You wanted to be there, you just didn’t know it. The Growth Hackers Conference was last week and it was awesome. The conference was way sold out and unlike most it was totally worth the investment! Thanks to the speakers for focusing on creating value for the user through the product – it’s not about spam or zombies or farm animals, it’s about creating value!
Thought I’d share my notes for all the growth teams at startups in NY, LA, London, and everywhere else who didn’t get to attend.
Were you there? If you posted your notes or blogged about the conference please leave a link in the comments!
Chamath Palihapitiya (Facebook growth, ICQ etc), Keynote: “How we put Facebook on the path to 1 billion users”
- It’s not about trying to generate extra-normal behaviors in people. Great products is not complicated, its just hard.
- Don’t shroud your product in complex veneer.
- Do what you think is right; not what you read about or what you get credit for. Otherwise you will fail. The hype (the latter) destroys product.
- Use your own product.
- Do three things: Measure, test, try things, throw out what doesn’t work, do what does.
- A simple framework: Acquisition (get them in the door); activation (get to aha as quickly as poss), engagement (deliver core product value as often as possible – allow it to emerge), virality (don’t worry about this – great products will do the first three, and then this will come – virality alone is not sustainable even if quickest ROI).
- Create a culture where you
- (1) Eliminate ego (be okay w/o short term rewards, detach emotionally, don’t conflate luck & ads)
- (2) Invalidate Lore by being clinical and understanding what’s really happening and why (e.g. gut feeling isn’t useful b/c people can’t predict correctly).
- people had all these theories about why Facebook is working. we tried to invalidate their lore. (the best thing that happens is they shut up, then next time they talk they try to know what they’re talking about.)
- we looked at a broad section of our most engaged users and backed into what lead them to getting there (engaged; to realizing core product value as often as possible). then find an axiom and then you do all of the sneaky viral stuff like getting them to “select all without realizing it”.
- k = spam. ruins the internet. ruins the experience. talk about being in the weeds. there is no context to this. don’t focus on abstractions that destroy long term value. don’t work on metrics in the absence of context of understanding acquisition, getting to aha, and delivering core value.
- core product value is elusive and most products don’t have any. know what you’re building & why. marry things that are nonintuitive for people.
- find a framework that allows you to restructure & redefine the ways that your product delivers value to different markets.
- The simple rule we discovered at FB: “get any individual to 7 friends in 10 days” – got people into a network w/ density.
- growth = engage in the core strategy of the product. ruthless prioritization. pivot around core product value, finding different ways to expose it.
- culture wins every time. don’t optimize for short term. work backwards from what is the thing that uses are here to do? what aha moment do they want? how can we deliver that aha as quickly as possible? you need to understand the core value.
- leaders recruit great people, give them a framework, and create discipline.
- core value was you being able to understand your friends – photos is the core way to deliver that.
10:30-11:15 Sean Ellis (lives in So Cal) (dropbox, eventbrite, xobni, qualaroo), startup-marketing.com, Keynote: “Building an Authentic, Effective Growth Engine for your Business”
- coined “growth hacker”. how to leverage metrics, testing, engineering to drive growth in a lean, creative way (that leverages the value of the internet).
- principles of growth hacking:
- leverage! engineering, user base, apis, networks.
- spreadable experiences into different environment to find people and onramp them into your product.
- obsessive optimization
- at the core of any successful, sustainable growth program is a great product. and that is hard to do. you have to have a must-have product at your core.
- growth = building a value delivery machine.
- Hook Promise –> identify the core of the MHX (must have experience)
- calibrate hook & promise messaging (this part is related to traditional marketing – put the promise in a familiar context that users can relate to)
- optimize funnel to deliver MHX as quickly as possible, as often as possible
- identify must-have users to understand the MHX
- qualaroo: would you be disappointed if you didn’t have access to this product?
- then (open ended) what is the primary benefit you’ve received from this product?
- then multiple choice after you have about 20 responses from multiple must have users, followed by “why did you select that benefit as your favorite?” (oe)
- this helps you understand what the must have experience is, you have insight to find these people and deliver the value in an optimal way
- Goals: maximize the % of people that reach the MXH
- engagement = desire minus friction
- increase the decrease the friction
- ask: what is the primary purpose of your visit today? (to understand intent – then figure out hooks based on intent via a/b)
- uncover points of friction (looking at user test)
- front load the MHX / aha experience as quickly as you can
- built channels to improve habit, loyalty, referrals, sales
- understand why people are dropping out of the funnel (e.g., installing but not using)
- building your growth program
- creativity –> engage users for breakthrough ideas. talk to them to find out how product fits into their lives. it’s critical to understand behaviors that users are already doing (without your product) and then shift it slightly.
