Lec 01

●​ A digital good has the following value-adding factors:

○​ Reproducibility
○​ Transmissibility
○​ Manipulatability
○​ Storability
●​ Unlike a physical good, it doesn’t depreciate
●​ Goods that have now been digitized: currency, books, newspapers, etc. - the value they
held was information - that has been extracted and digitized - it has improved efficiency
and cut costs - the physical means don’t have value anymore because info has been
extracted
●​ Steel market example - percent represents % of iron in steel
●​ US car market - was extremely overserved in 1980 by the premium/luxury focused
players - so Toyota Kia Hyundai entered in the lower segment - and then over time,
entered premium and luxury markets as well
●​ This process of starting with overserved or underserved market and slowly taking market
share in upper market segments as well - is known as disruption
●​ Amazon was a disruptive innovator - started with books only, despite making losses -
took market share of book stores - and then slowly moved upmarket
●​ Google was busy making search better and improving ad revenue and growing Youtube
and Cloud - that’s why they missed out on AI - because OpenAI was focusing on a single
problem statement and single market of AI, unlike Google doing many different things
(all of which were the right things to do) - and then Perplexity came and improved on
OpenAI, solving for the hallucination problem by combining AI chatbot and traditional
search
●​ 6 questions to ask to identify disruptors
●​ Sony and other music producers were selling whole albums when people often only
wanted to listen to one song
●​ With every disruption, consumer behaviours are changed - new markets and thus new
revenue streams are unlocked -- the entire ecosystem is improved
●​ Disruption doesn’t mean that incumbents can’t respond - incumbents have responded
successfully in certain cases - will be discussed in last class of course

Lec 02
●​ “Everything to do with AI is an efficiency game” - not necessarily
●​ AI is both a differentiator and a cost leader - it gives network effects to build scale, allows
increase in margin through differentiation, and cutting costs
●​ What is intelligence? - task execution, store information, problem-solving, reasoning,
exact decisions, creativity/ideation, context setting, analysing data, adaptability, learning
○​ None of these are exclusively human anymore, but all operating in silos
●​ ML is a subset of AI - machine’s ability to learn from past data, process of figuring out
right model is ML -- this constitutes AI - AI has decision making ability, there is a decision
making layer
 ○​ So phone unlocking process is AI because decision making is involved; the ML
part is when you’re setting up the face id and configuring it through different
angles of your face
●​ Counterfactuals - causal claims are based in these
●​ House of AI/analytics framework - data is the core of whatever you do with AI, apart from
RL side of business
○​ 4 pillars aren’t mutually exclusive, for example GPT has a bit of prescriptive and
predictive as well
●​ Fairness is subjective - everyone’s definition varies - 13 different types of fairness
according to research - depending on business context, your fairness will differ

