This is a quick dump of notes taken at the Growth Hacker Conference today, May 3rd 2013.
I unfortunately missed some of the talks (updated with notes from all talks!). Sorry about typos and incompleteness. Videos of the talks should be up on Udemy in June.
Value -> Growth -> Growth Hacking.
You have to create value first, prove growth at a first scale, before trying to do growth hacking.
- LinkedIn -> Indexing profiles in search engine.
- Paypal -> Buttons
- Youtube -> Embeddable Players
Growth Hacking is not only data driven, mostly an observation data. It’s a logical science.
Marketing is dead. Feature design is the new marketing.
Square growth -> The product was so good that it was a magnet and it’s what drove the initial growth
“My job is to add zeros to the dashboard”
Facebook had a growth team and it worked well so now everyone wants a growth team. People in the valley tend to think like that, investors included. It might not make sense in every company. Sometimes a product leader who can think like a growth hacker might be enough.
There isn’t a lot of growth hacking content. There’s a natural sense of timidity, it’s very hard to find good stuff.
Perfect growth team: 1 designer 1 engineer (+ 1 analyst). 2-3 person team.
How did you structure the marketing team at Square? Everybody has different definitions about marketing (and APIs).
Difference: 1) customer acquisition 2) brand marketing. 1) is more important than 2) in the beginning. But the right brand can help A LOT. Real hard question: how do you get the two to work together and prioritize it well enough.
Product/Market Fit -> Transition to Growth -> Growth
He spent a lot of time in the second phase. Examples coming.
This applies to most but not all businesses.
Mark Andreessen -> “Focus obsessively on getting to product/market fit”.
So what is product/market fit. “You know when you’ll have it.” “Products flying off the shelf.”
A lot of luck, inspiration, responsiveness, etc.
As a marketer it’s EXTREMELY RISKY to get in a company before product/market fit. That’s the founder’s risk. Only get in after product/market fit.
- Implement user analytics
- Identify “must have” benefits of your products
- Map benefit to an effective hook
- Optimize business model too
- Optimize conversions
- Hire growth team
Problem: slow growth. We stopped trying to grow. We mapped the funnel with very good analytics and found out that very few people had a good experience with the product.
Asked customers: “Is there anything preventing you from downloading the product?” People didn’t think it was free so no-one was converting. Changed messaging and tripled conversion. (They phrased it: “Download the free version or a trial of the paid version”).
At the end 1000% increase!! They also did this: when people downloaded the free version, they would give them the pro version automatically for 30 days.
MOST GROWTH HACKS WON”T WORK IF YOUR PRODUCT IS SET UP WRONG
- Increase desire
- Reduce friction -> this might be even more important
Asks first open ended question to customers, then drills down with closed choice questions to identify the “must have” of the product.
KissMetrics -> Optimizely -> Unbounce
(ex LinkedIn, now Wealthfront)
- How to build a growth machine
- Set growth goal
- understand drivers of growth
Goals: needs to be measurable. The growth team needs to have a measurable goal and it should be measurable quickly.
Facebook growth goal: “10 friends in 13 days”. Get as many users as possible to have 10 friends in 13 days.
You would have thought that it was monthly active users, billions of users etc.
If they had 10 friends in 13 days there was a very very high chance they’d become active users. The best part is that you can measure it in 13 days.
Impossible to measure MAUs in under a month. Ability to measure the user that signs up today, is going to be very slow: will have to measure users for months.
You need to get results of experiments very quickly.
At Linkedin they had signals to understand if you’d become MAU in just 1 day. Immediately measurable.
Goal needs to be aligned with product and business model.
Make the goal a SINGLE number. A single goal.
To understand growth drivers: work backwards from the goal.
Understand the steps/channels users use to get to the goal. If you have a diagram with all the steps and all the channels that users go through, each box will be optimizable.
Optimize each step and channel.
3 ways to optimize:
Make something simpler / easier to do. Reduce steps / clicks, etc. LinkedIn form went from 10 fields to 4 fields -> 20% better conversion.
If you need all of the data: split the form apart. Break it up into many small steps. (microquestions, 3 bullets and ok.) LinkedIn changed their data entry flow to microquestions.
LinkedIn endorsments: 1 click. As low friction as you can make it.
Give me a better reason to do something.
Make the product reasonate with the users, but something else you can do that doesn’t have to do anything with the product:
- Give aactual numbers of savings or something similar.
