Marchin Treder, founder of UXPin.com has published a book recently, "UX Design for startups". While not being a substitute for a good book on UX in general, it gives you a decent overview of relevant challenges that a startup may face, as well as high level overview of UX design. There are also some interesting notes on startup analytics, although nothing that Pirate Metrics wouldn't cover.

Following my general principle of converting books/articles into actionable items, I took some notes while reading this book. Have a quick look at those and decide for yourself whether it's worth reading the book.

If you interested in this field, and especially if you are just starting out, I highly recommend the "Human-Computer Interaction" course from Stanford/Coursera. It's awesome, and it's free. Best of all, prof. Scott Klemmer, who runs this course, is damn passionate about UX field.

Book is here: http://uxpin.com/upload/ux-design-for-startups-marcin-treder.pdf

Notes:

Intro

• Starts with a notion of business canvas, "getting out of the building"; directly mentions Steve Blank
• Constant iteration in "Customer -> Problem -> Solution" cycle

  • Define your value proposition in one sentence, that will include all of the Customer, Problem and Solution.
    “For people who are trying to design products with great user experience and are having problems with documenting their ideas quickly and clearly and sharing them with their teams, UXPin provides an online, fully collaborative app that helps them to go through the UX Design process together with their teammates.”
  • Each part of the C-P-S is just a hypothesis, that needs to be verified

 

Get to know your users

• UX design is inherently human-centric

  • it means you need to interact with your users all the time, not just sit and design

• Ask question for each of the CPS:

  • Does your design address your target group?
  • Does it solve their real problem, or a made-up one?
  • Does it solve it well?

• Go out and meet your users

  • they spent two weeks, having 3 meeting per day with users to figure their questions out

• Guerrilla User Research

  • refers to Steve Krug / Don't make me think
  • find your users and talk to them, show them your work

    • prepare a testing script

      • start with a bit of a story - provide context for research

    • prepare people

      • treat them like experts and say that results don't matter, their opinion is what matters
      • encourage them to speak aloud while they are doing something, ask "What are you thinking right now?" if they go silent

    • takes notes during the test

      • use only short catchwords

    • analyse results immediately
    • turn notes into actionable items

  • use Skype for interviews - people are more happy to meet strangers over Skype than over real coffee

• survey.io for online user survey
• Set up a feedback forum - and be the most active participant, answer to what people have to say, engage

  • User Voice

 

Efficient Design Techniques

• Use whatever tools appropriate for the given situation, don't get stuck in arguments
• Strategise - to pick the right tools (i.e. wire framing, sketches, interactive prototypes etc) - ask yourself

  • Are you in the rush or do you have some time that can be used for design? Time/Quality tradeoff.
  • Accuracy level - are you designing a complex product for regulated industry? Will you need to produce lots of docs?
  • Past experience - which design techniques work best for your team? Are they able to proceed without lots of questions? Were they happy with the result?

• UX designer shouldn't just be a "wireframer" - but must be able to uncover people's problems and design pleasurable, seductive and inspiring solution
• Paper prototyping

  • Don't spend too much time on initial prototypes - makes it easier to criticise & throw them away - for this, paper prototyping is the best
  • don't over design your paper prototypes
  • can have pre-printed elements, given out to team of non-UX people to encourage them to design UIs during the session
  • always do paper prototyping before starting on more expensive prototyping initiative, such as digital wire framing/interactive wire framing.
  • OK to use only paper prototyping - if time is limited & your team is happy to deal with scrappy documentation

• Wireframing

  • low-fidelity representation of the design
  • Should show:

    • Main groups of content
    • Structure of information
    • Flow of interface - description & basic visualisation of on interactions

  • Should contain a representation of every important piece of the final design
  • Don't spend too much time on elaborate details such as choosing right icons or colours

    • use only three, at most four colours white, black, gray, and blue for links if you must

  • Write descriptions/comments/notes for elements when necessary

 

Growth and design hacking

• Always measure things!

  • Economic metrics

    • Pre-revenue, no traction

      • How many users start to use product on regular basis?
      • Sign-up funnel conversions, sans the payment step

    • Pre-revenue, traction

      • continue to track engagement of users. Test whether they are willing to pay, if possible.
      • continue to optimise sign-up conversion funnel, sans payments

    • Revenue, no traction

      • optimise sales funnel
      • track number of paying customers
      • track # of people leaving the product(s) - churn rate
      • analyse your CPA (cost per acquisition)

    • Revenue, traction

      • track recurring revenue, per month
      • LTV - life time user value - how much on average do you earn per user in total, given the average length of paid product usage
      • ARPU - average revenue per user
      • track # of people leaving the product(s) - churn rate
      • analyse your CPA (cost per acquisition)

  • Behavioural metrics - to track specific actions of users

    • For every new feature you launch:

      • figure out what's the main use case
      • figure out how to track whether users are able to follow this use case
      • for example, for a "sign up " feature you could track:

        1. conversion rate
        2. # of successful sign ups
        3. # of various types of errors

    • Don't just blindly track everything, but only the metrics that validate/refute your design hypothesis

• Be aware of vanity metrics, such as number of visitors / page views - those need to translate to revenue

  • vanity metric is something that makes you "feel good" about it, yet it's of no value to your business and you cannot act on that data
  • constantly remote the vanity metrics from your analytics dashboard

• For product development, set goals based on metrics, and track your your progress with every release / weekly
• Metrics need to be actionable - what do we need to do so conversion rates goes up? Or more people convert from trial to paying?
• Keep doing qualitative testing, such as in-person interviews regularly to uncover UX bugs.
• Tools

  • Analytics

    • Google Analytics
    • Kiss Metrics
    • MixPanel

  • Usability testing

    • Silverback

  • A/B

    • Visual Website Optimiser
    • Optimisely

Get it optimised

• UXPin launches new version every couple of days, figuring out what to keep and what to kill
• Launching a product is not end of UX designer's work, but only the beginning of measurement phase so you can learn how to optimise the design
• Don't get focused on the release, hoping that everything will magically work - instead be ready to measure and act, since usually things don't change for good, they can even become worse
• Don't chase the feeling of relief you are hoping to achieve with release, instead focus on the goals
• Consider a feature only working when people start using it, not when you develop, test and deliver it
• Example of adding a "viral loop" for earning days of free trial by referring friends - this actually caused the # of conversions from free trial to paid to drop, since people were postponing the conversion decision indefinitely. Wouldn't be able to tell that if didn't have measures in place and pre/post release data.
• So, when launching something, think how you will measure the success

Tools - see PDF for all sorts of links

• Wireframing tools
• Diagraming tools
• User feedback tools
• Analytics
• User session recording tools