Lec 03
●​ Claim: Gen AI is only a sequence of predictions
●​ Logistic and linear regressions are models - and it can have many parameters with many
weights
●​ So when we say chatgpt has 5 billion parameters, it means it has learnt the weights of all
these 5 billion parameters and is giving an output based on the input of each of these
parameters that the user provides
○​ Each word you input is a vector (parameter) and is input into the equation and
accordingly chatgpt generated a prediction as output
●​ Learning these weights is the difficult part - the computational aspect which began to
develop decades ago - but the computational power to pull this off only came up now
●​ And when we say gen AI learns with every prompt - it doesn’t mean that the weights
change in the backend, it’s just that it applies more filters and transformations based on
what it learns
●​ RAG: a model that is only functioning in a particular context, you can fine tune it further
for your use
●​ Open weight models:
●​ ChatGPT only makes money through subscriptions - ChatGPT is an application, and
GPT-4 is a model
●​ Each AI company has many such models - applications are called ChatGPT, Claude,
Gemini and so on
●​ AI companies are trying to monetize the models itself by selling it as a backend product -
so anyone can make applications of their own using the GPT models
●​ Open weight models are where they give the model for free, but telling you to incur
deployment costs - so you have to have the infra and resources in place as a client to be
able to deploy it
●​ The other alternative is the companies that are charging you
●​ Deepseek is keeping the model open but still has a website through which customers
search for specific things - so they still collect data on you through their interface that
they can monetize - many avenues for monetization in this space
●​ Prompt structure fundamentals - need each of these in every prompt - otherwise it will be
wrong
●​ Zero-shot prompt: doesn’t give any data/example from past
 ●​ Few-shot prompting: giving example and asking for answer in a different context
●​ Chain-of-thought: step-by-step thinking - Khanmigo does this - instead of giving answer
directly
●​ A lot of organizations follow this mantra: “build for competitive advantage and buy for
parity”
●​ OpenAI incurred $80bn to develop ChatGPT 3.5 - one-time cost, monetized forever - so
is it right for Khan Academy as a non profit to pay a recurring price to OpenAI for a
product that OpenAI has already built?
●​ Khanmigo
●​ Fundamental shift: - adaptive tutoring - active mode of learning - changes org structure,
type of talent you hire, cost structure also changes; organization goes from being
content-centric to AI-centric
●​ But another way of looking at it is that it is an extension of existing offering - they were
already offering personalized, adaptive learning experiences -- so then where is the
fundamental shift?
●​ If you don’t move first, someone else will - and someone who has profit as incentive
moves first, then you will never be able to venture into this as non profit - by moving first,
you can control how this market shapes up, in accordance with your missions - so worth
taking the huge risks associated with deploying such an early-stage LLM
●​ AI does well when people are unsure about what needs to be done - so in a field like
customer service, highly skilled people don’t derive much value - but in science, top
scientists derive the highest value addition from AI, because scientists are trained to be
unsure in their profession

Lec 04
●​ Data’s bias is induced because you’re only training it on pictures of criminals and not
showing any non-criminals - so the data is trained on only black people that are
criminals, and not exposed to any black people that are not criminals
●​ Explainability is important - should be able to understand what is happening inside a
model - why is it making these decisions or giving these predictions - as opposed to a
black box model where you don’t know what is happening inside
●​ There is often a tradeoff between accuracy and explainability - but not always -
sometime explainable models like data trees are better than neural networks in certain
use cases - generally in language models, blackbox is better than explainable - but it
depends from situation to situation
●​ Neural networks and random forests are black box models
●​ Can also do post-hoc analysis of explainability - i.e. let the model learn on its own and
then once it’s ready analyse it and build-in the explainability - but this is a very costly
process
●​ ChatGPT and Gemini are blackbox models - that’s why hallucinations are difficult to fix

Lec 05
●​ Use cases for metaverse
○​ Simulating/training/upskilling - safe/scale/cost reduction
 ○​ Gaming - immersive/scale
○​ Social interactions - competitive use cases: where there is value in neutralizing
identity - interviews to eliminate gender bias, cultural bias (related to accents);
potentially in dating - can get to know more about person without revealing
everything and before meeting; collaborative use cases as well
○​ Fitness
○​ Shopping/retail - Nike did it
○​ Ads/spaces
●​ Metaverse is 3D experience, immersive, synchronous
●​ Early adopter use cases were gaming and then training, in that order

Lec 06: Blockchain

Lec 07:
●​ Where is value being created and where is it being captured?

