share and tell stanford 2016
TRANSCRIPT
Lessons learned after 130 interviews
Yegor Tkachenko, MS Marketing AnalyticsMachine Learning
Eric Peter, CS & MBAConsumer Insight ExpertManagement Consulting
Scott Steinberg, MBAMarketing Growth StrategyManagement Consulting
Karan Singhal, Undergrad CSWeb Development
User Interface Design
Share&Tell
Share&Tell
Yegor Tkachenko, MS Marketing AnalyticsMachine Learning
Eric Peter, CS & MBAConsumer Insight ExpertManagement Consulting
Scott Steinberg, MBAMarketing Growth StrategyManagement Consulting
Karan Singhal, Undergrad CSWeb Development
User Interface Design
Day 1 (Clarified)We create a way for
consumers to make money by actively sharing their behavioral data and
opinions.
Through this data, we help companies unlock
previously unattainable insights.
NowWe help retailers and CPG
companies understand online shopping behavior.
We do this by creating a platform for people to donate
their Amazon shopping history
to raise money for charity.
130 Interviews
3,500+ Survey responses
Cost StructureFixed - Infrastructure, servers, team of data scientists, corporate sales force, project managers & analysts, product & user experience development team
Variable - Payment to consumers for use of their data, profit-sharing model (dividends) with consumers, consumer service reps
Revenue Streams1. Custom research studies2. Per-feedback fees (surveys, video interviews, focus groups)3. Sales of raw data / data with automated analytics on top4. Subscriptions to the platformPricing based on sample size/type, data type/amount, number of questions, feedback time
Key Resources
Key ActivitiesKey Partners
Value Proposition
Customer Relationships
Channels
Business Canvas - Week 1Customer Segments
Consumers• Millennials/students• Lower income consumers with smartphones• Existing research participants
Enterprises• Marketing agencies, consulting• Marketing departments at large companies• Marketing departments at non-large CPG companies
• Panel acquisition, retention, incentivization, quality control• Automated seamless insights extraction • Data security• Empowered customer service (for consumer)• Sales force, customer service knowledgable about market research design & execution
• Historical granular data
• Automated platform for seamless insights extraction
• Expertise in market research methodology, execution, statistics
Consumers• Profit sharing• Targeted ads in line with customer’s tastes• Sense of empowermentEnterprises• Unique data,analysis• Easy and fast way to do it
Consumer• Website• Mobile app
Enterprise• Direct web portal• Resold through market research agencies• Custom consulting & research design services
Consumers• Getting paid for data that has already been shared, but from which individuals are not profiting• Provide sense of empowerment and control over data• Offers a natural, effortless way to share opinions• Feel heard and that opinion matters
Enterprises• Linking real-behavior with opinions (vs. stated behavior)• Ability to follow up with consumer• Faster turnaround
• Data API providers• Data aggregators• Marketing agencies• Panel participants
blue = consumerblack = enterprise
What we thought: Enterprise VPblue = consumerblack = enterprise
Enterprise Value Proposition:
Replace traditional survey providers by:● Linking real behavior with opinions (vs.
stated behavior)● Ability to follow up with consumer● Faster turnaround
Key Resources• Historical granular
data
• Automated platform for seamless insights extraction
Demographics● Age?● Gender?● ...
Behavior● Where did you buy?● What? How much?● ...
Emotions / Feelings● Why did you buy?● What matters to
you?● ...
Survey
Surveys are based on SELF REPORTED data
Replace with real data
What we did: Talk to companies who use surveys for market researchHypothesis: We can replace existing panel vendors if we have real behavioral data (as opposed to self-reported data)
What we did: 12 Customer Discovery interviews with companies that conduct market research using surveys
EnterpriseWeek 1-3
What we found: Not that muchpain with self-reported data...
“Self-reported data isn’t great, but it’s directionally good
enough.”
“With real data, we’d get the same insight as we do now, but perhaps we’d be slightly more confident.”
“In order to switch vendors, you need to be able to answer a question we can’t
answer today”
“We have to use [vendor] - we have a long term contract through our HQ."
EnterpriseWeek 1-3
What we found: Not that muchpain with self-reported data...
