Leave a comment
Get the GH Bookmarklet

AMAs

Edith Harbaugh is CEO & co-founder of LaunchDarkly, the leading feature management platform. Companies like AppDirect, GoPro and CircleCI use LaunchDarkly to control who gets what features. LaunchDarkly serves 20 Billion features every day for early betas and long term subscription control. LaunchDarkly has raised over $30M in funding. 

Edith was product and marketing at Tripit for Business and EasyBloom, and grew both from $0 to $1M in revenue in under 18 months.

She's also the host of To Be Continuous, a podcast about continuous delivery and software development.

She's also shared tactics for both starting a company and effective software development in this First Round interview.

Ask her anything about category creating, convincing people to buy vs build, SaaS price testing, or how to build software.

You can follow her on Twitter: @edith_h.

She will be live on Jan 23 starting at 930 AM PT for one and a half hours during which she will answer as many questions as possible.

  • TC

    Ted Carstensen

    11 months ago #

    Hey Edith,

    I'm hoping you can share some of your tactics around early startup marketing and thought leadership. What tips do you have for first time founders who want to get their names out?

    • EH

      Edith Harbaugh

      11 months ago #

      Effective speaking is a tactic to build market awareness, get customers, and lend legitimacy to your company. But when we started LaunchDarkly were a two person startup with no brand name — no one was clamoring for my TED talk*. In fact, one of our angel investors said “Edith, you’re a huge nerd – you’re never going to be the tap-dancing CEO on stage”. Which I took as a challenge.

      1. At the beginning, be willing to speak anytime to anyone, to build your profile / practice speaking. I gave a talk at Runway Accelerator to the only two people who showed up — but they were interested. With more practice, I became a better speaker & a more known quantity.

      2. Meetups are often looking for speakers who aren’t giving a product pitch. Everyone starts somewhere — I remember Drew Houston talking in 2009 to 25 attendees on marketing tactics. 

      3. Apply, apply, apply! Today I have a 1:3 ratio of talks I apply to vs acceptance. When I started it was more like 1:10 (or 1:20). I made a spreadsheet of all relevant conferences. I wrote out an abstract, long description & bio, and took 4 hours to apply to all relevant conferences, with appropriate tweaks to the abstract and description. Almost all turned me down, but some accepted, I got more practice, & could then apply to a new batch. Doing it as a batch is less work than sitting down for each conference with a cold slate. 

      In short, think of being a better speaker & getting engagements as a growth process. See what talks are getting accepted, practice them.

      *still no TED Talk, 4 years later, to be honest, though people think I'm a good speaker now and do ask me to give talks.

      3 Share
  • MN

    Muhammed Nasrullah

    11 months ago #

    The developer community is notorious for under-valuing their time and responding with "oh we don't need you, we can do that in-house". How do you overcome that obstacle as a DevTools startup?

    • EH

      Edith Harbaugh

      11 months ago #

      Combatting buy vs build – I actually love buy vs build because it means that the customer believes in what you’re doing. If they say “we don’t believe that we ever need that”, there’s no conversation, and no sale. For LaunchDarkly, our customers have to believe that feature toggling is a best practice they want to use. If they’re not on board with feature toggling, they’ll never build in house (or buy our solution). For buy vs build, I try to understand WHY the customer wants to build:
      Pride – “We can do it better”
      Cost – “Why would I spend $299/month on that?”
      Core - “This is too important to trust to a vendor”
      Customization - “No off the shelf solution can meet our needs”

      We actually love it when our customers try to build in house - after they’ve tried building it in house, they realize how much time/energy goes into maintaining a system like ours. We “publish our blueprint” - http://blog.launchdarkly.com/buying-vs-building-a-feature-flagging-system/ on how you should make this decision.

      4 Share
  • JF

    Javier Feldman

    11 months ago #

    Hola Edith

    Can you talk about any experimenting with pricing you'll have done that had any significant impact?

    Also, I was surprised to see that, unlike what appears to be standard elsewhere, you'll don't offer a discounted annual pricing option upfront as a default and make people contact you about this. Could you talk more about why you'll went down this "monthly only" path?