Lec 08: Tech Led Products and Product-Led Business Models

●​ First thing to assess: customer value proposition
○​ What are the problem(s)?
■​ Not able to access data whenever/wherever you want;
■​ Physical data loss;
■​ Sharing and collaborating on data
○​ There many players already solving these, but only one or two - not all. But these
problems occur in tandem. So Drew Houston offered an integrated solution that
solved all 3 problems - and the target customer were non-tech-savvy -
“individuals”
○​ Trojan strategy - get people used to the Dropbox product so that they keep using
it - forcing enterprises to adopt as well - “land and expand” strategy - so he
always wanted to get into enterprises
●​ After identifying problem: create a differentiated solution
○​ Differentiators: reliable, simple-to-use, incremental upload, bypass firewalls
■​ Every feature is completely in service of the problems of the customers
they have identified
■​ The key for differentiation is detecting the unarticulated user needs for eg.
if you ask users, they will say the problem is that “upload is too slow”, but
the unarticulated user need here is that internet capacity was low, so
needed to optimize for that
●​ Next, what is the technology and operations management plan? (How will they build it?)
○​ Hosting their storage on AWS
■​ Always build for advantage, buy for parity - investing money to build their
own storage capability will not help them build anything better than AWS -
no competitive advantage - and would cost a lot - and would take time
(hurting GTM) -- so use AWS services despite the high level of risk
involved at the time
 ●​ Then, GTM:
○​ Feedback from early adopters vs innovators - early adopters are the ones who
care about the problem - innovators just like trying new things
○​ Things that didn’t work - search (users didn’t know the problem) - partnerships
(because B2B has bureaucratic inertia, and managers who make business
decisions are not the employees who actually use it, so actually good,
user-centric products don’t work; enterprises also have inertia due to data gravity
- data migration is a large undertaking and risky - the risk of losing data is a major
detractor)
○​ Virality - R0: coefficient of virality - for every customer, they are recommending it
to R0 more customers - and these recommenders in the case of Dropbox also
became their CSR: because if someone is sharing a doc with you on dropbox, if
you have any tech issues, you reach out to the sharer first
●​ Financials:
○​ Loss $0.75M
○​ Contribution margin: revenue per paid customer: $120/year, cost per paying
customer: $3.18 * 12 = $38.16; cost per free customer: 0.11 * 39 * 12 = $51.48
■​ here we got 39 by calculating the number of free customers being
subsidized by 1 paid customer (and 97.5% customers are free, 2.5%
customers are paid), so = 97.5/2.5 = 39
○​ Also this $51.48 is essentially a part of your CAC (because it is how much you
are spending to convert a free to paid customer)
○​ So contribution margin is 120 - 38.16 - 51.48 = $30.36
○​ Projection: LTV/CAC for enterprise (based on estimates)
■​ Conclusion: enterprise would not make money, as of 2010
●​ Framework for making this decision: qualitative and quantitative factors; internal and
external factors
○​ Diversification only makes sense when company is generating FCF - in 2010,
Dropbox was a very young company - didn’t need to diversify as that would dilute
stake
Lec 09: Platforms
●​ Physical good vs digital good - from lecture 1
●​ Song (information) is the value-creating entity; CD is the value delivery mechanism
●​ CD ripping: mp3 format started in 1990s - were able to copy song from CD, extract it,
and put it on PC
●​ CD ripping had large-scale economic implications - it was able to separate the
information from the value delivery mechanism
●​ Industrial (traditional) economy - supply-side economy of scale
●​ Digital economy - demand-side economy of scale
●​ Uber in India is not a platform economy - a taxi service - not sharing cabs here, unlike
US - here in India, it’s a regulated market
●​ In a digital economy, both sides are demand-sides - both are demanding something from
the other - and platform is making money off that information asymmetry - both sides are
demand so you get to make money from both sides
●​ Digital economy: low marginal costs, demand side-economies, network effects
●​ What is a platform? 3 examples
●​ [Link]: employers/recruiters, job-seekers
●​ Benefits to job-seekers: rich information/volume; convenience: time-saving and saving
transaction costs; no spatial/temporal/sociocultural/socioeconomic implications
●​ Benefits to recruiters: same, more or less
●​ Naukri as a platform: builds the infrastructure and components, rules of pay etc.
●​ Value of Naukri is contingent to number of people in either side
●​ Cross-side network effect: preference of one side for more of the other side
○​ Positive: App store
○​ Negative: users don’t want too many advertisers on platforms
●​ Same-side network effect: preference of one side for more of same side
○​ Positive: social media for users (more users, more content)
○​ Negative: whenever there is a compete scenario (too many advertisers like you
on the platform)
●​ Amazon:
○​ Book publishers, book buyers (consumers)
○​ Where amazon is a reseller, it’s not acting as a platform -- in the value chain,
both arrows have to be pointing to the platform (in this case, amazon)
●​ In the platform model, there is absolutely no transfer of goods through the platform - the
good flow directly from publishers to buyers; there is only transfer of information from the
platform
●​ Going from resource ownership to resource orchestration - in resource ownership, you
have to make profit (markup) by optimizing internal supply chain - platforms which do
resource orchestration, have to optimize value of information across all the actors in the
value chain - maximizing value of ecosystem, not just internal value
●​ Example 3: Google
○​ Google is purely an advertising company - everything they do is in service of ads,
not search
○​ 3 sides to google: users, advertisers, websites
○​ Users: information, convenience, free
○​ Advertisers: global audience, target ads, deep attribution of every marketing
dollar you spend (know exactly which ad click generated how much)
○​ Websites: monetization, visibility, deep personalization
○​ 57% of Google revenue from Google search, another 10% from Google ads
●​ Mono-homing: android vs ios, playstation vs xbox - expensive/difficult to have both
●​ Multi-homing: ride-sharing - booking multiple rides at once - cab driver also looking at
multiple apps at once
●​ Switching costs - often expensive to switch
●​ Subsidizing means you’re charging one side below the actual marginal cost of serving
that side
●​ Who to subsidize?
○​ Price sensitive side
 ○​ Side that asymmetric data network efforts: both users and advertisers value each
other, but advertisers value users exponentially more than users value
advertisers - asymmetric network effects
○​ Side that demands quality - ladies’ nights at nightclubs - matrimonial websites
charging men, not charging women
●​ Mobilizing the google platform -
○​ Embeds
○​ Adwords auctions AND ensuring quality of advertisements: by assigning quality
score to every advertiser; and then final rank is a product of absolute bid amount
x quality score
●​ First price auction works well in open bidding systems
●​ But in a closed bidding system: winner’s remorse in first-price auction - if you get it at the
amount you first bid, then you think you overpaid - so it’s a second-price auction with the
coefficient of quality

