Transitioning to Lean at Infochimps

Transcript

1. Transitioning to Lean @DhruvBansal @JimEngland @TimGasper
2. We’re in a predicament with DATA.
3. DATA is the raw materials
4. DATA involves many markets and data types
5. DATA involves a wide variety of customers in undefined markets
6. how does Infochimps fit into the data landscape?
7. Data marketplace Social media influence IP intelligence Geolocated data our existing products
8. Our lean mission is to determine solutions to must-have problems for sizable markets.
9. foursquare data derivatives social media solutions new product hypotheses
11. We have a significant portion of the foursquare check-in firehose.
12. HYPOTHESIS FOURSQUARE DEVS Want venue hotness and venue ID matching for their applications.
13. How many 4sq devs are there?
14. How many would pay?
15. HYPOTHESIS DIGITAL AGENCIES Want venue hotness and brand comparisons for their clients.
16. How many brands want 4sq metrics?
17. How many would pay?
18. Must have Nice to have Don’t need
22. killing a feature is just as important as building one :)
23. Social media data
24. DEVELOPERS Wanted detailed and transparent influence metrics and rankings.
25. trstrank
26. HYPOTHESIS DIGITAL AGENCIES Want detailed and transparent influence metrics and rankings.
27. interviews later... 15+
28. Digital agencies actually wanted analytics research tools and dashboards.
30. That’s not our market. That’s not a feature set we’ve ever built. Leverages stuff we know, but little of what we’ve built. Sends the company in a very different direction. PROBLEMS
31. you’ve found a market and a product; should you build it?
32. Learn from and measure what you already have. Diagnose whether to pivot or persevere.
33. Product Strategy Vision Optimization Pivot
34. You shouldn’t define where to pivot to unless you know why you should pivot.
35. You can only know why if... 1 2 3 4
36. Clearly defined hypotheses. 1 2 Clearly defined actionable metrics. vs. vanity metrics
37. You’ve been experimenting and optimizing. 3 4 The experimenting and optimizing still has you short of your goals of creating a sustainable business.
38. Pivots are a change in strategy. The lean canvas provides a powerful strategy framework.
39. ERIC RIES STRATEGY Product roadmap Customer definition Partners & competitors Business model ASH MAURYA LEAN CANVAS Problem, solution & key metrics Customer segments & channels Value proposition, unfair competitive advantage Cost structure & revenue streams
40. the bottom line
41. Lean is more straightforward if you start with it from the beginning. If you haven’t been lean, you must revisit the past.
42. Move through the loop faster.
43. data web
44. TWO USE CASES, ONE WEBSITE developers data scientists
45. DATA SCIENTISTS business analysts, marketers, consultants, economists, statisticians, etc.
46. We provide data scientists access to data through downloadable data sets.
47. We provide developers access to data through RESTful APIs.
48. DATA SCIENTISTS? NO PROBLEM Downloadable data sets follow a typical web flow
49. homepage data set page search results blog twitter signup download
50. MULTI-STEP
51. DEVELOPERS? IT’S COMPLICATED API usage funnel involves many steps and is more difficult to track & define
52. homepage data set page search results blog twitter signup
53. new developer
54. new developer
55. active developer
56. A developer becomes active when they send over 1,000 API calls per month.
57. How do we convert a developer to an active developer?
58. HOWTOs CODE EXAMPLES
60. DATA SET PAGES API REFERENCE
62. homepage data set page search results blog twitter signup howto guide code example api reference data set page search results active user
63. PLUS
64. We need to connect user activity on the website with user activity on our APIs. TECHNICAL PROBLEM
65. GOOGLE ANALYTICS CUSTOM VARIABLES TO THE RESCUE SOLUTION
66. We use Google Analytics to track visitor behavior on our websites.
67. There are a lot of other options • Cookie, JS-based data collection • No servers to maintain • APIs for reports •Without JS, client-side behavior is a mystery •Lots of work •Potentially awesome
68. Web Data One identity is known to us. The other is known to Google.
69. ? Web Data
70. Google stores 3 cookies: VISITOR expires after 2 years SESSION expires after 30 minutes PAGE set on every pageview
72. Username First seen at Segment Original traffic source Milestones • Signed up • Downloaded • Searched • Bought • Queried Visitor Variables Milestones • Signed up • Downloaded • Searched • Bought • Queried Session Variables Category Completeness Pricing Distribution Page Variables
73. CUSTOM VARIABLES
74. Visitor Variables c=201110;ms=dev;ds=1;sd=1;u=dhruv First seen in October, 2011 (c) Is a developer (ms) Has seen a dataset (ds) Has done a search (sd) Username is ‘dhruv’ (u)
75. WE CAN GET THIS DATA OVER THE WEB.
76. WE CAN GET THIS DATA THROUGH AN API.
77. Web Data
78. developers developers developers developers
79. The web team’s goal is to increase the number of active API users. But we don’t have a traditional funnel for converting these users.
80. So, we started by focusing on the overall site experience for developers.
84. to Lean TRANSITIONING
85. We haven’t run any experiments on the website... yet.
86. web experiments brainstorm
87. Developer-centric email campaigns 1
88. Improved “getting started” dashboard 2 http://twilio.com/user/account
89. Improved search experience 3
90. thank you @DhruvBansal @JimEngland @TimGasper