- testing – quality of velocity of tests drive the results. you need to test all 10 good ideas, not just one.
- double down on what’s working. replicate it, improve it, accelerate it.
- you sometimes need to triangulate to understand what’s working.
- do qualitative research (including open ended questions on the site via qualaroo) to look for ideas that you can run tests
11:15-12:15 Erin Turner (moderator), Danielle Morrill, Hiten Shah, Nabeel Hyatt (KISS metrics, Zynga, AirBnB) Panel: “Growth for early stage startups”
- understand your users. growth assumes you’ve found P-M fit. what you want is an unbelievable experience.
- the further you go down your funnel the harder it is to understand. so optimize for metrics that are far up the funnel and fairly early in the user’s experience. you want your virality tests to happen at the top of the funnel – otherwise (b/c the funnel gets smaller at the bottom) you don’t have enough statistical power to do cohort tracking & A/B testing effectively – not enough users to really analyze and understand it.
- at zynga we never looked at k. we found out it was about “d1 retention – after their first use do they come back the next day?”
- to start your growth, focus on one number and optimize that. which number? what’s one leading indicator that leads to high product engagement. fb: 7 friends in 10 days. spend a couple of weeks finding out the highest correlation to an engaged user. what’s the one thing they do that leads to engagement? then drive users to that.
- Zynga: find the one number to focus on means that your whole team can try to gather intelligence (e.g., from friends about their apps) to help you benchmark and find different ways to optimize for it
- Kissmetrics: on the other hand… don’t worry about benchmarking. baseline against yourself. baseline against several metrics that you think are important. you may throw many of them away, but you’re looking through your data for a few metrics that represent opportunities (through correlations) – then look for hypotheses about how to optimize the funnel based on understanding users qualitatively about what happens between two steps in a funnel.
- really define success and tie metrics to the stuff that really matters. how to define success? think really big about the big problem you’re trying to solve. then boil it down to a really simple framework and really attack that.
- AirBnB – when you have very few listings you worry about quality, not quantity. there were 100’s of things about quality we could think about. we figured out that the most important aspect of quality was photographed. so we tried to optimize for that, and now we have this huge network of pro photographers.
- beware vanity metrics. it’s not about how many listings we can have photographed, we thought about which fraction of our listings really need to be photographed.
- quality = go to your users and remove their hurdles to help them create quality. (e.g., airbnb realized that great images helped the user but they are hard for most people to take on their own)
- make it simple for your users to push their own distribution into every channel
- cohort analysis data is really important b/c your product will keep changing. but you can’t do effective cohort analysis unless your database treats user action as an analysis.
- “in b2b, sales is actually growth.”
- Retention as a growth engine is about both value (of product) and spammy stuff
- how to create retention? off-line experiences; Airbnb we storyboarded the users entire offline experience because we wanted to associate our brand with an amazing offline experience.
- more on the offline experience – everybody in the company needs to be engaged in talking to users. understand why people who are actually using the product to understand why they really love the experience. how does it fit into their lives? get their love stories.
- focusing on metrics needs to grounded in a qualitative understanding of why you care about that metric. (“metrics should come from a qualitative place, not a quantitative place.”) I focus on retention because it aligns the value of the product and DAUs.
- AirBnb – two sided markets where we treat each side as its own growth engine with different needs. We at Airbnb worry about the fatigue of hosts, and reactivating them, they need to keep opening up their home again and again. how can we get them to recruit new hosts? or share their positive experiences with their friends?
- the moment after somebody signs up is a very powerful moment. at KISS the second after you sign up you’re given options about which path you want to take, right away. Fastest path? More advanced path? Talk to us? THis was such a powerful thing is increasing our activation rate. gives users clarity about what their options are. single biggest thing we did with our product growth. for social networks, its about getting your friends to sign up, because this is a social network and the value comes when your friends are on it.
- A/B is for wimps. of course, you should never release a feature without A/B testing it w/ 5-20% of your users first (even if briefly, depending on the feature. you always need a control – wait until the noise smooths out. make sure you’re doing cohorts). it’s a bad way to setup a culture for the early stage startup – one of asking questions not driving for the product they believe should exist. the A is what you had last week, and the B is what YOU WANT THE FUTURE TO BE, and your testing to see if it works.
- Power of the dashboard – data can align a lot of different parts of the org that speak different languages. have your most important metrics front and center.
- operational goals (the number you’re really trying to move) vs. KPIs/indicator metrics – which may not get you to your operational goal. how does it all fit together to help you understand if what your’e doing is helping you achieve your strategy?
