Leave a comment
Get the GH Bookmarklet

AMAs

Spenser Skates is the CEO and Founder of Amplitude, a web & mobile analytics platform that helps growth and product teams understand user behavior. 

Spenser founded Amplitude in 2012 to meet the analytics needs of the next generation of product and growth managers. He and his co-founder Curtis Liu built Amplitude to enable product leaders to discover the behaviors that drive retention, engagement, and revenue. 

Spenser first discovered the need for a better analytics solution while working on Y Combinator startup Sonalight, a voice recognition app for Android, which led to his idea for Amplitude. He realized that existing analytics tools didn’t scale well for mobile and weren’t flexible or powerful enough to deliver more than vanity metrics.

Prior to Sonalight and Amplitude, Spenser worked as an algorithmic trader for DRW Trading Group and also explored a few smaller startup ideas. He has a B.S. degree in Bioengineering from MIT and is a two -time winner of MIT’s largest programming competition, Battlecode.

Spenser is excited to discuss user behavior and how you can use analytics to drive growth and retention. In addition, he’s happy to answer questions about growing a SaaS startup. (For example, he’s made some big moves on transforming employee equity compensation).

You can follow him on Twitter: @spenserskates

He will be live on Apr 7 starting 930 AM PT for one and half hours during which he will answer as many questions as possible.

  • WD

    Walker Donohue

    over 2 years ago #

    Is analytics fundamentally reactive or proactive? That is, should you use your analytics to understand the results of your experiments, or should you use them to determine where you need to go and what you need to do?

    • SS

      Spenser Skates

      over 2 years ago #

      "Analytics" means a lot of different things depending on the context. I feel like I see a new analytics company pop up every week :)

      Our take at Amplitude is it's about observational data on your the interactions your users are having with your application. You're trying to understand "this is what has happened with my userbase" more than anything else. That said, the reason you look at this data is it has a so that you can use it to figure out what to do next.

      For example Facebook's famous "7 friends" insight was that it took 7 friends to get a retained user- that observational. They took that and then figured out that adding features to encourage their users to add friends was where they needed to go.

      • SS

        Spenser Skates

        over 2 years ago #

        Anuj- the first thing I'd look at is whether or not an event was performed in the first day and how predictive that is of your retention a week later. Look at that across all of the events a user can do and that will give you great insight into what events are important (in order to do that you do need tracking of the events that could be important- I'd say track the top 20-50). That's the most basic analysis I recommend starting with. Once you've done that then you can move onto more complex patterns of behavior.

      • AA

        Anuj Adhiya

        over 2 years ago #

        I get that the tool helps but is there a framework you use to arrive at conclusions like in the FB example? Put differently how would you even put the foundation for an observation in place when at the start you have no idea what you're going to find? Is it as simple as finding patterns within events in x time frame or something else/more?

  • DF

    Diane Florescu

    over 2 years ago #

    Hi Spenser! I've got some "technical Qs" related to Amplitude. Currently using Mixpanel, but I signed up for Amplitude as well! When analysing the cohort's behaviour would I be able to learn which users are the most likely to convert, before they do? Could you iterate more on this predictive feature?

    Q2- How can we make sure that Data Quality is right? Not just for Amplitude, but generally speaking for data management?

    Thanks,
    Diana

    • SS

      Spenser Skates

      over 2 years ago #

      I strongly believe predictive is the wrong way to look at behavioral data! It's not about being able to understand that one group of users will convert and another will not. Much more important is the question of WHY they will convert or not. Saying "I know this group will convert because of a black box algorithm" doesn't let you do much from a product perspective. What you want is to say "this behavior is what is leading users to convert" and then you know to focus on that behavior as the driver.

      Data quality is a huge issue. How do you know what the data you're looking at means or whether you can trust it? It's a big part of what makes analytics so difficult to use. The issue is that the people who set up the analytics, instrument it, maintain it, and consume it can all be different people! What works well is to categorize the events you're tracking into an event taxonomy document that's shared across the company. That way everyone is on the same page about what a data point means.