- Social proof – these other people that you know are doing this. Tried at LinkedIn in different places. Every time it DOUBLED the number of people that completed the step
- Material incentives: Dropbox give space, you can give discounts, etc.
Underutilized by growth people.
ASK ME TO DO IT MORE OFTEN, MORE PROMINENTLY and IN MORE PLACES.
Don’t expect that the user will find something do to, or that they will do it when you ask them. Keep asking all the time.
LinkedIn growing profile data: added “Improve my profile” button so people didn’t have to go change the data. Took it one step further and now they pop up blue boxes with questions and other stuff to do
When you have a user that has done action X once, ask them to do it again.
Don’t push this too far: you can vary how often you show it and vary what’s there. Test both. Vary what’s there a lot. (eg. people you may know, always different people which is suboptimal in the short term but optimal in the long term to avoid blindness.)
LinkedIn’s goal was signups that imported an address book.
Constantly retry things and experiments that have failed, but have other people work on it.
(LinkedIn – Circle of Moms – 500 Startups)
Circle of mom grew on Facebook to 6m users. The Facebook channel had dropped sharply. Traffic down 50%. 1m active out of the 6m.
Had an email digest that sucked.
Email was extremely successful for them.
Lots of steps
- inbox delivery – getting into people’s inbox. MAKE SURE EMAILS GET DELIVERED. Sendgrind vs Return Path. Simplicity vs effectiveness.
- One item focus! In your email have one prominent piece of content + customized subject based on that.
- Quality – initially all user generated content, so quality kind of sucked. Hard to remove bad content and surface the good one.
- Subject optimization – Not a small project. They built their own full A/B testing product.
- Winners and losers – distinguishing
- Personalization – they build a predictive model for each user and email, then run tests and pick the email which has the best odds of converting.
- Frequency – test different options holistically on a huge number of different stats. Not just the clicks. If quality was high, they could do up to 4-5 times per week.
Quora has a very high quality email flow.
(Goldman Sachs -> Stanford MBA -> Facebook (end of ’05) -> Executive in Residence @ Greylock)
Goal at FB was growth. MySpace was the enemy.
3 obvious things: look at the data (people navigating from mobile and other countries, out of college, high school), find easy wins (carriers were seeing increases in traffic to fb and myspace), BD for mobile growth
Really understand users. Who are you building for?
- pay as you go internet access
- screen size
Lyft is probably the fastest growing local marketplace
growing faster in LA than SF.
Strategy is very different than what is usually talked about.
Future of growth hacking is being successful at branding as well as customer acquisition.
True brand building: Lyft uses pink mustaches and fist bump.
(Uber: multi-time entrepreneur, Sidecar: venture capitalist and entrepreneur, Lyft: people with transportation experience).
Secret: true understanding of the customer.
Service originally for women, because they felt unsafe with cabs: -> pink: feeling of safety and fun, even when a random car picks up you with a person you’ve never seen before.
It wasn’t a marketing play.
After you get in the car it’s super super awkward. Fist bump to break the ice.
That was not the brand.
The brand was what was underneath: the community.
In order to make this work we need thousands and thousands of suppliers, and we have to vet and interview every single driver. VCs freak out at this. BUT, they were able to create a community out of that.
More rigorous insurance and background checks than the taxi industry. More regulation out of the regulation.
User life cycle.
Before they buy -> First purchase -> Repeat purchases -> referral -<
In LA pink mustache is not an advantage anymore. So what worked in SF didn’t work in LA.
LA grows faster because of the understanding of the life cycle.
I hate change -> what is this -> what was the name of that app again? -> oh dude you should totally check this out -<
advertising -> conversion -> retention -> virality
Customer development applies to growth too: get out of the building.
LA: pitch people in the streets with a pink mustache all night for 3 hours.
After that mobile ads, which out converted every body because he spoke with all the customers and knew them really well.
(Twitter, Director Growth Engineering)
- Multi channel
- Quality metric – try to quantify quality or find other metrics that can give a sense of the quality
Teams: full stack engineers, problem solvers, metrics driven, #shipit
The products you build are really only as good as the people who build them: hire world class teams.
50 people on growth team: product managers, designers and engineers
If growth team builds a product which has success, who manages it? For now growth team still manages everything they build.
Keep track of all metrics, see the whole picture (eg maybe some feature increases follows but decreases tweets).
Is the success of a feature very country dependent? Usually yes.
Always slice metrics by country.
(Product Manager, YouTube)
Video and media is accelerating much faster than anyone could imagine.
