Tuesday, 5 March 2013

NSConference 5: Day Two

Thriving in an App Store World

Michael Jurewitz @jury

Jury used to be an Apple Developer Evangelist. He is now Director of Product Development at Black Pixel.

work well with apple

  • need to be looking forward - focusing on future (tech/features/hardware)
  • stay on the radar of apple contacts
  • Apple looks at devrel as “animal husbandry” :-)
    • Apple wants to sell devices & make customers happy
    • so apps should feed into that
  • Apple has laser focus on future and simplicity
    • expecting & embracing change
    • they force themselves to keep going forward
    • won’t look at apps that don’t fit in to that
    • must stay current
  • no secret agenda, but often hard choices
  • can get featured by taking advantage of new OS features
    • but it’s a time limited offer…
  • example:
    • Eventbrite getting featured on Passbook feature page increased new user sign up by 664%…
    • got in right at the beginning – only 15 apps on the feature page
  • “if users aren’t upgrading their OS, they probably aren’t buying your software either” - Wil Shipley
  • Gatekeeper vs App Store: e.g. Kaleidoscope
    • if you buy Kaleidoscope on Mac App Store then you can also download direct
    • app will notice that it’s already been purchased
    • will check receipt and unlock automatically
    • useful for sending customers beta builds when testing fixes
  • file bugs
    • also for requesting to open APIs that are currently private
  • localisation:
    • Apple sales in China increased 400% last year
    • Germany has a strong software market, even though it’s limited in size
  • accessibility creates loyalty

properly value your work

  • Jury did some research over the past weeks…
  • top paid apps have much lower mean & median prices than top grossing apps
    • mean: paid $12.46 vs grossing $49.13
    • median: paid $6.99 vs grossing $29.99
  • two separate markets going on here
  • top grossing has:
    • 200% more Finance apps in top grossing at the moment
      • because it’s tax season at the moment in the US
    • Utilities drop by 25%
    • 22% more Games
    • 50% more Graphics
    • Social drops in half (and would be nothing without Tweetbot)
    • No Weather apps
    • 3-5 Business apps vs none in Top Paid
  • four free apps in Top Grossing
    • in-app purchases
    • freemium can work
    • talk to Kevin Hoctor who did in-app purchases in Moneywell Express
  • cheap apps get downloads, but higher priced apps pay the bills
    • which would you rather have…?
economics 101:
  • price elasticity of demand
  • how does adjusting the price of a product affect the number that you sell?
  • elastic (e.g. social apps):
    • there’s a common price that people expect
    • if you increase the price beyond that, then demand massively decreases
  • inelastic (e.g. drugs… or photoshop…):
    • people will pay pretty much whatever you ask
    • keep increasing the price and maximise revenue
  • demand curves can move due to marketing campaigns…
  • if you double the price, and you lose less than 50% of your customers, you’ve just made money
    • and fewer users means less support costs
    • also increases perceived value
  • don’t be a commodity — charge what your software is worth

price your app intelligently

  • market segment:
    • lots of basic research
    • category prices
    • start research before starting developing
  • research competitors
    • do you have a big enough differentiator?
  • make a guess
  • then try an experiment
  • example: Kaleidoscope 2
    • developer tool segment: avg price $30.11
    • other apps $70-100
    • app is useful but not crucial
    • lots of alternatives
    • guessed $34.99 intro and $69.99 ongoing
    • got various peak sales that affected average
      • great for recouping costs, but not for calculating ongoing revenue
    • evaluation:
      • elasticity = % change price / % change in quantity
      • with a couple of price changes, you can work out the ratio
      • then apply to price changes to predict quantities and therefore revenue
      • (real world demand curves aren’t linear, so elasticities aren’t actually constant)

Talking to Hardware

Alasdair Allan, Babilim Light Industries @aallan

  • Apple’s External Accessory Framework is missing most of the useful stuff
    • rest is protected by Made for iPhone program
    • which is protected by massive ranks of lawyers
    • because Apple want to protect their platform

crazy stuff

  • jailbreaking
    • average time between a jailbreak release and Apple shutting the hole is about 7 days…
    • can’t release an app to the store
  • MIDI
  • simulate capacitance touch using a piece of foil stuck to the screen
  • PeerTalk: using the USB sync cable
    • uses TCP sockets
    • same protocol as iTunes and Xcode