    • EH

      Edith Harbaugh

      11 months ago #

      I was product manager and marketing at EasyBloom, a gardening sensor that we started selling in November, 2007 – not only the beginning of winter, but the start of a global recession. Original plan was to sell ~5,000 units in that first month. Actual sales: less than 100. As it pretty much literally couldn’t get worse, our CEO let me run many marketing & pricing experiments.

      1. Price: we started with a price of $79, as that gave us a good gross margin. Dropping to $49 made a significant difference, as then we slotted into a different mental bucket of spending. Further drops to $44/$39/$29 didn’t make enough of a difference.

      2. Shipping: We had our own drop shipper, as well as listing on Amazon. I bought Google AdWords campaigns. Pointing the campaigns at Amazon did much better (even taking into account Amazon’s cut) as it significantly reduced the buying friction. Forcing people to enter a credit card on our own site took us from a semi-impulse buy to thought. Also, because we were low-volume, shipping for us was $12 unit, Amazon was much cheaper.

      3. Paypal Checkout: We were selling a SaaS upsell of unlocking “fertilizer sensing” as a $2/month add on. Having a PayPal checkout button depressed our conversion. When I asked our users why, they said PayPal made us seem fly by night (this could have changed since 2007).

      2 Share
  • PH

    Pradyut Hande

    11 months ago #

    Hey, Edith. Glad to have you here!

    For a new SaaS company trying to compete with established players, a good product backed by competitive pricing can become a short-term differentiator. In such a situation, offering freemium accounts becomes an avenue of rapid client acquisition, and the hope is that these accounts will later move to a paid account. However, the conversion rates for these are quite low. How do you recommend this be tackled?

    • EH

      Edith Harbaugh

      11 months ago #

      Hi Praydut, great question! You have several assumptions here that you should validate.

      1- Competitive pricing can be a differentiator. If a product is a commodity, yes, underpricing can lead to acquisition. However, there are many products where under-pricing can be a deterrent to some customers. IE, I have no desire for “cheaper LASIK”. Positioning yourself as “low cost” is a very different marketing stance than “premium, stable, elegant”. Here’s a great talk by Michael Dearing on Pricing & Quality (with transcript): https://www.heavybit.com/library/video/harrison-metals-michael-dearing-on-pricing/

      2- Freemium can lead to rapid acquisition. There are many other costs besides money (switching costs, learning curve, network effects). Most apps in the iOS & Google app stores are free, but struggle to get any traction. At LaunchDarkly, we would not easily switch payroll providers even for “free” as our team has all set up their direct deposit, and it would be a burden on the team to re-enter, us to re-validate.

      3- CAC & COGS can be balanced on a premium base. You assume can segment out some functionality such that customers will transition from free to paid, and that these paid customers will pay enough to make your business a going concern. You need to keep a very, very tight eye on both your ongoing costs of goods (including support!), one time customer acquisition costs, and make sure you’re not ending up massively in the red. A mistake I’ve seen people make is only including the one time CAC and neglect the support/infrastructure costs. Here’s an article on Evernote & Freemium https://www.cmswire.com/digital-workplace/evernote-falls-prey-to-the-freemium-model-catch-22/

      4 - Upsell - finally, that you can find some features that premium users will pay for, but freemium users can exist without. Typical ratio of conversion is 2-4%.
      I hope this helps you decide what market you’re in, and whether freemium is a viable tactic for you.

      4 Share
  • MN

    Muhammed Nasrullah

    11 months ago #

    Heya Edith, how does a young, unknown startup ask well established players to add them into their production pipeline or into their production app?

    • EH

      Edith Harbaugh

      11 months ago #

      Getting early customers to use a young unknown startup (particularly in production) was one of our biggest early challenges. LaunchDarkly is core to what software a company does – they need to really, really trust us. At LaunchDarkly, we “laddered up”. I started with my former coworker startups who would use us for small, limited use cases. Then, when they were comfortable, they would start using us for more essential tasks. Flite: http://mechanics.flite.com/blog/2014/10/14/feature-flags-at-flite/ was one of our first users. After we had our first customers, it was easier for slightly larger others (who didn’t directly know us) to trust us as we had references. Now we serve 20 billion features a day, but it started with literally customer by customer groundwork. In the early days, we’d also give our customers a private Slack channel which was an “always-on” hotline directly with us. Our early customers LOVED this, and are still a bit sad when we got big enough we had to discontinue. Here’s an article I wrote on getting our first ten customers: https://medium.com/lean-startup-circle/how-to-get-your-first-10-customers-81170a5e65a9

      2 Share
  • SE

    Sean Ellis

    11 months ago #

    Hi Edith, thanks for doing this AMA with us. Any tips for increasing testing throughput with a mobile app? App Store approvals on iOS really seem to slow down the testing tempo.