“Self-reported data isn’t great, but it’s directionally good
enough.”
“With real data, we’d get the same insight as we do now, but perhaps we’d be slightly more confident.”
“In order to switch vendors, you need to be able to answer a question we can’t
answer today”
“We have to use [vendor] - we have a long term contract through our HQ."
EnterpriseWeek 1-3
Adding behavioral data alone does not make us 10x better.
We need to be able to answer a specific question that marketers can’t answer today
So, we focused on changing the value prop to answer new questions for marketers
How should I identify my
consumer target(SMB Businesses)
How do I better understand my
consumer target?
What is the path to purchase for online
and omnichannel shopping?
What are current online shopping trends?
Customer Needs Identified through Customer Discovery:
EnterpriseWeek 1-3
So, we focused on changing the value prop to answer new questions for marketers
How should I identify my
consumer target(SMB Businesses)
How do I better understand my
consumer target?
What is the path to purchase for online
and omnichannel shopping?
What are current online shopping trends?
Customer Needs Identified through Customer Discovery:
EnterpriseWeek 1-3
Value Proposition
Enterprises• Linking real-behavior with opinions (vs. stated behavior)• Ability to follow up with consumer• Faster turnaround
Value PropositionEnterprises
• Identify target consumers to increase marketing ROI
• Deeper and more accurate behavioral understanding of consumer segments
• Understand online/omnichannel path to purchase
• Understand online market trends at consumer level
Week 1 Week 3
✘
What about the consumer?
Cost StructureFixed - Infrastructure, servers, team of data scientists, corporate sales force, project managers & analysts, product & user experience development team
Variable - Payment to consumers for use of their data, profit-sharing model (dividends) with consumers, consumer service reps
Revenue Streams1. Custom research studies2. Per-feedback fees (surveys, video interviews, focus groups)3. Sales of raw data / data with automated analytics on top4. Subscriptions to the platformPricing based on sample size/type, data type/amount, number of questions, feedback time
Key Resources
Key ActivitiesKey Partners
Value Proposition
Customer Relationships
Channels
What we thought: Consumer VPCustomer Segments
Consumers• Millennials/students• Lower income consumers with smartphones• Existing research participants
Enterprises• Marketing agencies, consulting• Marketing departments at large companies• Marketing departments at non-large CPG companies
• Panel acquisition, retention, incentivization, quality control• Automated seamless insights extraction • Data security• Empowered customer service (for consumer)• Sales force, customer service knowledgable about market research design & execution
• Historical granular data
• Automated platform for seamless insights extraction
• Expertise in market research methodology, execution, statistics
Consumers• Profit sharing• Targeted ads in line with customer’s tastes• Sense of empowermentEnterprises• Unique data,analysis• Easy and fast way to do it
Consumer• Website• Mobile app
Enterprise• Direct web portal• Resold through market research agencies• Custom consulting & research design services
Consumers• Getting paid for data that has already been shared, but from which individuals are not profiting• Provide sense of empowerment and control over data• Offers a natural, effortless way to share opinions• Feel heard and that opinion matters
Enterprises• Linking real-behavior with opinions• Ability to follow up with consumer- Faster turnaround• Give additional context in traditional surveys
• Data API providers• Data aggregators• Marketing agencies• Panel participants
blue = consumerblack = enterprise
Consumer Value Proposition Hypothesis:
Get paid for your dataFeel in control of your data
Feel heard and that opinions matter...and, that consumers are willing to provide all these data types:• Social media likes & posts• Email purchase receipts• Credit card purchase
history• Amazon.com purchase
history• GPS location history• Web and search history
First consumer testHypothesis: People will provide their data and opinions for money
Tested through: ~25 Customer Discovery focused consumer interviews
ConsumerWeek 1-3
Experiment: Take an MVP on an iPad to the mall
ConsumerWeek 1-3
What we learnedHypothesis: People will provide their data and opinions for money
ConsumerWeek 1-3
Findings:
People will provide data and opinions for money, BUT
Only younger and poorer consumers were interested
Cash-based model had other problems too:● Doesn’t support retention and engagement● Misaligned incentives● Not scalable to get to large # of consumers
Tested through: ~25 Customer Discovery focused consumer interviews
As a result: What if we offered equity instead of cash?
