An Exclusive Interview with Pinterest Head of Developer Relations

Mark Boyd
Nov. 19 2013, 10:35AM EST

After months of anticipation and eager watching from the sidelines by developers across the board, Pinterest has begun releasing access to a new suite of API endpoints aimed at building strong business relationships and generating value flow to Pinterest end users. ProgrammableWeb spoke with Pinterest Head of Developer Relations, Jason Costa, about the new APIs. We also spoke with Jinny Kwon, from Random House, about how they created a recommendation engine with the Pinterest APIs. "The initial APIs that we have opened up allow partners to better surface their content to their users", Costa said. "It's like what Will Young from Zappos Labs said in the Wall Street Journal the other day: they have 150,000 objects on their website, so using Pinterest APIs is a great way to supplement their product directory and to surface items that customers might otherwise have missed." The newly released Pinterest APIs are aimed at established businesses - particularly those centered around an online content or product catalog - to provide new ways to engage with their end users by integrating into the Pinterest social media platform. The Pinterest APIs connect a partner's domain content with APIs that surface a site's:

  • Top repins
  • Most recent Pins
  • Related Pins
  • Most clicked-through Pins and
  • Pins from specific search terms.

Initial partners included AllRecipes, Whole Foods, Zappos, Target, Random House and others, but Costa is adamant that - with a creative and well-thought out integration plan - any business can gain access to the API toolset. To sign up, developers must register on the developer portal, provide mockups or wireframes, and explain their overall integration strategy goals. "It is by no means an exclusive API, we just want quality integrations," Costa said. "It's a lot less about getting into our API access queue and more about making quality mockups that enable us to see what it looks like and to be confident about the quality experience for end users. We're basically going though and vetting the applications. It's not exclusive to brands, if a smaller startup has clearly thought through the integration, we will give them access." Costa recommends that developers interested in using Pinterest's APIs first look at the initial range of partners to see how they are using the APIs and then consider carefully the value a business' end users will gain from Pinterest content integration. "Looking at these initial integrations is a really good barometer for the type of partnerships we are looking for, [we suggest] developers using that as a reference point and seeing how they are using Pinterest APIs to surface content to users - that really sets the tone [of what we are looking for]." Costa points to Target's creative use of the APIs: "They are populating their 404 pages with suggested pins. A 404 page for a product that is no longer stocked, that tends to be a page with a lot of user disappointment and instead Target is using it to give suggestions for other items customers may be interested in viewing." Costa also highlighted the way Random House is using the Pinterest APIs to create a recommendation engine for their site. ABOVE: How Random House are using Pinterest as a recommendation engine Jinny Kwon, Vice President and Director: Design & Development at Random House discussed what it was like for their development team to be working with the Pinterest APIs:

"We all appreciate the existing user experience Pinterest has built, so we were excited to work with the APIs, which gave us access to more ways of mashing up data and presenting in a creative format for our own users. We learned that it helps to have a clear strategy or plan from the get-go so developers can quickly make decisions around how to leverage the data. I’d say one challenge for the developers was mapping out Pinterest data to Random House data since the landing page isn’t just the stream of pins but the merging of data around the object (which is the book). The way in which Pinterest provides pin data also led us to dedupe pins to minimize repetition and improve user experience. There’s still some repetition but that also reflects the popularity of a book. The final execution is a single page, but there’s quite a bit happening behind the scenes as well. A middle tier was built to optimize performance and new front tools were also used. There was a lot of learning and building beyond the API and in a short amount of time – a “sprint” in fewer words."

Kwon echoes one of Costa's fundamentals for API signup: that potential partners think about the integration strategy first, and then submit the request to access the Pinterest APIs. "We had a team brainstorming session, and the concept of a discovery tool took shape. It was a very simple idea – let the Pinterest community curate the world of Random House books to other readers. We also wanted to enhance the Pinterest data by leveraging Random House’s database to provide readers with additional relevant information around a book, such as available formats and other books by the author." The result for Random House has been an enhanced user experience for their customers, Kwon said. "The response has been positive all around. Visitors to the page see what books are bubbling up through the Pinterest lens on a range of topics and interests as well as reflecting the breadth and depth of the Random House catalog. It’s wonderful to see a book surface that was first published a few years or even decades ago." Working towards this outcome is a dealbreaker for Costa. "We want to see how integrations are adding value to our business customers, and - most importantly - to Pinterest end-users," he said. For those businesses that do have a clear Pinterest integration plan, Costa assures developers there are a range of supports in place:

"The way it is set up now, you go to the developer site, and there is an overview of the APIs: end points, parameters, the technical documentation. "It is a RESTful API. We have standard HTTP response codes. We will send back an HTTP 429 if you exceed your rate limit for the hour, or send back a 404 if the method is not found. We also include a descriptive message in a response message. "The good news is that also, when we do whitelist a partner, we provide a number of additional tools, including an API explorer for example that allows realtime calls in a console to see what the JSON response body looks like. This all helps developers to scale the learning curve and we will continue to give them the support they need."

Costa was unable to confirm any roadmap for a more generic API that might be available to third-party developers looking beyond using Pinterest in conjunction with their own content. "The most exciting thing for us is to see the creativity of our partners, how they are using Pinterest datasets to power their audience engagement," Costa said. Kwon is already there, thinking about what is next for Random House's partnership with Pinterest: "Some additional ideas for future release are to let visitors start reading a book in the modal, sorting and filtering, as well as building an embeddable widget of the discovery tool that would showcase the feed of pins in real time. Also, this week we plan to launch a discovery tool on Tastebook.com, our cooking community and recipe management site, to surface recipes that are popular in the Pinterest community." Developer teams working with businesses interested in partnering with Pinterest via their APIs can sign up now.

Mark Boyd is a ProgrammableWeb writer covering breaking news, API business strategies and models, open data, and smart cities. I can be contacted via email, on Twitter, or on Google+.

Comments