  • GH

    Glen Harper

    11 months ago #

    Thank you for joining us today, Edith.

    I really like the idea of having the team write a blog post at the end of the week.
    Have you discovered any unintended/unexpected benefits of doing this?
    Anything you've done to tweak this process since you started it?

    • EH

      Edith Harbaugh

      11 months ago #

      In the early days of LaunchDarkly, we required every new employee to write a blog post about their first week. This was my Tom Sawyer way of trying to make blogging fun. It got new employees familiar with how to use our blogging software and so if they ever wanted to blog again, they knew how. We promised them that they could write whatever they wanted, we would only edit for grammar. The early entries were amazing & eye-opening to see what our newest members thought of the company, as your culture is only as strong as your newest members. I also heard that it helped potential new employees get a feel for what it was like to work at LaunchDarkly.

  • EH

    Edith Harbaugh

    11 months ago #

    I'm out of live time, but I'll be back to answer some more! Thanks for all the great questions.

  • SB

    Schuyler Brown

    11 months ago #

    Hey Edith, could use some advice for a fellow early stage startup. When did you feel like you'd found product/market fit? How defined was your buyer persona/target prospect? How efficiently could you qualify leads? How well defined was your acquisition funnel? When did you feel ready to invest more aggressively in each acquisition channel?

    • EH

      Edith Harbaugh

      11 months ago #

      Schulyer, great question about early stage product/market fit. I think product/market fit is a continual process, but there are some distinct milestones I remember.
      1) Getting anyone (anyone) to install our SDK for the first time felt good, but they were people in our own network. However, other startups couldn't even get their own friends to use them. Even a free product has a cost!

      2) When someone signed up on our landing page, I called them and asked what they were looking for, and we had exactly what we wanted. After multiple in-person visits they became a customer.

      3) When someone international went start to finish from signing up to installing our SDK to buying w/o us having to visit in person, I felt like we had achieved significant momentum.

      • EH

        Edith Harbaugh

        11 months ago #

        We are always simultaneously iterating on both our product and our marketing, which really helped us achieve product/market fit.

        Originally we thought we’d be all bottom-up like Atlassian or TripIt. Instead, our buyers turned out to be Director/VP of Engineering, who wanted more reassurance of our stability, performance. When we felt pulled by enterprise, we built features like audit log and SSO, and targeting enterprise shifted how we marketed.

        At the beginning, I talked to everyone – our acquisition funnel was more like a cylinder. As we progressed, we could get better at qualifying leads based on how they matched our target persona (now that we had one), and could actually build out distinct funnels.

  • AL

    Arsene Lavaux

    11 months ago #

    Bonjour Edith,

    How growth and product iterations should be tied?

    Merci!

  • AA

    Anuj Adhiya

    11 months ago #

    Hi Edith - so cool to finally have you on!

    Can you talk about any best practices with respect to running A/B tests with feature flags?
    Any considerations/gotchas that aren't obvious?

    Also, are you'll considering any integrations with Google Optimize?

    • EH

      Edith Harbaugh

      11 months ago #

      A feature flag is a way to run different code paths (or branches). It could be a simple as an a/b, or you could have a/b/c/d (etc) multi-variate paths. The great thing about feature flags is it allows you do testing of application functionality, not just superficial website changes. For example, you could do a two step vs three step registration, or experiment with putting different fields (name, credit card) at different stages. This is very powerful for lead forms or checkout flows.

      Tips:
      How much traffic does a flow get? If you’re a B2B app with 100,000 monthly active users, not every flow will get enough traffic to get meaningful results.
      Do users rely on a certain flow? If you have power users who are used to a certain sequence, a/b testing can be very disruptive for them
      a/b testing can also be used for making sure that a new system works as well as the old.

      Google Optimize – great idea! We already integrate with Optimizely.