Solves all business needs! ● panel retention and engagement● identity verification● quality of data
ConsumerWeek 4
Google Consumer Survey: n = 500
Oh Wait… Need to Isolate VariablesAlways be skeptical of your data!
Consumers aren’t interested in concept of being a partial owner - they cared about the
extra cash!
Designing a good experiment just saved us 49% of our equity...phew!
ConsumerWeek 4
Value Proposition
Consumer:• Getting paid for data that has already been shared, but from which individuals are not profiting• Provide sense of empowerment and control over data• Offers a natural, effortless way to share opinions• Feel heard and that opinion matters
By Week 4, We Had No Idea What Consumer Value Prop Should Be
Value Proposition
Consumer:• Getting compensated for data that has already been shared• Provide sense of empowerment, control over data• Partial ownership of company
Week 1-4ConsumerWeek 1-4
Consumer:• Control over data• ???
Value Proposition
Week 1 Week 3 Week 4
Let’s first focus on narrowing down enterprise value prop to see what data we need.
What we did: Customer Validation!
How should I identify my
consumer target(SMB Businesses)
How do I better understand my
consumer target?
What is the path to purchase for online
and omnichannel shopping?
What are current online shopping trends?
✘ ✘
EnterpriseWeek 4
14 more enterprise interviews to (in)validate our hypothesized value props and identify the most acute needs
“Great value prop guys, but I challenge you - if you had to do something tomorrow as an MVP,
what would it be? This is a LOT to do!”
Note: Quote paraphrased, concept of “Big Idea” was likely referenced
Key learning: A startup can’t do everything. It needs to do one thing well!
EnterpriseWeek 4
Well, why not focus on data that’s easiest to get?
Most Sensitive
Least Sensitive
Google Survey
ConsumerWeek 5
And heard from companies that Amazon data is big pain point
EnterpriseWeek 5
As a result: An aha moment...
Share & Tell…...helps better understand your target's online & omnichannel shopping & purchasing behavior
• What is purchased on Amazon.com?• What is my online/omni market share?
Why?• Where else does my target shop? Why?• What does my target do before they buy?
What is their shopping path? Why?• What products does my customer buy /
not buy? What do they buy with my product? Why?
...helps better understand your target's persona / where to reach them
• What online behaviors (sites, apps, etc…)?• What media consumption habits?• What do they search for online?• What activities, interests, hobbies?• What demographics?
...provides ability to more directly and narrowly communicate with your target
• Direct messaging / promos on S&T platform
• Better targeting on existing ad networks
Focused value prop and segments:
Help CPG and retailers understand consumer
trends on Amazon
EnterpriseWeek 5-6
Cost StructureFixed - Infrastructure, servers, team of data scientists, corporate sales force, project managers & analysts, product & user experience development team
Variable - Payment/donations for use of their data, consumer service reps
Revenue Streams1. Subscriptions to insights / platform2. Per-survey fees3. Custom research studies4. Linking data to client databasesPricing based on sample size/type, data type/amount, number of questions, feedback time
Key Resources
Key ActivitiesKey Partners
Value Proposition
Customer Segments
Customer Relationships
Channels
Resulting Business Canvas
Consumers
• Smartphone using consumers who shop online• Millennials• Existing research participants• People who currently give to charity
Enterprises
• Retail (traditional)• Retail (e-commerce)• CPG with online sales
• Panel acquisition, retention, incentivization, quality control• Automated seamless insights extraction • Data security• Empowered customer service (for consumer)• Sales force, customer service knowledgable about market research design & execution
• Historical granular data
• Automated platform for seamless insights extraction
• Expertise in market research methodology, execution, statistics
Consumer• Website• Mobile appEnterprise• Direct web portal supported by research-experience B2B sales force• Projects sold through market research & strategy firms
Consumers• Get: Charities send invitations• Get/Keep: Shopping discovery + targeted discounts app• Keep: Reports / comparisons of your data
Enterprises• Get:partnership,telesales,PR • Keep: Unique data, analysis• Easy and fast way to do it
Consumers• Feel good by donating data to charity• (potentially) Service to discover, get discounts on, and buy stuff online
Enterprises• Understand purchasing trends on Amazon by demographic group
• Data API providers• Data aggregators• Marketing agencies• Panel participants• Charities/non-profits
EnterpriseWeek 5-6blue = consumer
black = enterprise
• Understand purchasing trends on Amazon by demographic group • Retail (traditional)
• Retail (e-commerce)• CPG with online sales
As a result: Develop low-fi MVP
EnterpriseWeek 5-6
Now, how do we incentivize consumers to provide Amazon data?