Lecture 10: Platforms contd. and Simulation Game

●​ Irrationally burning money is a rational strategy if you believe it’s a winner-take-all market
●​ How do you decide if it’s a winner-take-all market?
○​ Are multi-homing costs significantly high for at least one user side?
○​ Does this side experience significant positive network effects from the other side?
○​ Are the opportunities for differentiation limited? - if there is significant
differentiation, there might be a compelling reason to multi-home - which would
reduce the winner-take-all
●​ Winner-take-all doesn’t mean 100% market share - it means a significant gap between
first players and next
●​ Is Search (Google) a winner-take-all dynamic?
○​ Consumers are the more significant side (due to network asymmetry - advertisers
want users more than users want advertisers)
○​ So for consumers there is no significant costs for multi-homing (costs here are
money, time, convenience - but not the quality of search)
○​ There is room for differentiation
○​ Bing took share from Google in 2010-12 - was 30% market share in the US - they
did vertical search strategy - specialized in travel data - signed deals for
exclusive access to flights and hotels info, that Google couldn’t display - Google
responded aggressively - went and bought the companies they had an exclusive
deal with
○​ Google is going after competitors in all vertical search specializations (Dropbox
trying enterprise search, Google responding with Gemini for enterprise; Pinterest
for image search; Amazon for shopping search) - so Google keeps coming back
and fighting - because it’s not inherently a winner-takes-all market - but they keep
engineering the network effects by working on product quality and responding
aggressively from competitors who take market share

Lecture 11: Merchant Models; Myntra Case

 ●​ Lower mileage/transaction costs; variety; discounts
●​ Traditional/spatial boundaries; richer data; returns; privacy (for discrete purchase
categories like hygiene etc.); price comparison and transparency; allows for better price
discovery
●​ Unit price of most items is lower online than in offline channel
●​ Cost of changing prices: menu costs; menu costs offline are very high, need to change
labels, train people, making changes to PoS; online it is very easy to change prices
●​ In online: price changes are more frequent, but quantum of change is much lower
●​ But variation of prices across retailers/platforms is greater online than in offline channel
●​ So there is greater price discrimination online → so then we can’t say that price is a
source of competitive advantage between online platforms
●​ For higher selection/variety, need to have supply chain optimized and ensure that all the
range of products are discoverable
○​ Ensure that product discovery workflow is meaningful - so that you can quickly
sort through so many options - pages, store-in-store
●​ Customer experience: most important aspect is delivery and return, not ordering process
as much
●​ Fulfilled by Amazon - gave rise to a whole new business “Amazon Transportation
Services” - PrimeAir flights - and they lease trains
●​ Myntra case
●​ Adoption (devices/internet); customer base is large; retail very unorganized; structural
issue; channels
●​ Challenges of Indian e-commerce: trust, last-mile, payment (COD), low ticket size/ASP;
fashion-specific challenges: returns; experience/trials; seasonality of fashion trends so
inventory has to keep up
●​ 3 metrics to assess if Myntra is successful: price, selection, customer experience
●​ Not competing on price
●​ Experience: virtual try-on technology; lookbooks/themes/professional stylists; return
policy for customers; UI/UX of app → clearly helping them as 70-75% of their revenue
from repeat customers, large number of active users
●​ 40% own inventory; 50% consignment model; 10% marketplace
●​ Doing all this because to be an e-commerce player you need to have large selection - to
serve the tail - so need to be in all ranges of the market - and to be able to have
competitive advantage i.e. serve the tail at cost - so need different models for each
●​ But they have a logistics problem - high fulfilment costs and thus unprofitable - so for
them there is upside to getting acquired by Flipkart - because Flipkart can help solve the
logistics problem
●​ Takeaways:
○​ Price is not a sustainable advantage
○​ Optimized supply chain is critical
○​ Product discovery is essential with a wide assortment
○​ Ability to serve the tail at cost with the right level of service is crucial

Lecture 12: Simulation Debrief; Creator Economy; Melissa Wood Health Case
●​ Scenario 1: Likely winner-take-all market; quality is more incidental here, only matters in
the beginning; keep bleeding till the market tips over
●​ Scenario 2: Not a winner-take-all market; quality is more important throughout, so need
to burn a lot of money; developers are the money side, need to attract them, then
competitor will copy you, developers will switch, then again you’ll have to get them back;
here you need to burn more cash in the beginning, returns only come much later; the
player who effectively monetizes the most number of developers is the one that wins
●​ Strategies to influence/engineer network effects: Google does this, creates network
effects by giving high-quality products that are free; even Amazon reviews are a way to
create network effects
●​ Strategies to influence multi-homing: penalty strategies like lack of compatibility (Xbox
players can’t play with PS players; Apple’s lightning charger when rest of the world is on
type c); incentive strategies like add-on benefits (Amazon’s Prime program for
consumers; incentive to suppliers through Fulfilled by Amazon - telling them you don’t
have to worry about logistics, and also given them better commission rates on every
transaction); Amazon also uses a penalty strategy on sellers is purchase price parity:
can’t charge a lower price on any other platform - so you sell the most on Amazon
●​ Information transparency promotes multihoming
○​ Eg. Groupon: LivingSocial was scraping the Groupon website and then only
advertising the most selling deals, so users were switching to LivingSocial i.e. it
was a high multi-homing market - so Groupon decided to intentionally
misrepresent the number of deals being sold - so instead of showing exactly how
many coupons were sold, they would show “x+”; and started randomly reducing
the numbers displayed by a random percentage that was randomly generated
daily - and LivingSocial’s strategy stopped working so Groupon was trading on
information asymmetry to beat competition - this may have impacted their own
user revenue a bit as well, but they made that tradeoff to reduce the multihoming
●​ Melissa Wood Health
○​ Deconstructing this business model and is it even a business?
○​ To break down any business model, ask these questions:
○​ Who are the stakeholders? Users (followers, paid subscribers); Brands
○​ How is value created in this business and for who?
○​ For followers: creating aspirations; giving tips for brands and lifestyles; discount
codes; emotional (parasocial) value and convenience/time/money-saving value
created through content; value derived from community
■​ Note: the convenience of at-home workouts is not necessarily unique to
Melissa
○​ For subscribers: access to Melissa Wood method (unique type of
workout/lifestyle); everything else she gives to followers; personalized workout
experience;
○​ For brands: brand imagery/awareness; sales; high conversion at lower cost;
Melissa posts more than stipulated if she likes the product (brands love her);
partnering with smaller, early-stage brands so they need good quality feedback
which they will only get from consumers who are deeply engaging with the
 product which Melissa can drive through her users because they trust her so
much
●​ How is value being captured?
○​ From followers: attracting brands to her business; also followers help increase
followers and increase subscribers → so followers are creating indirect value
○​ From subscribers: subscription fee
○​ From brands: partnership model (upfront money for a specified number of
posts/stories) or affiliate model (revenue-sharing)
●​ Is this a product business or a platform business?
○​ It is a platform because: there are 2 sides with distinct needs (followers and
brands); there are network effects; Melissa Wood (the platform) acts as
intermediary providing the rules of engagement (telling you which products to buy
or not)
○​ There is a product business in the follower/subscriber model where the content is
the product and it is operating on a freemium model
●​ Doing the LTV/CAC calculation and discounting, we find that monetary value of every
future subscriber is $185
●​ To ascribe monetary value to followers: taking brand revenue and dividing by number of
followers (after subtracting paid subscribers) = $0.18; followers also become subscribers
= number of subscribers/number of total follower = 16.5%, so value of this is 16.5% *
current number of followers * $185; so total value per follower per year is $32
●​ And there is intangible value created by followers that we aren’t able to calculate
(network effects etc.) → do these intangibles compensate for the difference between
tangible subscriber value and tangible follower value (185-32, as calculated above)?
●​ Her subscription growth is plateauing, churn rate is about 5.8-6% → leaky bucket
problem
●​ So to keep growing business at the same rate while making the same amount of money,
she has to grow followers

