Upload
aaron-stannard
View
1.537
Download
0
Embed Size (px)
Citation preview
Startup Product DevelopmentBY AARON STANNARD, CTO PETABRIDGE
HTTP://WWW.AARONSTANNARD.COM/
Idea Stage
Define a product hypothesisParameterize it:
Who are the customers?What’s their case for action now?How are they going to buy?How are you delivering the product?How might all of these change over time?
“We’re going to build a tool to help large, .NET enterprises monitor and manage commercial deployments of distributed systems”
Parameters:
Customers: large, .NET enterprises (banks, healthcare, oil & gas, mining, gambling, insurance, retail, manufacturing)
Case for action: developers are expected to deliver more; don’t have tools or knowledge to do it (yet!)
How they buy: direct, inbound sales
Delivery: standalone installation; bundled with professional services
Change over time: annual subscription license; deliver updates and renewals to keep existing co’s happy. Expand sales through increase in deployments per account. Might add SaaS option to cloud environments. Maybe resellers.
Failure Conditions
1. No urgent case for action (idea is probably stupid)
2. No clear idea on how to deliver experience (can’t validate)
3. Unclear target persona (can’t target)
4. No vision (if idea can’t evolve, then it has no future)
Validation Goals
Validate / Invalidate Hypothesis How does it resonate with my personae?
Quantify that.
Tune parameters Determine which parameters are optimal for first implementation
I.E. what’s my most profitable group of users and what will they pay for?
Quantify that.
Figure out how to save as much time and money as possible Refine product idea down to limited scope. MVP.
TacticsValidation Tactics
B2C
Pre-Orders
B2B
Interviews
Professional Services
Landing Pages
List Building
Surveys
Tools for Validation
Github Pages – static HTML, super fast, easy to host
Mailchimp – basic email tool
Drip – hard-core email tool
SumoMe – analytics + email capture
Optimizely – easy A/B testing
Validation Results
Best metrics translate most closely to revenue Pre-orders (actual money)
Letters of intent (some will convert to future purchases)
List opt-in (I want this!!!111!)
Picking an MVP
LOW CUSTOMER VALUE HIGH CUSTOMER VALUE
LO
W B
UIL
D C
OS
TH
IGH
BU
ILD
CO
ST
SweetSpot
Technology Stacks (Web)
.NET / C# / Windows (Enterprise Stack)
Java / NoSQL / Linux (Big Data Stack)
Node.JS / JavaScript / Linux (RAD Stack)
Ruby on Rails / JavaScript / Linux (RAD Stack 2)
Technology Stacks (Client)
Objective-C / Swift / Xcode (native iOS)
Java / Eclipse (Android)
.NET / Xamarin (native Android and iOS)
JavaScript / Adobe Cordoba (RAD stack)
Factors in Selecting a Tech Stack
Labor availability Who do you know who’s proficient with a particular stack?
Do you need to hire people in your local area?
How much do consultants and full-time laborers cost?
Infrastructure Requirements Some stacks are “heavier” than others
Some are better late stage than earlier stage
Find a compromise and have a transition plan
The Process
Develop specification for an MVP Wireframes
Workflows
Find engineers you can trust and gather estimates Rough estimates initially
Pick engineers who suggest using off-the-shelf tools where appropriate
Develop a “workback” plan with engineers Trust the engineer’s judgment on where they need to begin
Set milestones and concrete deliverables
Hit milestones until MVP is finished
Workback Plans
Break tasks down until they’re no larger than “half a day” in size
Total amount of work to hit milestone = sum of the parts
(This estimate will still be wrong)
General rule of thumb for work estimation: multiply the number of units of work by 1 and increase the unit of work by an order of magnitude 2 days = 3 weeks
2 weeks = 3 months
3 months = 4 years
4 years = pick a new MVP
Facts about your MVP
It will have bugs
It will have rough edges
If it doesn’t, you’ve waited too long
It will still impress customers if you’ve done your validation right
When the MVP takes too long….
CUT features
Communicate with your users; don’t let them forget about you
DON’T hire more engineers
Deployment
Your engineers must have a deployment process in-place
Should be able to roll back to a previous version of the productUse a source control systemUse continuous integration (CI)
Launch Procedure
Always deploy your product before your marketingDeploy the night beforeMake sure it worksPostpone marketing materials if it doesn’t
Make sure your product is functional before you schedule major PR
Quantify how well your MVP proved your hypothesis
Most important: be able to collect information about bugs and product failures
Product analytics: track key product metrics like unique users, user retention, recommendations, etc..
Conversions: how well does the product convert users to customers?
Qualitative feedback: what works and what doesn’t? Break down by cohort.
Starting the Cycle Again
Change delivery / product based on feedback
Have specific goals for next iteration in mind: Improve user retention by X%
Improve conversion by X% or $N per month
Does this require a new feature? Often, the answer is “no”
Most of the time it involves fixing something customers already use