ConsumerWeek 5
We identified a few possiblealternatives to cash...
Pay cash
Provide a valuable service
$5 / $10 cash
Donate your data
(to benefit a charity)
Receive targeted
promotions
Personalized product recommen
dations
✘Had learned previously consumers more
willing to share data if they get some intrinsic value
ConsumerWeek 5
What we did: 10+ Customer Discovery interviews...and 2,000+ survey
responses
ConsumerWeek 5
What we found: “Donate your data”best meets the business’s needs
Gets Amazon data?
Retention / engageme
nt? Quality? Large #? Outcome
$5 / $10 cash
✔ Cash is king! ✘ May be transactional / one-shot deal
✘ Limits to low income
✔ ~>50% interested
Kill for now or use in combo w/ donations
Donate your data
✔ Interest in ‘doing good’
✔ Donation implies opp to ask for future donation
✔ Consumer leads verified through charities
✔ ~27% interested
Focus for class; need to understand impact of bias
Targeted promos
✘ Does not solve major pain, already available
✔ Creates clear gain w. reason to come back
✔ Can verify respondent behavior
✘ Quant test running, qualitatively poor reaction
Test for “keep / grow” insteadProduct
recs✘ Limited interest - does not solve pain, not 10X better than others
✔ Creates clear gain w. reason to come back
-- Unclear if able to verify respondent• Need 0.75% of TAM to register (1M /
150M)• Of those interested, ~3% will register• Implies >25% interested
ConsumerWeek 5
What we found: Consumers skeptical of donation scams
“I’d donate my Amazon data to raise money for charity X,
but only if that charity asked me too”
“I probably would not donate to a random
startup unless I knew for sure that they
were legit”
Nonprofits should send out communication asking people to donate their data
Nonprofits are a customer acquisition channel and a new customer segment
ConsumerWeek 5
As a result: 3-sided marketConsumer
Week 6
Value Proposition
Consumer:• Control over data• ???
Consumer:• Feel good by donating data to charity• Doesn’t cost money to donate
Value Proposition
Week 3 Week 5
Resulting BMC changes (I)
Consumer:• Millennials & students• Lower income consumers with smartphones• Existing research participants
SegmentConsumer:• Millennials• People who donate to charity
Segment
ConsumerWeek 6
✘✘
Value Proposition
Non-Profit:• A new revenue stream• A new way to engage with donor base• A way to get donations without pushback
Value Proposition
Week 3 Week 5
Resulting BMC changes (II)
Segment
Non-Profit:• All non-profits
Segment
ConsumerWeek 6
Resulting BMC changes (III)Consumer
Week 6
Consumer:• Targeted ads in line with customer’s tastes• Sense of empowerment
Cust. Relationship Consumer:
• Get: Charities send invitations
Cust. Relationship
Need to test this
✘
eCommerce Data & Insight
Companies
Data aggregators
Online Donation Tools and Platforms
Slice, Clavis, Profiteero, One Click Retail, Profiteero, Return Path, Paribus?
Data Wallet, Datacoup, Infoscout, Axciom, Experian, LiveRamp, SuperFly
Razoo, CrowdRise, Causes, Survey Monkey, One Big Tweet, GoodSearch, AmazonSmile
Share&Tell
Marketing research agencies
TNS Qualitative, ,Conifer Research,Horowitz Research,Nielsen, Kantar, IPsos,dunnhumby
Our Competitive Set Has Evolved too
Removed through pivotsOnline Survey ToolsTraditional survey panelsOnline qualitative research
Behavioral Consumer Panels (w/ or w/o surveys)
Nielsen, NPD, IRI, LuthResearch,VertoAnalytics, RealityMine, comScore
SHARE & TELL
ConsumerWeek 6
Nonprofits might not be the right routeWhat we did:
Interviewed 10+ nonprofits
Tested email campaign to 60
nonprofits to gauge interest
What we learned:
● Only nonprofits who value smaller donations (<$100) from larger base of people were interested in the model
● Nonprofits are slow to make decisions and risk-averse
So what?
