• #17: I want you to have the same big win by the end of this course. First, decide what key metric you’d like to learn more about or improve in your business. Maybe it’s seeing who your best customers are, learning where most of your new signups are coming from, or decreasing your churn. Decide what metric you want to learn more about with SQL, and then do the above steps. Let’s take a second and have you do 2 things to get the most from this course.1. Email your developer and ask him or her to give you read-only access to your database, or to set up a ‘slave’ server that you can run queries against. 2. Ask your developer to show you quickly how to run queries against this database. You’re likely using HerokuDataclips, phpMyAdmin or some other relatively standard database management tool. Be sure to ask if you’re using SQLite3, PostgreSQL, MySQL or some other type of SQL engine.Really. Take 5 minutes to have someone on your team set this up for you. If you’re on your own, check out the links in the resources on how to set up HerokuDataclips or phpMyAdmin, two of the more popular solutions out there.Resourceshttp://sixrevisions.com/tools/top-five-best-database-management-tools/http://www.sqlmaestro.com/products/mysql/http://www.sqlmaestro.com/products/postgresql/http://www.sqlmaestro.com/products/sqlite/https://devcenter.heroku.com/articles/dataclips (only if you’re using Heroku)SQL database to test statements and play around with so you can see how things work together - http://www.w3schools.com/sql/trysql.asp?filename=trysql_select_all
• #18: I want you to have the same big win by the end of this course. First, decide what key metric you’d like to learn more about or improve in your business. Maybe it’s seeing who your best customers are, learning where most of your new signups are coming from, or decreasing your churn. Decide what metric you want to learn more about with SQL, and then do the above steps. Let’s take a second and have you do 2 things to get the most from this course.1. Email your developer and ask him or her to give you read-only access to your database, or to set up a ‘slave’ server that you can run queries against. 2. Ask your developer to show you quickly how to run queries against this database. You’re likely using HerokuDataclips, phpMyAdmin or some other relatively standard database management tool. Be sure to ask if you’re using SQLite3, PostgreSQL, MySQL or some other type of SQL engine.Really. Take 5 minutes to have someone on your team set this up for you. If you’re on your own, check out the links in the resources on how to set up HerokuDataclips or phpMyAdmin, two of the more popular solutions out there.Resourceshttp://sixrevisions.com/tools/top-five-best-database-management-tools/http://www.sqlmaestro.com/products/mysql/http://www.sqlmaestro.com/products/postgresql/http://www.sqlmaestro.com/products/sqlite/https://devcenter.heroku.com/articles/dataclips (only if you’re using Heroku)SQL database to test statements and play around with so you can see how things work together - http://www.w3schools.com/sql/trysql.asp?filename=trysql_select_all
• #19: Some SQL queries can take a long time to run. And, if you don’t have much experience, that can slow the performance of your actual app or product.
• #20: We now know that SQL is a language that allows us to pull information from a database. But how does a database work?Check out the example database table from W3schools. A database consists of multiple tables, just like an Excel file can consist of multiple spreadsheets. Each of these tables contains columns and rows with unique customer information you can use to look things up.
• #21: Easier to query – can filter, sort information and do it all automaticallyVisualize – can automatically update queries and data using different design librariesCan dump SQL into Excel if you want
• #22: Walk through what this query is doingnote some database systems require ending each statement with a semicolon ;
• #23: Walk through what this query is doingnote some database systems require ending each statement with a semicolon ;
• #24: There’s a lot you can do with SQL. In this course, we’ll only cover how you can extract data from a database using the language. As you saw on the last slide, databases are a lot like Excel sheets. What we’ll do with our SQL queries is simply write them in a way that tells SQL what data to select and show us, where to access that data, and how to display it. Can also use “Select Distinct” to pull only unique values from your database. The query above will pull in every user from that time period. If someone signed up 2x, they’d show up 2x. Not so with ‘Distinct’
• #25: Mention that I’ll cover “count” statement later
• #26: This one needs some more explaining.http://www.tizag.com/sqlTutorial/sqljoin.phpRight join – will list every row that exists on the right sideLeft join – same but lists every row on the left side of the table. In general these are a bit complex, often not necessary
• #39: Explain what stack is.
• #40: Explain what stack is.
• #41: Explain what stack is.
• #42: Explain what stack is.
• #43: Explain what stack is. All these are tools that will make it easier for you to get data from a SQL database
• #44: Explain what stack is. All these are tools that will make it easier for you to get data from a SQL database
• #45: Explain what stack is. All these are tools that will make it easier for you to get data from a SQL database
• #46: Use this information to write your queries. Remember, you have to name the tables and columns you want to pull data from.Show – displays all databases or tables (depending on your selection)Use – tells SQL which database you want to use to run queries onDescribe – describes column data
• #47: For this part, we’re going to show you how to construct some basic SQL queries using the test SQL database from W3 school. Though these examples will be more fitting for an ecommerce store, you should get a good idea of how to construct these queries by scratch, and you’ll be able to see the output they generate. In real life, you’ll be able to query your database for whatever parameters you want, then dump that data into an Excel or other spreadsheet for easy, non-programattic analysis.First, let’s look at how you can find customers who’ve made repeat purchases at your ecommerce store. Write a query that shows the repeat buyers and how much they ordered in the test SQL database
• #51: SQL is especially useful for cohort analysisA cohort analysis is a method of analyzing a metric by comparing its behavior between different groups – also known as cohorts – of users. These are generally based on the dates that a user started using a service or application.This kind of data allows you to see if you’re attracting users with higher LTVs, measure the efficacy of marketing campaigns over time, and see how retention compares between different cohortsIn this case, we’re looking at all users who generated an error in the month of March, 2014. Then, we can compare this to other groups of users from previous months to see if new users are sending more errors (aka using the product more)
• #52: This graph is showing how different marketing campaigns lead to orders months after the customer has been acquired. As you can see in this example from Chartio, Facebook ads lead to orders throughout the customer lifecycle, while customers acquired via TV advertising don’t place any more orders after 4 months. All of this data is exceedingly useful in planning your next marketing campaigns.