How Google’s crawlers, Uber’s drivers and Instagram’s selfies power the platform economy

| February 5, 2014 | 19 Comments

This article is #4 in an ongoing series on the design of platform business models

What do Google’s crawlers, Uber’s drivers and Instagram’s selfie-clickers have in common? Or for that matter Elance’s freelancers, Dribbble’s designers and Twitter’s tweeters.  In a world of platforms, they create the fuel needed to facilitate economic and social exchanges that power business and society.

More on this in a moment! Let’s set some context first!

There are two defining characteristics of today’s economy: it’s networked and it’s built around information. YouTube’s open-access network allows anyone to create videos but it is the information/data in the video that determines who it is served to for consumption.

We noted in the last article how the movement of goods and services as well as of money is increasingly being determined by the movement of information. Today’s digital platforms enable the exchange of information to power the exchange of goods, services, money, reputation and attention; essentially all social and economic exchanges.


Info + Money consumer-led


Uber’s drivers, for example, announce their availability and coordinates on the platform by sharing various parameters on location, availability etc. Drivers create value units: information about themselves that helps them to be matched to the right consumer. When a consumer pulls out his phone and requests for a cab, he creates a filter i.e. his location. The driver that is most relevant to a consumer’s location and time of request is sent across.

Once this match is made, everything else clicks into action. The taxi turns up, the traveler is taken to his destination, the appropriate charge is transferred out of the traveler’s account and the taxi driver is compensated.

All subsequent exchanges occur once the information exchange takes place.


The information exchange has three key components:

  1. The producer (taxi driver) creates a core value unit (information about availability, location etc.)
  2. The consumer (traveler) sets up a filter
  3. The unit(s) that pass through the filter are notified and one of them fulfills the need

(Some parts of the above interaction are simplified to illustrate the commonalities across platform businesses.)



This is a recurring pattern that we see across information businesses, in general, and platforms, in particular. Units are produced and served to consumers based on how they pass through certain filters.

Even if you haven’t used Uber’s taxi booking platform, you’ve probably used Google’s search engine at some point. Google’s search engine acts in a very similar way.

Google’s crawlers crawl the web and create web page indices (value units). Users type in a query which acts as a filter. Units are passed through this filter and served on the basis of relevance to the filter. (Google’s algorithm includes more inputs into the filter including social signals but this serves to illustrate how units and filters work together.)


Instagram-wielding teens click selfies all day and create core value units on the platform. These units are then served to other users based on network connections (filter) with the creators.


Facebook’s news feed makes this dynamic clearer. Your entire network creates status updates, pictures, comments etc., all unitss added to the platform. Your news feed algorithm, then, acts as the filter that determines which untis go through and are shown to you and which ones are not. This filter is built based on signals you’ve given in the past through your interactions with the news feed content.






Platforms businesses are useless without value units. One look at an empty search results page will rapidly prove that.

No search results

The core value unit is the fuel that powers the platform. At any given point, a platform is only as useful as it’s ability to match it’s units to the filters that consumers have.

The value proposition of a platform revolves entirely around its core value units.

From a producer’s perspective, a platform is:

  1. An infrastructure to create or store units. Android provides an infrastructure for the creation of applications. YouTube provides an infrastructure to host videos.
  2. A marketplace to find an audience for the goods/services represented by the units. Airbnb acts as a marketplace for listings that represent rooms, Uber serves as a marketplace for information that represents available taxis.

From a consumer’s perspective, a platform is:

1. A repository of units where the consumer can discover units based on her filter.

The value proposition revolves entirely around the value units. A platform without units has little or no value proposition. A platform that fails as an infrastructure or a market for value units will not attract producers. A platform that fails to match the right units to consumers will fail to hold on to them.



You may have figured by now, especially given the Google crawler illustration, that the concept of the core value unit is central to all information businesses. Even when information businesses behave like pipes, they transfer value units from the point of production to the point of consumption.

But value units are especially interesting in the case of platforms. Unlike pipes, platforms don’t control quality and quantity of their core value units. Platforms are information factories without control over inventory. They can create the factory floor i.e. provide the infrastructure. They can create a culture of quality control. But they do not have an iron hand in controlling the quantity and quality of what is produced and consumed. The challenge of building and managing platforms is that you do not exert direct control over your value proposition.

As a result, a focus on the core value unit is doubly important if you’re running a platform. Deciding who can create such units, how they are created, what differentiates a high quality unit from a low quality one are all critical design decisions while building a platform. We’ll explore these further in subsequent essays.



The age of the industrial economy accorded inordinate power to those who held the means of production. In the age of platforms, production is distributed. Whether it is the distribution of manufacturing through 3D printing, the distribution of marketing and journalism through social media or the distribution of service providers in the collaborative economy, the means of production are no longer limited to a large company or entity.

With distributed production, the platforms that enable and aggregate this production are the new winners. Successful platforms create the right tools, incentives and rules of participation to attract producers who create units reliably and regularly.

Further, in a platformed world, the people and processes that determine quality and quantity of units are the keys to success. Google’s ability to serve results and solve the world’s problems collapses if its crawlers don’t work well. Uber’s ability to serve consumers collapses if there aren’t enough drivers available within a particular area. Instagram’s ability to engage users in conversations and keep them away from competition collapses if teeny boppers stop taking selfies every time they visit the bathroom.



A Platform’s work doesn’t stop with the production of Value Units.

Successful platforms need these value units to pass through filters as well. This is where data becomes important. 



There is a lot of hoopla around the ‘big-ness’ of data. But big or small, an understanding of Value Units and Filters helps us understand the role of data in today’s economy.

Value Units pass through Filters because of data. 

Consider how a YouTube video is served to you.