Fundamental shift of how people think about media, it’s now a two way street.
YouTube doesn’t have a top of the funnel problem. 72h uploaded every min. 1b monthly actives.
But TV is much bigger, YouTube is missing something.
For a long time they optimized around number of views.
They then shifted to watch time. Very difficult cause everyone really understood views but not watch time. Partners, users, etc.
Choosing the right metrics matters a lot. Painful to change.
Started a forum called YT stats, where executives managers and engineers would talk about all the metrics, and anyone could ask questions?
12 to 15 people meeting, went to 150 people. All the important decisions made there.
Culture change starts with great questions.
YT has teams siloed by feature (watch, homepage, etc.)
So they started a growth team, horizontal.
Failed. They reorganized everyone into horizontal, everyone is on the growth team.
Align yourself around the user.
It doesn’t need to be complicated.
- Goodreads: how do you make people write reviews?
They originally put them in chronological order. They then realized English majors students were writing most of the reviews. Many people will read the same book at the same time.
If any of your friends wrote a review, it will now show up first.
Now people know that their friends will read what they write.
5x more reviews.
Invites went out! People write a review and want to show to everyone how smart they are. People who are invited read the review immediately.
All they did was change the order of how the reviews were displayed.
- Facebook photos: conscious decision not to compete with photo sharing sites.
Everyone thought that having the original image was the most important thing. Facebook was throwing away originals! But they did do tagging. You get an email saying you were tagged.. There is no way you’re not clicking that email.
Lyft and PayPal have different flows for merchants and users because they have different teams. People think of themselves like people so there should be one flow and one teams.
Growth is not just one simple thing. It’s manu small and complex things.
Volume is a competitive advantage: more volume means faster optimization.
Growth is not only about the onboarding flow.
Neither about gamification.
Gamification should only be used when it provides utility.
Use existing members who enjoy your product to drive growth. 70% of members should come from there.
Inviting people from the onboarding flow has limited utility.
- failure (all the time)
takes years to get growth. no overnight successes.
All of the following notes are by Dinesh Gomes
Greg Tseng, Tagged
Growth v/s Engagement
Tagged as a case study:
Launched feature (newsfeed)
Learned an insight, people didn’t want to network, wanted to meet new people
Theme of Social Discovery
Itereated and learned how to position this
Pre-product Mket fit learnings
- Sustainable value only comes from REAL user engagement
- Focus on engagement
- A STARTUPS FIRST GOAL: PROVE PRODUCT MKT FIT
- Only do growth to help achieve Product Market FIT
Ensure enough sample size when doing A/B Testing
(exception is when the product is the network, like messaging)
Viral marketing is a tactic not a religion
Noah Kagan, Mint, Blogger
How to make your first dollar
- Asked question on shoe lace v/s spending half an hour for a dollar
NO ONE WANTS TO PAY TO TIE SHOES
- Do each of you have a dollar in your pocket? Do each of you have a problem in your life?
Try the coffee challenge> app sumo.com/coffee
Go to starbucks and try to get a 10% discount on your next cup of coffee
The #1 problem with starting a Biz? FEAR
HOW DO YOU VALIDATE THAT PEOPLE WANNA USE YOUR PRODUCT
ie. Will they pay me? Will they use it?
Eg. How would you validate AirBnB?
A. Put an ad on CL
Eg. How would you validate square
A. Go to a farmers mkt , put a fake master card sticker, photograph the card, and tell your friend to purchase the pmt
If people will pay you for that information, then you have a business
Josh Elman, Twitter, LinkedIn
Habits – from Inception to Adoption
Noticed that his talk is listed here
Not just what do you need to solve to ? but…
Who are you solving for
How is it solved,
How often is it solved
Case study – Twitter signup: Autofill had “I don’t get twitter” before “I don’t get drunk” – how to fix?
Learned that Twitter isn’t for everyone. Twitter turned out to be “whats happening with people and orgs you care about”
“Come join me because <This is the value I bring to you>
The only metric that matters: How many people are really using your product
- At the frequency you expect
- with the confidence that they’ll convert
- They will pay (someday)
Figure out who comes back
State diagram of twitter users types
Curious -> New + + -> Casual
- Define your purpose
- Incept the right understanding of product
- Teach people how to use as they adopt
- Create real habits and understand how to keep them
- Define & monitor only what matters
- Look for A-Ha moments and design around them
Thanks again to Dinesh Gomes for providing the notes on the talks I missed.