less crazy stuff

  • wifi is possible, but getting setting up is a pain
    • and there’s lots of support issues for different network situations…
  • acoustic coupling via the headphone jack
    • Square does this for a card reader
    • can even use the audio to provide 7.4mW of power (see Hijack board)
  • Redpark cable
    • dock connector to RS-232
    • comes with an SDK
    • but won’t let you put apps in the store — have to approve the app and the hardware
      • you can approach Redpark and ask them to request approval
      • a lot less expensive than going through MFi program
  • XBee & Zigbee (802.15.4)
    • mesh networking for low data rates
    • dock connector to XBee adapter soon available from redpark
  • bluetooth 4 (low energy)
    • can run for months with bluetooth active powered by a coin cell
    • introduced with the iPhone 4S
    • lots of boards available with Android & iOS SDKs
    • e.g. red bear labs
    • use CoreBluetooth plus board’s SDK
    • easy integration:
      • Alasdair did a live demo in just a few minutes
      • connected to arduino over Bluetooth 4 from iPhone and toggled an LED


Matias Piipari @ms2


  • CouchDB has been renamed as CouchBase Lite
  • TouchDB is document db with CouchDB-like API
    • but uses SQLite under the hood
  • sync & share with CouchDB (or TouchDB)
  • concurrency controlled like git
    • when you’re saving you have to be up to date first
  • lightweight
    • < 500Kb in app binary
    • 0.1s startup
  • Document <=> Model, with versions
    • can also contain attachments
  • use Views to define queries by property
  • once you’ve configured the replication it will keep going
    • don’t need to worry about network availability
    • do need to think about conflicts
    • do need to think about concurrency with data changes (get notifications on changes)
  • can create push and pull separately and to/from different destinations
    • e.g. combine pull from bundled with pull from remote
  • sync handles https with basic auth or OAuth
  • can combine db from iCloud/Dropbox with TouchDB

Step away from the screen

Nathan Error, Empirical Development @neror


  • your body is a tool too: improve your coding by improving your body’s effectiveness
    • exercise, diet & sleep…
  • scientometrics
    • measuring rate of change in science
    • rate of output increases by 7% each year
    • output doubles every 10-15 years
    • so whatever we think now will probably change several times over the next 10-15 years
  • maybe start looking at meat more as a side dish rather than a main
  • recent research has linked high intensity aerobic exercise to increased brain performance
  • missing one hour of sleep for a week is equivalent of a blood alcohol level of 0.2%

Subscription pricing

Manton Reece @manton on ADN


  • in 1999 working on a Mac app that cost $199
    • actually considered pretty cheap for the time
  • benefits of subscriptions
    • happy customers:
      • unhappy customers can cancel at any time
    • automatic paid upgrades
      • everyone is on the latest version
      • paying for the service
  • to justify subscription, app and service need to be one
  • examples:
    • adobe: switching to creative cloud monthly subscription
    • microsoft: office is now $9.99/month (or $99.99/year)
    • billings pro:
      • free for 1 invoice/month
      • 5 invoices/month = $10
  • focus on the consistent predictable part of the graph rather than the spikes
    • even if you don’t make more sales, the revenue is consistent
  • billing periods
    • payment percentages affect revenues: charging less often means less percentages to payment provider
    • Manton found that 57% of customers preferred yearly billing
  • hosting costs
    • use Amazon reserved instances if you’re committing to a year (saves money)
  • Stripe is leaps and bounds better than paypal…
    • but only available in US and Canada
    • beta coming to the UK this week!
  • Apple in-app purchase types:
    • non-renewing subscriptions: cancel is the default — will probably lose a lot of people
    • auto-renewable subscriptions: better, but more restrictions on review — including privacy policy & description
      • Apple are very cautious about letting non-magazine apps do this

The “Simple And Intuitive” Fallacy

Why we need standards for complex UX, too

Joerg Schweider @cooliopenguin


  • iPhones & iPads are not just accessory devices — they are becoming the main device in a lot of cases
  • apps need to be feature complete
  • if you simplify and cut out features then a lot of users will be left out in the cold
  • what makes a more intuitive UI?
    • can’t always find out by getting people to compare UIs
    • they will rate familiar schemes higher