    • AV

      aditya vempaty

      over 2 years ago #

      Diane, are you talking about a single source of truth or actually ensuring the data you capture is accurate?

      As these are two different things, often times setting up your analytics from day one with the things you want to track is key to ensure data capture is correct. This means instrumenting correctly, dedicating resources to doing the architecture of the event schemes as well.

      If you are thinking about single source of truth, then we have seen some customers setup things such that there is just one location for everything vs multiple systems of access. But to have this the product you use has to enable access to raw data, but have dashboard access for all that need it, enable sharing of data that is easy to consume to everyone regardless of ability to do SQL or be a master at spreadsheets.

  • LS

    Logan Stoneman

    over 2 years ago #

    Hey Spenser! You have had quite the education background: studying at MIT, learning at Y-Combinator, and building Amplitude. While each have their own merits, what period of life & what during that period would you say you gained the most amount of knowledge? If that's too broad, you can narrow it down to:
    -Knowledge about growth/marketing
    -Knowledge about yourself and what career & life paths to pursue

    • SS

      Spenser Skates

      over 2 years ago #

      Hi Logan, thanks for the kind words! In college I had massive ADD around what career path I was going to take. I started out planning to do biology research as an academic, then got excited about nonprofits, then finance, and then startups. I remember my mom at the time was very worried that I was never going to be focused enough to be successful.

      I spent the year after I graduated from college trying to figure out whether I was good enough to start a company of my own. I read everything I could about starting a company and talked to as many entrepreneurs as I could. I applied to TechStars (see this gem: https://www.youtube.com/watch?v=4PIM5wWut5Q )

      The biggest thing I learned was 1) almost all successful startup stories have a point about a year to a year and a half in where things looked hopeless and they probably should have given up and 2) if you were willing to stick it out for at least 2 years you'd likely come out the other side with something successful. That convinced me that as long as I was willing to not doubt during that time then I could make it worthwhile.

      • LS

        Logan Stoneman

        over 2 years ago #

        Thanks Spenser - I definitely have to concur with what you said about sticking it out with the startup. It's still tough to know when you are intelligently sticking it out or just being naive to the idea that your startup isn't going to work.

    • HQ

      Hila Qu

      over 2 years ago #

      great question, Logan!

  • SS

    Spenser Skates

    over 2 years ago #

    Hi GrowthHackers! Excited to be answering questions for the next hour and a half. Happy to talk about analytics, startups, or anything else. Fire away.

    • SS

      Spenser Skates

      over 2 years ago #

      Signing off for now, thanks for all the great questions! Really like how thoughtful this community is when it comes to analytics.

  • SE

    Sean Ellis

    over 2 years ago #

    Hi Spenser, thanks for doing this AMA with us. I'm curious when you started Amplitude, what was the opportunity in the analytics space that you saw as untapped? There are clearly a lot of analytics companies and have been since the early days of the Internet. But I've heard from people who understand the space well that Amplitude is doing something special. It would be great to get some context on that initial vision and if that continues to be your main focus or you have shifted significantly as the market has evolved. Thanks!

    • SS

      Spenser Skates

      over 2 years ago #

      Thanks for letting us be part of the community, Sean!

      There were two big openings we saw when we started Amplitude. The first was that what growth teams were doing at large companies was 10x more sophisticated than what was out there at the time. Instead of continuing to improve on their analytics products, most analytics companies had pivoted to doing advertising or marketing products. The second was that there was about to be a massive shift to mobile and no one was really focused on the unique challenges of mobile. We're still focused on pushing the boundary of what analytics can be (I answered part of this above) and think there's a long way to go.

  • AA

    Anuj Adhiya

    over 2 years ago #

    Hey Spenser - so stoked to have you on.

    I promise I wasn't paid to ask this q but it's one I've been chewing on for a while.

    As someone who's been trying to wrap their brain around existing solutions like KissMetrics and MixPanel, what would you point to as they key differentiators to make the switch to Amplitude?
    What are the key things anyone should be looking at to make an accurate comparison between all of these tools to decide what's best?

    • SE

      Sean Ellis

      over 2 years ago #

      To add onto this question... Should companies "make the switch" or is it sometimes appropriate to have multiple analytics solutions? In my experience, most companies have at least two (Google Analytics and something else).

    • SS

      Spenser Skates

      over 2 years ago #

      Thanks Anuj and Sean! There are two big ones. The first is what questions you're trying to answer/what you're looking out of an analytics solution, and the second is how much data you have.

      If you're looking to understand user behavior deeply, like what behaviors lead to conversion or retention, then that's something only Amplitude can do. If you're looking to get an "all in one" analytics/marketing/A/B testing platform than Mixpanel would be a better choice. The second is how much data you have. If you don't have many users (less than 10k monthly actives) than Amplitude/Mixpanel/Kissmetrics are the same. Once you get beyond that and scale becomes an issue then that's where Amplitude can perform about 10x better. I'm biased on that one but I encourage you should look at each pricing page to see how they compare when it comes to data at scale.

      Sean- it's good to use multiple analytics solutions. We often see a free product (Google Analytics/Flurry) running alongside us. If you're deciding between Amplitude and Mixpanel then we have a number of ways to help make that transition easier.

  • WD

    Walker Donohue

    over 2 years ago #

    Spenser, what's your take on the statistic that says (I can't find a link right now) something like 68% of companies struggle with their analytics because they don't understand the key drivers of their business? It was a surprise to me, but maybe not to you. How can analytics help companies figure out what they should be measuring, or can it?

    • SS

      Spenser Skates

      over 2 years ago #

      This is really hard to get clarity around without the right framework. We call this coming up with your "magic threshold" metric to understand what is ultimately driving your business. We have a blog post that explains it fully here: https://amplitude.com/blog/2014/07/29/find-the-key-to-your-apps-growth-without-an-army-of-data-scientists/

      The idea is you want to figure out what outcome you care about for your users (for most businesses this is retention). Once you've done that, see what the differences are between users who retain and users who don't. Is it a particular pattern of behavior? Is it what they're buying? Is it where they're coming from? Slicing your retained vs churned users along all these behaviors and attributes will allow you to understand what the key drivers of retention and ultimately your business are.

  • YP

    yadav prasanth

    over 2 years ago #

    Hi Spencer!

    Thanks a lot for sharing your valuable words here.

    I have a simple question.

    I have a very important question according to our startup, we have a healthcare product. And we have lot's of data put in our App expecting the user need's all of it. My question is we are dealing with such app with more data than usual apps, how can we measure user dropouts and use that in gaining traction as well as customer retention ?

    Best Regards
    Prashanth

    • SS

      Spenser Skates

      over 2 years ago #

      Prashanth, thanks for saying hi! I'd actually say you want to understand which feature the users need and use the most in your application, and just focus on that one. Send me a note at spenser@amplitude.com and I can get you set up with some of the basics on your app.

  • AV

    aditya vempaty

    over 2 years ago #

    How do you create a culture of using data?

    • SS

      Spenser Skates

      over 2 years ago #

      It's all about making it easy for anyone in your company to do it. Analytics products across the board are too hard to use.

  • AT

    Andrew Tate

    over 2 years ago #

    Hey Spenser,

    Thanks for doing this! It always worried me when companies say they are "data-driven" because it sounds like they are chasing the numbers. Do you guys have any framework that you stick to which allows you to use data effectively, but still allows for your own experience, insight, and intuition?

    Cheers,

    Andrew

    • SS

      Spenser Skates

      over 2 years ago #

      Andrew, if I could only pick one of between talking to my users and using analytics to see their behavior, I'd pick talking to users every single time, no matter what company I was at. That sort of qualitative data is so valuable because you get insights around what people's pains are and how they're using your product that you'll never get out of the data. That said, data is invaluable once you're at scale and you need clarity because you have too much anecdotal data that points in different directions. Here's how I think about it: generate hypotheses from qualitative experiences, then use data to test those hypotheses.

  • JK

    James Kaiser

    over 2 years ago #

    Hey Spenser!

    I read about your changes to the traditional equity compensation model (extending the exercise period from 90 days after termination to 10 years after), and I was wondering how that has affected Amplitude's growth? Was it harder in your early growth stages, or did it present difficulties as you matured, or has it been smooth sailing the whole way?

    • SS

      Spenser Skates

      over 2 years ago #

      Hi James! It hasn't presented any difficulties at all so far. It means we're slightly more generous with equity and we have to keep in mind that people will be able to hold onto it for a long time without exercising. I'd definitely recommend it as it's a way to standout during recruiting. I know a few people who have interviewed here have mentioned it as a positive.

  • BW

    Brand Winnie

    over 2 years ago #

    Hey Spenser!

    Thanks for the doing this AMA! What is Amplitude's biggest growth challenge today and why?

    • SS

      Spenser Skates

      over 2 years ago #

      The biggest challenge is having people who are evaluating other solutions know we're a possibility too. Once someone engages with us they "get it" pretty quickly and we're good at turning them into a successful long term user of analytics. The hardest part is to make sure we're in the conversation in the first place.

  • AL

    Arsene Lavaux

    over 2 years ago #

    Hi Spencer,

    Thanks for doing this AMA. I only have one quick question.

    What's the most artistic part of data analytics?

    • SS

      Spenser Skates

      over 2 years ago #

      This still hasn't been solved, but how do you design an interface that is sufficiently flexible to answer questions for power users but not overwhelming to someone who's new? It's a really, really, hard problem to get right. We haven't cracked it yet but are getting there!

  • DC

    David Chen

    over 2 years ago #

    Hi, Spenser!

    - Even though Amplitude's users tend to be companies (rather than individual users), are there aspects of your product that you use to drive your own decision-making processes in shaping Amplitude?
    - In your opinion, what are the most important qualities to look for when hiring data scientists?

    • SS

      Spenser Skates

      over 2 years ago #

      David! Finally getting back to you after two months :) Thanks for letting me stay at your place in New York! To answer your questions:
      - We use Amplitude a bunch internally to understand two things: feature usage and what types of users there are. We're actually rolling out a new user interface based on the common workflows that people perform.
      - Data scientist isn't a well defined role in many places yet. The biggest thing I'd look for is resourcefulness- can someone figure out what the right problem is to solve, solve it, and get something done without a lot of guidance.

  • HQ

    Hila Qu

    over 2 years ago #

    Hi Spenser,

    Very excited to your AMA.

    1) Read the article about startup compensation, applaud you for taking initiative to drive that change. If someone is new to start up world receives an startup job offer and his company didn't offer so much information as Amplitude does, how should go about understanding and evaluating equity piece of the offer?

    2) What do you do to relax and re-charge?

    • SS

      Spenser Skates

      over 2 years ago #

      1) Ask about a specific scenario by saying "if the company sells tomorrow for X, what does that mean for this equity?" Sam has a good way to do this: http://blog.samaltman.com/employee-equity You can also let them know that you expect it because companies like Amplitude/other competing offers you have are doing it.

      2) Good question! I love reading history on Wikipedia. Sounds crazy but you can spend forever there.

      • HQ

        Hila Qu

        over 2 years ago #

        Very practical advice on 1), and totally unexpected but yet fun answer on 2). I guess you can balance your brain a little bit by reading history :)

  • GB

    Gautham b

    over 2 years ago #

    I'm just a beginner in the field of analytics. What resources would you recommend for me to start on the path to becoming to become a good data analyst? What core concepts do you think I'd have to know?

Join over 70,000 growth pros from companies like Uber, Pinterest & Twitter

Get Weekly Top Posts
High five! You’re in.
SHARE
51
51