      2 Share
  • JP

    John Phamvan

    11 months ago #

    Great to have you here, Edith.

    a. What tools is the team using at LD for experimentation & analytics right now?
    Has the team added any tool(s) to the stack recently? If yes, why?

    b. As the CEO what data/metrics (specifically from a product use perspective) are you most interested in looking at daily/weekly/monthly? What reporting tool do you use to get this data?

  • DH

    Dani Hart

    11 months ago #

    Very excited to have you here!

    What is LaunchDarkly's biggest growth challenge currently?
    How are you tackling this?

  • TN

    Tri Nguyen

    11 months ago #

    Hi Edith

    Does category creation necessarily involve giant breakthroughs or can they come from incremental innovations? If yes, can you provide examples of companies that have done this?

    On the other side, can you point to any examples of any companies that created categories, especially through the use of some unusual/breakthrough business model?

  • MD

    Mark Anthony de Jesus

    11 months ago #

    Hey Edith

    a. I noticed that you'll had a secondary "get demo" CTA on various pages.
    Why not just have the primary "Start Trial" CTA?
    Under what circumstances do people choose to get a demo?

    b. Why is your free trial for 30 days long and not, say, 15 days or any other time frame?
    Has there been any experimentation around trial time you've done that threw up any key learnings?

    • EH

      Edith Harbaugh

      11 months ago #

      We have both a Get Demo & Start Trial as we found that our buyers were Engineering Leadership (Directors, VPs) and our users were developers. Though you can get started using LaunchDarkly in less than five minutes as a developer, a VP Engineering wants to ask questions about how LaunchDarkly fits into their org and helps them overall.

      At a prior company, we heavily tested free trial length, and found it’s very much product dependent. We had a theory that with longer trials (60 or 90 days), we’d get more conversion. In fact, the longer trial, the less likely people were to convert. Our theory was that with a longer trial, there’s less urgency to evaluate in a timely manner.

  • DO

    Danielle Olivas

    11 months ago #

    Hi Edith,
    It seems everyone has a podcast these days.
    You have been doing it for a while.
    What benefits have you gotten from doing this - obvious and expected to the unexpected and not-so-obvious?

    • EH

      Edith Harbaugh

      11 months ago #

      I’m cohost of “To Be Continuous”, and enjoy doing it. Benefits:
      -have a friend to talk software and startups with
      -excuse to hang out and chat with interesting guests like Jocelyn Goldfein, Martin Casado, Kris Gale, Kevin Henrikson http://blog.launchdarkly.com/to-be-continuous-continuous-delivery-and-mobile-development/ still one of my favorite episodes on how to do weekly releases on mobile.
      Unexpected –
      Many VCs say they listen to it, which surprised me. I always assumed I’d have only technical listeners.
      -international reach –when I was in Japan, CTOs said they listened to the podcast!
      -Microsoft asked us to broadcast live from MSFT Build, their developer conference

      2 Share
  • JD

    James Dunn

    11 months ago #

    Hi Edith,

    What do you think was the biggest difference between marketing at TripIt was vs at LaunchDarkly?

    • EH

      Edith Harbaugh

      11 months ago #

      Tactics that worked at TripIt didn’t work at LaunchDarkly (or vice versa) but it was good to have a toolbox to draw from.

      Content Marketing never really worked for TripIt – people have been writing about travel since Marco Polo, every newspaper has a travel section, people love to blog about travel. OTOH, with LaunchDarkly we could see immediate results when we would put out valuable content.

      TripIt could grow virally – people would invite their friends to see their trips.

      Conferences weren’t cost effective at TripIt, as our ACV was $50/month. LaunchDarkly is B2B, so meeting our customers in person is important for both them and us.

      Press was easy for TripIt – reporters travel, they instantly got why it mattered. LaunchDarkly is a much harder thing to explain to mainstream press.

      For both, we focused very much on Net Promoter Score – I firmly believe that happy customers are the best marketers.

      3 Share
  • SK

    S Kodial

    11 months ago #

    What are the biggest lessons you've learned about fundraising from the early rounds to the last one?
    Is there something you wish you knew before any of those rounds before you decided to go for it? Anything yo'd have done differently?

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

Get Weekly Top Posts
High five! You’re in.
SHARE
44
44