Focus more efforts on testing viability of direct to consumer route.
Key hypothesis to test: Can we build enough trust through social media and website?
NonprofitsWeek 7-9
Non-profits may not be most
efficient consumer acquisition path.
What we did: Tested ‘direct to consumer’ using a high fidelity MVP...
https://www.datadoesgood.com
ConsumerWeek 7-9
What we learned: ‘Direct to consumer’ might be a viable route
Arrived to the landing page
Clicked ‘donate now’
Logged in with Facebook
Shared Amazon data
Filled out demographics
100%
~18%
~6%
~6%
~5%
~80%
~95%
~55%
Choose a charity ~11%
~60%
25%
ConsumerWeek 9
Cost StructureFixed - Infrastructure, servers, team of data scientists, corporate sales force, project managers & analysts, product & user experience development team
Variable - Payment/donations for use of their data, consumer service reps
Revenue Streams1. Subscriptions to insights / platform2. Per-survey fees3. Custom research studies4. Linking data to client databasesPricing based on sample size/type, data type/amount, number of questions, feedback time
Key Resources
Key ActivitiesKey Partners
Value Proposition
Customer Segments
Customer Relationships
Channels
Consumers• Online shoppers• Current charity givers• Millennials• Existing research participants
Enterprises• Buyers at e-commerce retailers • Marketers at CPG with online sales
Nonprofits??• Hungry for donations and values small donations from large # of donors• Private donations are main revenue stream
• Donor acquisition??• Donor retention and engagement??• Data quality control• Data security and storage• Automated analytics• Custom analytics• Sales force• Legal
• Physical - workspace, servers• Additional human (short-term) - Full-stack software engineer, Database architect, Security consultant, Legal Consultant, Advisors/Industry Movers(long-term) - Sales team, Analytics team, Security team, Engineering team, Advisors• Intellectual - Trademarks, Contracts with clients, Proprietary analytic tools, Software copyright• Financial - angel/venture funding
Consumers• Website• Mobile app
Enterprises• Web portal supported by B2B sales force• Projects through market research & strategy firms
Nonprofits??• Web portal
Consumers• Get: Social media campaigns & charities send invitations• Keep: Reports / comparisons of your data Enterprises• Get:partnership,telesales,PR • Keep: Unique data, analysis• Easy and fast way to do itNonprofits??• Get: telesales, PR
Consumers• Feel good by donating data to charity• Donating is free & easy
Enterprises• Understand purchasing trends on Amazon by demographic group.brand preference
Nonprofits??• A new revenue stream• A new way to engage with donor base• A way to get donations without pushback
Short Term:• Charities/non-profits• Nonprofit hubs/associations• Legal• Other collectors of online purchase history
Long Term• Data API providers• Data aggregators• E-commerce retailers• Ad networks and programmatic ad buyers?
Final Business Model Canvas Week 10
So...what’s next...
We are going to continue working on this after the class.
Can we gain traction with consumers?
Several additional experiments we want to run incorporating feedback from our MVP.
● Facebook “nominations”● Linking more directly to causes● Many improvements to the MVP
Can we get a letter of intent from any businesses?
We continue to hear companies say they are interested and that this data is valuable. Is one willing to sign a non-binding letter of intent
First Priority Second Priority
Thank you, George!
Appendix
What we learned: Refined value proposition for enterprise...Share & Tell…...helps better understand your target's online & omnichannel shopping & purchasing behavior
• What is purchased on Amazon.com?• What is my online/omni market share?
Why?• Where else does my target shop? Why?• What does my target do before they buy?
What is their shopping path? Why?• What products does my customer buy /
not buy? What do they buy with my product? Why?
...helps better understand your target's persona / where to reach them
• What online behaviors (sites, apps, etc…)?• What media consumption habits?• What do they search for online?• What activities, interests, hobbies?• What demographics?