Being Naive

Rob Rhyne @capttaco http://martiancraft.com/

  • “just build something so I can show it to the client”
  • users don’t care about engineering
  • iterate and test
  • at martiancraft, within three weeks of a new project you’re going to see something
    • it won’t be finished, but you can play with it
  • brent simmons: anatomy of a feature
    • the feature is the smallest part
    • it’s all the edge cases and polish that take the time
  • the naive implementation
    • demonstrates the feature
    • most obvious solution
    • takes the least amount of time to develop
  • example: histogram of live video
    • can use Accelerate framework to get histogram data from an image buffer
    • what about drawing the graph?
      • 3rd party charting lib: not obvious; little work; may demo but might not animate
      • OpenGL vertex buffer: not obvious; lots of work; will demo
      • CoreAnimation: obvious (when experienced!); little work; will demo
    • used CAShapeLayer
      • had used previously in Minds of Modern Mathematics
      • 14,000 pixel wide scroll view on an original iPad
      • can set up style once and draw separately
      • path was animatable property
  • http://giveabrief.com
    • codeless prototypes

Working on Sketch

Pieter Omvlee, Bohemian Coding

  • Sketch: vector art app
  • have to persuade users to try us rather than Adobe
  • don’t want a public feature list with voting
    • sets unrealistic expectations
    • there will always be business goals or technical issues that mean that highly requested features remain unfixed at the top
  • listen to your customers, but only to a certain degree
  • started giving away beta versions of app
    • got lots of testers & feedback
    • keep the fans happy
    • they’re very vocal and will do marketing on your behalf
  • don’t leave refactoring until the next big update
    • you’ll want to focus on new visible features to justify the upgrade price
    • customers don’t care about engineering
  • be practical with your time
    • don’t spend time on rewriting git history
    • focus on pricing, attracting the right customers, etc
  • put in crash reporting early
  • videos are excellent promotion & support
    • they take time and lots of takes
  • try and keep in contact with your customers
    • Apple gives you no way of getting in touch with App Store customers
    • but you could ask for details within the app
    • if no newsletter, then Change Logs are your only communication medium
    • “bug fixes” is a waste of an opportunity
  • try and contact bad reviewers

Introducing CoreValues

Scott Morrison (Chief Cook & Bottle Washer), Indev Software @smorr


  • Indie developer, less Independent, more Individual
  • but not just single person
  • instead individuality & personal
    • personal investment, principles, impact, payoff & risk
  • personal & professional roles are mixed
  • indie company has the heart of the indie developer
  • values are important and define you
    • but sometimes not thought through
    • can lead to inefficient decisions
  • Edward de Bono: if you want an out of the box solution, get out of the box first
    • PO statement — an unconventional (silly) idea used to generate ideas
  • introducing CoreValues: an objective-c framework for describing and defining personal and professional values…
  • interfaces:
    • personal ≠ private
    • professional ≠ public
  • (there was much more in this vein: thought-provoking metaphors, but I didn’t write them down)

Slightly Unsupported — Finder Code Injection

Steve Flack, Bromium UK

  • wanted to add icon badges & contextual menus
    • not based on file type
    • animated
  • used class-dump, x86 disassembler plus lots of trial & error
  • inject code with mach_inject
  • Finder has four methods to swizzle for icon badging
    • one for each view (desktop is fourth)
    • talk gave code specifics for each
  • only two methods to swizzle for contextual menus
    • desktop & grid view share
    • again, talk gave specifics

The Rise and Fall of a Mobile Startup

Emily Toop @fluffyemily


  • Tiny Ears 2011-2012
    • app to teach reading to 4-6 year-old children using speech recognition
  • Emily and her partner Ian both got startups accepted to Start-Up Chile
  • had to get speech recognition
    • Google Voice API was best accuracy by far
    • PocketSphinx with OpenEars was only option with no network
    • but all were bad for children
  • can improve recognition with a better model
    • 20 hours to customise a model
    • no existing models for children
    • 20,000 hours to create a new model!
  • Startup Weekend
    • pitch your product
    • then recruit audience to work on it for 48hrs
    • make an MVP
    • really useful
  • talk about your product
    • to everyone
    • until you bore everyone
    • even yourself…
  • used AVAnimator to play movies with alpha channels
  • …but speech recognition model would cost $100,000 and take 18 months
  • then animators didn’t want to continue without speech recognition
  • make your partnerships sound!
  • don’t go alone
    • guides recommend that you have a hacker and a hustler
  • mothballed Tiny Ears
  • joined another startup to learn how startups work
  • getting more involved in education
    • joined Code Club
    • applying to be a reading assistant
  • approached by dreamthinkspeak to help with their “large-scale, site-responsive theatre production inspired by Leonardo Da Vinci, The Book of Revelations and the world of Mechatronics”
    • go see them while they’re still in London (until March 30th at Somerset House)
  • nothing you learn is ever wasted

No comments: