9. Data Products...what are we talking about?
1. “...any application that combines data and algorithms...”
2. “A data application acquires its value from the data itself, and
creates more data as a result. It’s not just an application with
data; it’s a data product.” -Mike Loukides
3. “Data Products are self-adapting, broadly applicable economic
engines that derive their value from data and generate more
data by influencing human behavior or by making inferences or
predictions upon new data.” -Benjamin Bengfort
Read more…https://www.districtdatalabs.com/the-age-of-the-data-product/
10.
11.
12. Assumptions & Biases: A bit of pop-pysch...
Assumptions can arise from a lack of knowledge or a fear of
asking questions.
Biases can begin as lazy mental models that we slowly and
unconsciously adopt.
Both are substitutes for critical thinking and arise from a
natural human desire for a reduction in feelings of anxiety.
13. Assumptions & Biases: Cause problems...
Create friction for communication, teamwork, collaboration,
and coordination.
Increase the chance of failure through compounding
unknowns.
An uneasy truce: We rely on assumption and bias (formal
education, society), yet believe success comes from an
independent, questioning mind (entrepreneurial culture).
15. Product Management -> Questioning assumptions...
Customers: “I need this feature then the product will work the
way I want”
Sales/Biz Dev/Leadership: “Deliver this feature set and we’ll
be successful”
And many other examples….
16. Successful PDM’s are curious skeptics...
Value Experiential Data: Product Usage, Customer Support
Tickets, Sales Calls, and shadowing users.
Always Questioning: Re-Analyze Customer Data & User
Experience Research. Repeatedly ask “Why?”
Product & design frameworks work by quickly & iteratively
testing and validating solution hypotheses based on real
world observations.
17. Quick Review: What is a User Persona?
“Personas are fictional characters designers use to reflect
user types, pinpointing who they are and what they do with
products in relevant contexts. Designers create personas
from user data, to understand user characteristics, needs,
goals, etc. and gain valuable insights into user journeys, and
later, test prototypes.” Interaction-Design.org
18. Multiple User Personas -> Get used to it...
“You can’t build a data product that will make both your
customer and their end user happy.”
Multiple Personas are often found in...
● Most Enterprise SAAS products
● B2B products
● Worldwide consumer products (GDPR)
● Complicated products...
19. User & Buyer Personas: Critical for MVP success
User Persona: Usually refers to the end user of the product,
who is often the ‘customer’ of the buyer persona.
Buyer Persona: The person responsible for making sure the
value of the product is greater than its cost.
MVPs must focus on meeting the needs of both users but if
the Buyer cannot see or measure success then the product
will fail.
20. Focus on the top line KPI for your customer...
What are they looking to improve?
● Revenue?
● User Growth?
● Engagement?
Determine how to support end user behavior to drive this KPI.
Success there proves out product value… and allows for broader
feature work to support more advanced use cases.
21. After the customer is (mostly) happy with MVP...
Re-evaluate problem space & user needs, perhaps your
original analysis needs updating.
Remember reuse wherever possible and look to create
layered feature solutions (especially for data generation).
…. But sometimes it’s best to re-invest in original value
proposition and optimize.
23. New Data Product -> Old Product Assumptions...
“You can’t build a new data product in a company with an existing product and
revenue stream.”
● “It’s not in our DNA”
○ It’s about the narrative, which can be changed.
● “You’ll distract us from focusing on the core product”
○ Revenue is revenue.
● “You’ll never convince existing customers to buy it”
○ The marketplace is judge.
● “It will never scale”
○ Needing to scale is a good problem.
24. “Raw” Data = The Truth
“Raw data is both an oxymoron and a bad idea; to the
contrary, data should be cooked with care.” Geoffrey Bowker
“Data doesn’t speak for itself -- it echoes it’s collectors.” Lena
Groeger
“If you torture the data enough, nature will always confess.”
Ronald Coase
Read more… https://www.thenewatlantis.com/publications/why-data-is-never-raw
25. “A Data Product has to be complex!”
Simple solutions have less operational cost and are easier to sell,
support, and build.
Complexity makes it harder to improve, change, or maintain a
product.
Simplicity forces the PDM to focus on rigorous problem definition
and creating solutions that consist of only what provides value.
Speed to market: Being able to test with customers almost always
trumps waiting for something better.
26.
27. Good Complexity: AB (or Multivariate) Testing
Provides a way to truly test assumptions against real world
conditions.
“The basics”, but often first on the list of items to de-prioritize.
Encourages data feedback loops and analytical thinking from
the whole team.
Provides operational, sales, and other value outside of just
product development.
29. “You must use Big Data and Machine Learning!”
Big Data: Large quantities of streaming, real world, log, or
event data that requires extraordinary processing and storage
techniques. It’s complex to extract meaning from this data
due to not only to size, but also velocity, and raw (sometimes
error filled) state.
30. “You must use Big Data and Machine Learning!”
“Machine learning (ML): is the scientific study of algorithms
and statistical models that computer systems use to
effectively perform a specific task without using explicit
instructions, relying on models and inference instead. It is
seen as a subset of artificial intelligence. Machine learning
algorithms build a mathematical model of sample data,
known as "training data", in order to make predictions or
decisions without being explicitly programmed to perform the
task.” -Wikipedia
31. The goal is the best product not the best tech...
More Signal or more Noise… Big but at what operational, organizational,
and mental cost?
Challenge complexity: If we do less can we launch now? Can we start
over if this fails? How will Sales & Customer Service support?
Privacy: Understand the what and why behind data before customers,
regulators, and stakeholders ask first. Privacy isn’t always a problem.
“Is ____ stable enough I would run my life on it?” because your
customers will...
32. If you do use Big Data tool sets...
● Be sure to define what ‘big’ is and why it is important. Is this is a
product need, an engineering need, a contractual need?
● If possible leverage existing work:
○ Is there already an internal system you can repurpose/copy?
○ Can this be solved by buying from a services provider?
○ Would waiting X weeks/months help?
● If you must build new: Choose stable, open source, and heavily
supported tool sets.
● Be aware of any product or business requirements to keep un-
aggregated or un-analyzed data!
33. If you do use Machine Learning tool sets...
● It’s difficult to know what ML techniques will work for a given project or
product until you build a POC. Get something simple working with
open source tools ASAP!
● The model output & product depend heavily on the quality and
quantity of training data in most cases.
● If it’s hard to get two human experts to agree on the ‘right’ answer to
the problem, then it might be a bad fit.
● Figure out what to do when the algorithm fails. (Remember it’s ok to
rely on humans for the last 10-20+%.)
34. More thoughts on product success...
● Creating the group narratives needed for success:
○ A strong data focused culture.
○ An achievable product roadmap.
○ Buy in from key stakeholders.
○ Trust, patience, and an appetite for failure.
● Quickly launching simple data products in the market to
validate user value, revenue, and the data
ecosystem/economic feedback loop.
● Resilience and a sense of humor. This is all new stuff,
have fun!
35. www.productschool.com
Part-time Product Management, Coding, Data and Digital
Marketing courses in San Francisco, Silicon Valley, New York,
Santa Monica, Los Angeles, Austin, Boston, Boulder, Chicago,
Denver, Orange County, Seattle, Bellevue, Toronto, London and
Online
Notas del editor
Give examples of real world problems with browswers etc