ProductTank AMS - Werkspot Product Culture - Antonio van der Weel
Data Driven Product Management - ProductTank Boston Feb '14
-
Upload
quantopian -
Category
Technology
-
view
531 -
download
0
description
Transcript of Data Driven Product Management - ProductTank Boston Feb '14
Data-Driven Product ManagementPractical Ideas and Tools PMs CanAnd Should Use to Make Decisions
1
About Me
Who here lives in Arlington? (Vote Dunn!) MIT mechanical engineer (but I never used it) 7 startups in 15 years Career path from support to implementation to
QA to PM
<date> – Confidential 2
3
Most PMs Aren’t Visionaries
Ideas come from customers, colleagues, and prospects
Steve Jobs isn’t walking into this product meeting
PMs probe, interpret, and synthesize
4
Ideas Are Not the Scarce Resource
Ideas come in sizes: markets, features, bug fixes, and optimizations
They have different motivations Increased salesHigher retentionLower cost of goods
Unlimited resources, you could do it all – but we don’t have that
Someone has to decide what is nextThis is why PMs get paid the big bucks
5
Optimize for Enterprise Value
The PM’s job is to prioritize What’s the North Star for your company?
Stars are directional – you can’t make a map to get to starHow do you know if you are pointed in the right direct?How do you know if you are making progress?
How do you compare apples to oranges? And compare that to bacon?
6
“That is a knowable fact.”
What the advocate says“No one uses that feature”“Everyone wants this!”“That breaks all the time”“You're not fixing enough bugs”“This problem happens to everyone!”“I’ve heard this request a million times”
What the data says15% of users click that every weekWe’ve had 3 customers ask for this feature5% of support calls are associated with a bug
7
Know Which Facts Are Knowable
Carefully separate opinion from fact, known from unknown
Huge, immediate reduction in complexity of the decision
Develop a third and fourth category1. We really don’t know2. Knowable fact3. We can know if we do . . .4. Before we decide, we really should know
A good PM uses all 4 categories to make a decision
This talk is more about 3 and 4
8
Know Your Data. Wallow In It.
9
Your Application Database Knows Your customers using your app are telling you
how they use it. You need to get the data reproducibly
You need data, not reportsKnow what you need to changeKnow if your changes actually worked or not
10
Measure It From The Start Your application database can’t tell you
everything Make an early change that adds data and
measurementPipeline speedFunnel shapeDaily activity
Measure the Good and the Bad You have to know what the problems are You have to know when they get worse
Make a Dashboard of It
When Do You Have to Decide?
Most of the time, the answer is “later” Don’t decide until you have to
This is where the art meets the scienceKnow your downsides and worst-case scenarios, and mitigate them
Watch, and monitor Agile (“agile”) really shines here
You will have the development bandwidth when you need itUnfortunately frustrating for many customers and colleagues
14
Time to Invest!
15
Keep Investing!
16
How Do You Decide?
Most decisions aren’t reduced to a time seriesComparing apples, oranges, and baconYour company needs all three
Collect all the data you canRead what the customer said (or potential customer). Talk to them
directly.Talk to the people who interacted with them (support, consultant, sales
rep, account manager)Look at the usageLook at the market and the competition
17
Find a way to order the data
Whiteboards and stickies What themes can you find What time ordering can you find What pre-requisites can you find Which ideas are both cheap and enable discovery
18
Build a framework
Whiteboards and stickies – and Excel Just make one up
10 points for data loss1 point for annoying1 point per customer affected3 points per big customer
You are the most qualified person to do it See what maps to your intuition, what doesn’t Know the limitations of what you built Iterate
19
Customer Pain (in thousands)
<date> – Confidential 20
My Tools
SQLYou need access to the data, not reportsNoSQL has query tools, too
Text editorUltraEdit. Python, Perl work tooTurn dross into data
ExcelYou do know how to make a pivot table, right? Find the lumpy parts.Can you do vlookups in your sleep? Integrate your data sources
Tableau Whiteboards and stickies
TheBrain mind-mapping software
21
<date> – Confidential 23