The New Moats — Jerry Chen from Greylock Partners Notes & Thoughts

The story starts with a quote from Warren Buffet:

In Business, I look for economic castles protected by unbreachable ‘moats’

Defensive moats -> Sustainable & profitable business. In his blog, Jerry Chen from Greylock Partners surveys traditional moats that are used by tech companies. These old moats are getting disrupted while the tech startup today needs to build new moats. Jerry states that these new moats are “systems of intelligence”, AI powered applications. The content of the blog is mostly focused on enterprise applications. Consumer applications can also be inspired by his analysis.

Photo credit: Javier Arce ☄ via Visualhunt.com / CC BY-NC-ND

Incomplete list of old moats surveyed

1. Economies of scale: massive scale to drive unit cost of production down, making it prohibitive for new entrants. SaaS and cloud services are effective in building scale.
2. Network effects: when a new user accrues more value for every other users in the network. (For a more detailed breakdown see detailed notes based on A16z’s presentation here)
3. Deep Tech/IP/Trade Secrets: novel solutions to hard tech problems, new inventions, new processes, new techniques, patents. These would then evolve into unique accumulated insights into a problem or process
4. High Switching Costs: Stickiness is built from standardization, lack of substitutes, integrations to other apps and data sources, entrenched and valuable workflow that the customers depend on. (Vendor lock ins)
5. Brand and customer loyalty: With each positive interaction customers have, the brand is strengthened. However, it is vulnerable and would quickly dissipates if customers lose trust.

Moats are breakable

Moats help survival, which is different from prosperity. In the long run, they can even be traps — see Clayton Christensen’s disruption theory.

These traps combined with massive platform shifts (cloud and mobile) create new openings for new players. Startups should:

1. Attack legacy moats
2. Build own defensible moats

Bottom line: riding the new wave — this answers the question from Peter Thiel’s 7 questions: “Why now?”

Notes: A specific trend in tech:” The use open source is making it harder to monetize technology advances while the user of cloud to deliver technology is moving defensibility to different parts of the product.” Despite these new trends, moats can still be built in the old way. However that means — picking a technical problem with few substitutes — this requires hard engineering and operational knowledge to scale. Current condition in the tech market: market favors full stack companies. SaaS offers integrated application logic, middleware, and database. Customers are increasingly buying full stack technology in the form of SaaS applications instead of buying individual pieces and build their own. Whole enterprise application experience should be evaluated through another dimension — stack of enterprise systems.

The building of the moat

Stack of enterprise systems

Systems of record

This is the database on top of which applications are built. When the data and app power a critical business function, the combo is the system of record.

3 major systems of record in an enterprise:

1. Customers — CRM
2. Employees — HCM
3. Assets — ERP/Financials

Generations of companies have been built around owning a single system of record and in every wave there’s a winner. Applications can be built around a system of record but are usually not as valuable as the actual system of record.

Examples: Oracle, Sieble Systems, Salesforce, Netsuite— CRM; PeopleSoft, Workday — HCM; SAP — ERP/Financials

Systems of engagement

These are the interfaces between users and the systems of record. They control the end usr interactions. The system of engagement changes more rapidly than systems of record. The most strategic advantage of being a system of engagement is to coexist with several systems of record and collect all the data that passes through the product.

Example: Slack, Facebook, Whatsapp, Instagram, Snapchat, WeChat, Alexa, Google Home.

Systems of Intelligence — The new moats

It crosses multiple data sets, multiple systems of records. Building intelligence on a single data source or single system of record can be okay but the position becomes harder to defend against the vendor that owns the data. To thrive, the startup need to combine Oracle, SAP, and other data sources to create value for the customers. Vendor can create their own system of intelligence around their own data. E.g Salesforce’ Einstein.

The next generation of enterprise products will use different artificial intelligence techniques to build systems of intelligence. It can transform applications as well as data centers and infrastructure products.

System of intelligence can built around 3 areas:

1. Customer facing applications around customer journey
2. Employee facing applications like HCM, ITSM, Financials
3. Infrastructure systems like security, compute/storage/networking/monitoring/management

The bottomline is the movement from the sources of data to what to do with the data.

Photo credit: ashleyg via Visualhunt.com / CC BY-NC-SA

Reflections

Questions

These are great conceptual models in thinking about old moats and new ones. Jerry Chen’s article is very high level and strategic and of course, he did this intentionally. The following is a set of questions that I’d like to find answers to after reading his post:

1. Can these systems of records/intelligence/engagements apply to consumer companies/applications? How would that look like?
2. How to build the system intelligence, tactically speaking, how to productionalize AI techniques in ways that are useful?
3. How the system of intelligence layer interacts with system of engagement/records?
4. Are there any examples of companies trying to build system of intelligence that crosses multiple systems of records? The Veeva example isn’t evident. Veeva owns the life sciences CRM and marketing data but where the intelligence aspect comes from isn’t immediately clear.
5. Theoretically speaking, the virtuous cycle built on top of applying AI on system of data/engagement would make the product better but how does that process look like? Is that done through the combination of manual analysis and synthesis of insights by applying techniques and then applying those insights in improving the product or is the AI system generating the insights and automatically improve the product? How does the AI system and the human agent interact in this process?

How this framework would apply to the consumer space

If you start to see information as a different systems of record at the individual level, the comparison isn’t lost. In the enterprise space, the system of record is built on top of customers, employees and assets. At the individual level, the system of records can be built on the following:

1. Employment records: LinkedIn currently owns the employment records for most people
2. Learning & education records: no one owns the records here except of highly fragmented pieces. Transcripts and college/grad school record. Other types of learning records aren’t managed nor connected with the individual.
3. Assets records: Mint has functionality that would somewhat approach “owning” an individuals’ records here but their functionalities are still very very far from actually owning people’s records here.
4. Health records: EPIC systems and various other EHR companies are trying to own a person’s health records but their systems are far from actually owning the records in the space.
5. Social records: Facebook, Instagram, Snapchat, Whatsapp, Wechat and various others social media websites own the social records of an individual. Facebook comes the closest to owning the social records here in the Western world while Wechat owns the social records for individuals in China.
6. Criminal records: Government owned and prepared internal data bases manage the records here.
7. Political records: Government owned
8. Entertainment records: Highly fractionalized. There are a variety of entertainment providers and each own a piece of the pie. There are many sub categories here. Movies, music, TV shows etc. Within each category, the chain again becomes very convoluted. Netflix has emerged as owning TV shows + movies but again due to the power of suppliers in this space, Netflix’s ability to obtain more information about the viewers is limited. Content producers have a lot of power but again there are a lot of content producers and a lot of different types of content producers. In the music category, Spotify, Google Music and Apple music have emerged. But their domain is limited to owning digital music entertainment.
9. Consumption (products & services) records: Amazon owns this space but again they only own a very small fraction of all the consumptions out there, namely ecommerce. Other types of consumption for example — grocery, restaurants etc, aren’t quite captured.

This is a first crack at the thought of applying the framework to the consumer space.

The original article is here: https://news.greylock.com/the-new-moats-53f61aeac2d9

All good ideas are shamelessly stolen and all bad ideas are mine alone. — Jerry Chen