Lecture 13: Guest lecture: Ganesh Krishnan, Serial Entrepreneur; Founder of TutorsVista,
Bluestone, BigBasket
●​ For any business model, there’s a north star, that moves the needle most substantially
for them - for IPL, it was converting non-viewers to viewers
●​ Easiest way to accelerate growth - convert non-consumers to consumers
●​ Easiest lever for BigBasket’s growth - how do you convert the people
●​ Subscription business model - reduces sticker shock; increases TAM because
consumers would rather pay smaller amounts on a periodic basis rather than a larger
upfront cost at once; CAC reduces substantially; can track consumer behaviour more
constantly and capture it as data so allows for better targeting; offer greater value to the
consumer than they actually consume, so charging them for more than they’re
consuming
●​ Breaking into duopoly - typically has ‘winner-take-all’ dynamics
●​ How did you use the answer I gave you? Critique it part-by-part, see if it addressed
specific problems listed in the case itself
Lecture 15: Teams vs Zoom; Incumbent’s Response; GE Case
●​ Why did Zoom become successful during COVID?
○​ Ease of use; reliable
○​ New segment of customers emerged during COVID: education, religious
institutions, startups, personal
○​ Freemium model - free for 40 minutes
○​ Feature-rich
○​ Microsoft was busy competing with Slack (on the collaborative workspace model)
so didn’t try to compete with Zoom
●​
Teams Zoom

