How to Build Killer Products: Theories of Why Google Struggles. pt 2.

Link Texting
How to Build Killer Products
5 min readNov 18, 2014

tldr: This is part 2 in a series of blog posts about how to build killer products. Google is an immensely valuable company by any stretch of the imagination, but they struggle to build killer products. This is my analysis of why they struggle.

While this is not an exhaustive analysis, in the spirit of Cunningham’s law, I put my two cents out there and look forward to critiques & expansions to my theories.

A Short List of Google Products that I Love

I recently tried recalling every product that google has made that I love and consider mission critical. The list was surprisingly short. I found that my love of these products declined over time and that the switching cost became low.

Google search changed my life. I began exploring obscure fascinations with foreign music, learning languages I couldn’t dreamed have learning in a library, and more. It wasn’t until recently that I’ve started defaulting to reddit and quora as a search service.

Products like Google Plus had me frustrated and shaking in anger at Google for refusing to take in qualitative feedback in favor of “statistically significant” analytics.

  • Gmail. This is slowly declining in value especially since they removed calendar invites from the compose function.
  • Search = Amazing.
  • Earth — Grew up next to NASA. I’ve been addicted for years.
  • Translate. — Language obsession’s best tool.
  • Code— Has solid documentation.
  • Patents— Patents easily searchable. You can save money on a patent search by at the very least doing some simple searches here.

Google’s ‘Failed’ Apps

I use the word ‘Failed’ here to describe products that google launched that never reached most of the American public. This isn’t meant to describe their functionality. They didn’t become Huge.

  • Helpouts — Looks like Clarity.fm beat them out.
  • Bookmarks- Evernote and ReadItLater plus 100 others in the ‘bookmarking’ space. I used to love ohlife.
  • Maps — They bought Waze. They didn’t build the killer maps product, they bought it.
  • AdWords — Adwords provide diminishing returns and X% of ad traffic is bot traffic. Any level of searching through BlackHat forums will yield you this data. AdWords provides a lot of explicit value to small businesses, but it has its flaws.
  • Picasa — This product has so many problems with the way it interacts with google plus. It’s also difficult to understand who has access to which photos.
  • News — I stopped using this product. The LTV was low as a news aggregator. Now the market is flooded with news aggregators. For a while, before I became a technocrat, I thought FlipBoard might take over.
  • Books — Not sure where this went. They never built a quotes product out of it, which I thought was inherently unusual given the success of brainyquote and
  • Panoramia — meh. Not really great for exploration. The left and right keys don’t work to toggle between the two.
  • Trends — Kind of cool, but the LTV isn’t as high as searching reddit and quora to get an understanding of the reality of a trend.
  • Sites — There’s now 100's of players in the website building space.
  • Google Keep — Evernote dominates.
  • Google Wallet —- http://techcrunch.com/2012/09/09/the-mobile-payments-fustercluck/ — The mobile Payments Fustercluck lays it all out nicely.
  • Analytics — The existence of Segment.io as a player in the analytics space shows the failure of google analytics as a product. They never fully abstracted event tracking to be simple. Google analytics is something I use regularly but despise for its lack of abstraction.
  • Google Voice—- Better alternatives in greater numbers.
  • Google Hangouts — Google hangouts sucks so much that there’s a company that writes a blog post on how to use google hangouts in detail and then offers their solution at the end.

Google acquires 1 to 2 companies every week. Why can’t they build it internally?

1.) Their teams are too damn large. The ‘rent’ is too damn high.

The pressure to swing for the fences on every product decision and iteration is enough to mess up everything. Moving fast, breaking things, and being ruthless about it is better managed by small teams. The number of dependencies becomes too damn high.

2. ) Building with large scaling stuff is hard.

Using highly scalable things like BigTable instead of janky Rails is hard. It’s restrictive. The best way to build things at the onset is to build using the tools that you feel in high performance mode with unless the ramp up learning cost is low. Google used to force employees to use their ‘stuff.’ Not sure if this is still the case.

Scalable tools are hard to use and the ones google provides are far more complicated than others. Postrank was built in rails and had a huge corpus. There’s a reason Ilya Grigorik, the founder is now a web performance engineer at Google.

3.) Multi-platform is tough.

Being compatible on over 700 browsers is a huge undertaking for Google. The desire to make a product extremely accessible which is orthogonal to speed and growth, the fodder of a startup.

4.) You’re not allowed to break stuff and see how it fails and works in front of your users.

I can’t imagine bringing a product to the masses without the ability to break stuff and iterate in a very real way in front of my users. Being hyper compatible with every form of browser is tough. Every 2 weeks when users log in to LinkTexting they see changes and new updates to the product.

LinkTexting’s main website knowingly breaks on the opera mobile browser. It’s okay though. Hypercompatability is an extreme form of optimization.

Our failures and successes are very public. Our most recent push was an integration with BranchMetrics. In a more ideal world, Google wouldn’t be concerned with backlash from prototype failures, errors, and bugs.

Ways google can fix the problem.

  1. Buy more companies.
  2. Give small hackathon style teams more independence. The current decision tree for new products has way too many dependencies and sign-offs.
  3. Allow developers to use whatever tools and softwares that they’d like to use.
  4. Don’t worry about browser compatibility at the onset.
  5. Don’t brand skunkworks projects with the google name. Release them as if they were built in a basement. Don’t tie the brand to the prototype.

Why does Google struggle to build great products?

Tweet to us @linktexting, @datarade, or @blaurenceclark or hashtag #googlestruggles if you have ideas. We’ll be adding them periodically.

--

--