- airbnb – we quantify qualitative data when we see patterns emerge and we get the loops going from there.
1:15-1:45pm Mike Greenfield, Britely advisory board member, Growth Hacker in Residence at 500 Startups, was at Circle of Moms/LinkedIn/Paypal, “The Art of A/B Testing”
- A/B testing can lead to a local maxima if you’re only changing the details (like button color) trying to optimize something small. To avoid leading to a local maximum, you should do holistic testing (e.g. pit two totally different home pages against each other – this helps you make an informed decision about how they’ll perform differently)
- Good to do A/B testing w/ things related to sharing, email subject lines, and redesigning.
- Do split testing when you have something to lose (users = data). At early stages w/ few users you have nothing to lose, and you don’t have enough data to split test anyway, so just change things.
- “Test only if you can get to statistical significance.” (google “split test calculator”)
- “Test only if you can get to business significance.” (only if the resulting optimization it can make a material difference to your business)
- for virality, small changes matter if you’re close to k=1. only test if you can get to that range.
- Build your own (testing) framework once you’re at scale. Until then use optimizely or unbounced, but also store the data yourself. Then you can understand the holistic effect of an A/B test. E.g., how are people behaving differently over time as the result of being assigned a different condition (e.g., in an email split test) over the lifetime of that user.
- bit.ly/artofabtesting (code details for building own platform)
- Emails – Focus on 1 item
- clear subject focused on that item (e.g., why the giants will win the world series)
- body should focus on that item too (peripheral content okay in the periphery)
- clear, big click-through action in the body that requires click through to get the full story
- Signup flows – minimize distraction
- provide context/messaging of what the product is. but don’t make it clickable.
- clear next or continue steps to guide user through the process.
- remove unnecessary navigation. remove the header. don’t let them return to the homepage, etc. this makes an enormous difference.
- social proof face pile (a type of context as to why you should sign up) can result in a 35% increase in signup completions. most important are your friends names that are using the product, not the name of the product. people care about what they know – which is their friends. they will use what their friends, who they know and trust, are already using.
- A/B testing helps your company’s culture & performance.
- fosters a culture where data trumps opinions.
- encourages rapid iteration – it’s fun and exciting to see how things are making a difference.
- improves everyone’s ability to predict when their products will be effective. builds better intuition.
- to get started, practice A/A testing, just to get a sense of what the statistical noise looks like.
1:45-2:15pm Aaron Batalion, “The LivingSocial Story: Always Be Growing”
- How we went from 60 million users
- I can’t do this talk justice with notes, Aaron is an incredible story teller! He has the whole audience focused on the stage and not on their screens – a first for this conference.
- know thy API – spot early changes before other people. use changedetection.com to find changes before your competitors – find out what new features you can exploit (before they get turned off!).
- don’t talk, build. JSIO = just shit it out. don’t talk about what you should build, just build.
- apply growth strategies (leverage what’s available to you) to everything, including hiring for rapid scale up.
2:15-3:15pm Aaron Ginn (moderator), Blake Commagere, Simon Tisminezky, Matt Humphrey, Jesse Farmer, Panel: “Harnessing networks for distribution”
- You can basically only focus on distribution on one platform (fb, google, etc). FB is predictable – they’ll kill you if it makes sense to kill you, unless you’re so big (#1) that they can’t afford to kill you (like zynga they had to make a deal – but they will kill #2 & #3)
- you can’t slap on a distribution network. you have to make a strategic choice about what’s a natural fit w/ the platform. test w/ the platform to see which you can use in a unique way & what works. you want to be first.
- very hard to do both growth & retention at once… but you have to.
- don’t forget about the value of a great email. for most people they hate their jobs and every email is a welcome escape – if its good.
3:30-4:30pm Andrew Chen (moderator), Eric Florenzano, David Lieb, Linda Tong, Panel: “Growth on mobile”
- mobile is hard b/c its hard to iterate quickly.
- find the newest channel then maximize that before it gets tapped out.
- don’t ignore word of mouth.
- paid users – tap joy. need to be able to understand the product in the first few seconds.
4:30-5pm Elliot Schmukler, “So, where should I start optimizing, anyway?”
– build on your strengths. optimize for things that are already working well. it is easier and faster to move the needle there.
– focus on getting already engaged users to do more actions, rather than getting inactive users to do their first action.
– here’s where my battery crapped out. Luckily, Sandi MacPherson posted some awesome notes from this fantastic session on Quibb.
5pm-6pm Jeremy Liew (moderator), Josh Elman, ChenLi Wang, Yee Lee, Panel: “Building a culture & skill set to get results”
– #ghconf on twitter