Target Consumer Business users/Enterprises SMBs/Individual

consumers/Educational/Religiou
s

Value proposition Integration which supports Simple and reliable video

collaboration conferencing

Business Model Bundled or $9.99 without Freemium

bundle

Security Security and privacy Moderate security

Internal focus, within External focus

ecosystem

Difficult to use (need an Easy to use (no need for an

account etc.) account)
●​ So, are they even competing? They clearly serve different segments of users
●​ ISB is overserving recruiters: because graduates expect higher pay than what recruiters
want to offer and recruiters need lesser than what we are offering in terms of skills
●​ Is Zoom a disruptor?
○​ Targeting overserved and new markets? Yes
○​ Simpler/cheaper/more convenient? Yes
○​ Business model Moat? Moderate
○​ Tech enabler? Unclear
○​ Incumbents propensity to ignore? Yes, because they believe there is less money
in the business model, it won’t deliver on the metrics of success Microsoft look at
that
○​ Is offering lower quality than incumbent’s as measured by incumbent’s matrix?
Yes
●​ So, there is a higher tendency for Zoom to be a disruptor in this market
●​ Late mover
●​ Realplayer: first streaming service
 ○​ Real server which is housed on Windows OS and there are stream creators and
stream consumers -- stream creators are the money-side
○​ Real server works only on Windows OS -- so there is an overlapping consumer
base with existing product of Microsoft
○​ So they will envelope this new product onto existing features and create
Windows Media player - as part of their operating system - so now money-side
switches immediately - because now there is a double margin problem - creators
being charged for Windows OS and Realplayer when they can just pay for OS
and get the same capability
○​ And this switching will happen even in winner-take-all markets with high network
effects, as long as this multi-platform bundling i.e. envelopement happens
●​ Pre-requisite for platform envelopment is either partial or complete overlapping customer
bases - this is an extremely aggressive strategy and not easy to do, need a lot of money,
resources, and time
●​ 2007 iPhone was launched: 5 companies died - LG, Nokia, Moto, Sony, Blackberry;
Samsung survived; which tech company was spooked when it was launched? Google
○​ Google was spooked because economy was shifting from browser-based to
app-based economy - earlier you were going to Google to search, now you have
an app for the same thing → that was the hypothesis Apple was operating under;
from searching for weather/game score on Google to instead using the weather
app/football app for the same purpose; so if this happens, then browser is
disintermediated
○​ At the time of launch: Apple said it is a closed ecosystem with an in house ads
platform iAds → so Google no longer has access to ads as well
○​ Essentially, Apple was trying to envelope Google
○​ Google responds by investing in Android → and the killer move was that they
made it open source → so as many developers can be onboarded
○​ They knew they will win both browser economy and app economy by focusing on
this open distribution
○​ They also invested in Admob which became the enabler of ads on all android
apps
○​ Then Apple turns around and says it will be open ecosystem as well and invest in
admob so even they use the same platform for ads
○​ Google saw this flip from browser-based to app-based happen 7 years ahead
and started working towards it
●​ Zoom is trying to copy Teams on everything - from a market strategy POV, it makes no
sense; looks like they don’t know what to do, because you can’t win against Microsoft -
but they’re possibly trying to maximize their exit value: if they build as many assets as
possible, they are increasing the potential valuation they can get
●​ Feature parity doesn’t matter because the audience is a B2B market - Zooms is better
than Teams in video conferencing, but it doesn’t matter - because no one cares for
customer insights in enterprise-serving businesses
●​ In business, any problem that can be solved with money is not really a problem
●​ GE
●​ Instead of being a one-time product sales company, become a recurring service sales
company and increase your margins
●​ Digital transformation:
●​ IoT: ability to capture data through instrumentation of a physical object and then transmit
data
●​ So GE decides to do IoT - why? Now they can offer their customers:
○​ Predictive maintenance
○​ Revenue-sharing model through outcome-based pricing → if I increase your
revenue by x%, I get y% of that
●​ And revenue-sharing model is great for margins because it has no absolute ceiling - if I
help you achieve more and more revenue, then there is more money in this
●​ But the problem with solution-based business models is that you can’t scale this -
everyone requires a unique solution, it’s a linear scale not an exponential scale - very
high cost
●​ So to make it scalable you make a platform - one side: companies requiring services -
other side: software service providers who have specialized domain knowledge
●​ Incumbents are judged differently by investors: they are looked at for current period
profits, markets penalize for doing new things; for disruptors: future profits are looked at,
valuations reward doing new things
●​ In any market, you will see that both incumbents and disruptors survive - it’s not that
incumbents die
●​ Possible incumbent’s response:
○​ Fight back: NYT/MSFT/GE
○​ Double down: Disney/MSFT/WH Smith
○​ Retrench:
○​ Move away: Fuji (from photography to medical imagery)