...provides ability to more directly and narrowly communicate with your target
• Direct messaging / promos on S&T platform
• Better targeting on existing ad networks
EnterpriseWeek 4
...for 3 generic enterprise segments
EnterpriseWeek 4
Retailers
Traditional
E-Commerce
1
2
CPG
With online sales
Without online sales
3
What is market research?
Comes in many forms...
1.Surveys to understand consumer opinions / emotions
2.Data to understand market trends
Initial hypothesis:“disrupt” survey-based market research
A quick primer:How do surveys work?
What features do my
customers care about?
1 Business asks a question about their customer
What does my most valuable customer look
like?
What drives customer loyalty?
A quick primer:How do surveys work?
2 Market research team writes a survey that will inform the answer
Demographics● Age?● Gender?● ...
Behavior● Where did you buy?● What? How much?● ...
Emotions / Feelings● Why did you buy?● What matters to you?● ...
Survey
5 - 10 minutes of questions
10 - 15 minutes of questions
A quick primer:How do surveys work?
3 Survey sent to consumers through a ‘panel provider’
Demographics● Age?● Gender?● ...
Behavior● Where did you buy?● What? How much?● ...
Emotions / Feelings● Why did you buy?● What matters to
you?● ...
Survey
$ / person
Panel ProviderMarket Research team
Demographics● Age?● Gender?● ...
Behavior● Where did you buy?● What? How much?● ...
Emotions / Feelings● Why did you buy?● What matters to
you?● ...
Survey
A quick primer:How do surveys work?
4 Consumers answer survey based on their memory
Panel ProviderMarket Research team
Self reported data
A quick primer:How do surveys work?
5 Market research team analyzes data to develop an answer
Market Research team
Insight & recommended business action
Demographics● Age?● Gender?● ...
Behavior● Where did you buy?● What? How much?● ...
Emotions / Feelings● Why did you buy?● What matters to
you?● ...
Survey
...Where we thought we fit in
4 Consumers answer survey based on their memory
Panel ProviderMarket Research team
3 Survey sent to consumers through a ‘panel provider’
Why can’t this be based on actual (vs. self reported)
data?
Demographics● Age?● Gender?● ...
Behavior● Where did you buy?● What? How much?● ...
Emotions / Feelings● Why did you buy?● What matters to
you?● ...
Survey
...Where we thought we fit in
4 Consumers answer survey based on their memory
Panel ProviderMarket Research team
3 Survey sent to consumers through a ‘panel provider’
...let’s be a “next gen” panel provider that merges real
data with opinions
...Where we thought we fit in
What data?• Social media likes &
posts• Email purchase receipts• Credit card purchase
history• Amazon.com purchase
history• GPS location history• Web and search history
Opinions how?• Record short video /
audio clips• Take <5 min surveys• Write reviews• 1-1 text chats
Other learnings
Presenting
Share the key insights that led to a decision or answer.
Don’t just share the answer
Example: Equity IdeaWe learned a, b, & c...therefore we want to do “x”
VS.
We want to do “x”. Here is some rationale for why.
Preempt question the audience might ask and prepare
responses.
Don’t bullshit if you don’t know the answer. It’s okay to say need
time investigate it.
1 2
Group work
1. Set up regular recurring meetings at least twice a week
2. Carefully consider if the task is best performed by a group or by an individuala. Everyone wants to participate in decision making, but it is often
more efficient if a single person completes 80% of the task and the group then finishes the rest
3. If there is any tension, discuss it explicitly
4. Don’t take criticism of your ideas personally
5. Humor helps
Launchpad Methodology/Process1. Applying the scientific method to business model is extremely
usefula. treating all ideas as hypotheses prevents attachment to bad
ideasi. also encourages rapid iteration to get to better ideas faster
b. using MVPs as tests of ideas rather than finished products avoids wasting tons of development time
2. Interviewsa. what people initially say is not what they would actually do
i. need to push commitment to see what they actually dob. interviews with experts are a quick way to get a lay of an
industryc. it’s surprisingly easy to get interviews with experts with a warm
intro, student status, and the purpose of learning as much as we can
d. need to clarify customer segment as early as possible to interview the right peoplei. early interviews should focus on figuring out who they are