The video (Core Value Unit) has various forms of data associated that include information on WHAT it is (title, description), HOW GOOD it is (votes etc.) and WHO it can be served to (explicit content etc.).

The video is served in various ways.

  1. It may show up on your news feed. The news feed acts as a filter which takes into account data on WHAT you have consumed in the past.
  2. It may show up in response to a search query. The search query contains the data that serves as the Filter. You could then slice and dice through your search results further by adding parameters to the Filter (video length etc.)
  3. It may be shown as a related video at the end of a video that you consume. In this case, the similarity to the current video, as well as your past consumption behavior, work together to create the Filter.

In all these cases, the Core Value Units has data and the Filter has data. Units pass through Filters when the data of the two match with each other.



So how do filters work?

Filters, essentially, are data about what the consumer finds relevant. For a location-based application, the location is a filter. For search engines, search queries act as filters. The News Feed creates a filter based on past actions and signals. The Twitter feed’s filter relies almost exclusively on recency.

Filters can pull Units to themselves. That’s what search queries do.

Units can also be pushed through Filters. That’s how the news feed works.

JP Rangaswami recently wrote a great series on Filters that I’d highly recommend.



In steady state, a platform is an engine that works on the following heuristic:


Three things are required to make this happen:

  1. Value Units need to be created
  2. Filters need to work well
  3. The platform needs good data about units as well as filters to ensure the right units pass through the right filters.

In the next essay, we’ll look at how platforms enable this engine.



Google’s crawlers, Uber’s Drivers and Instagram’s selfie-clickers are the new producers in the Platform Economy. Tweet

The new economy runs on data. Data matches value units with filters. Tweet

The Platform Economy runs on Distributed Production and Personalized Filtering. Tweet


To read the rest of the posts in this series, click here


About the Author:

  • Sangeet Paul Choudary

    Yes, the sources of value definitely change. I agree with your observation.

    On a Pipe, you bundle and subsidize things while on a platform, you may get unbundled. On a pipe, you differentiate by owning exclusive access to supply. On a platform, that assumption may break down.

    Search costs definitely reduce on platforms. In fact, post-Google, platforms compete on their ability to provide a better discovery or personalization experience. All that never happened in a Pipe world.

    So yes, and I have a very visual way of showing this which I will share soon, you’re right. When you move from Pipes to Platforms, your sources of value, monetization and competitive advantage completely change. And that’s why so many companies in so many industries are failing. Because they fail to grasp this.

    Thanks for the great comment. I hope we have more such exchanges and happy to connect off-blog as well.

  • Josh Blandy


    I am talking about producer-side curation and consumer-side filtering in the same breath and perhaps the point was as simple as producers need to curate what the consumer actually wants or there is a ‘matching’ problem and the platform struggles, (just like for a pipe).

    Where, in my view, the challenge lies is that if you are pursuing an opportunity that involves migrating a pipe to a platform it is not as simple as ‘copying’ the pipe value chain onto a platform enabling technology.

    In relation to the first part of my comment, the sections of the pipe that contain the most value are not necessarily the same when migrated to a platform, just look at how most print directories failed to migrate to an online search environment when they effectively copied their books (and business model) onto the internet.

    In relation to the jobs-to-be-done lens, my (developing :-) ), hypothesis is that in a pipe there may often be a greater level of information asymmetry that reduces the ability of the consumer to make a choice that lets them achieve more of the outcomes they value. These could be quality and social outcomes as well as utility, which for Uber could be riding in a high end town car in place of a taxi, where quality may have been more of a compliance activity. Hence the value drivers of the consumer filter are inherently weighted differently for a platform than for a pipe because the reduction of asymmetry (e.g. built in feedback loops like consumer ratings, access to multiple suppliers in parallel, etc) on the producer side means they can be.

  • Sangeet Paul Choudary


    This is a great comment and my apologies for the delay in getting back. I wanted to understand this better before I responded.

    Your point about bottlenecks and gatekeepers on the pipe (high costs) is very valid and I intend to explore that in a subsequent post.

    I didn’t quite understand the second point from a jobs-to-be-done lens. Could you please elaborate on that further? I wonder if you’re talking about producer-side curation and consumer-side filtering in the same breath as I see the two as being different. The first is a decision by the producer on what he wants to put out. The second is what gets served based on the consumer’s intent and context.

  • Josh Blandy

    Hi Sangeet, this is a great series of posts. There are a few things that I have been thinking about while reading them in relation to other frameworks.

    One is the change in transaction cost enabled by technology within a ‘Pipe’ that makes a platform more valuable from a ‘Consumer’ perspective. As you have written we are in the early (but accelerating) stage of this change. Many ‘Pipes’ implicitly rely on high transaction costs at different locations in the pipe both as barriers to entry and as an architecture for the cost structure of an industry.

    The other relates to the relationship between the seed and the filter. I fully agree that the ‘quality’ of the seed as it relates to the filter will make or break a platform. However, if you think about the platform through a ‘Jobs to be done’ lens, arguably the value of the ‘Consumer’s’ filter in terms of the jobs they trying to get done is the key to the viability of the platform opportunity. The seed, in this context is all about execution. This can be evident in terms of platforms that are ‘Producer’ built to solve a producer problem and end up pushing seeds that are essentially pre-filtered with limited understanding of how that filter lowers the end-to-end transaction cost incurred by the consumer to complete their jobs to be done.

    Conversely, trying to seed too many filters over time can have the same effect which is why unbundling can be a rewarding strategy in some categories.

    Looking forward to what you write next.

  • Jim Preston

    Seed sounds a bit like those 1950′s SciFi movies with seed pods showing up everywhere. Maybe sperm would be better, they move toward a “platform” participant / user, but both contain the information / DNA.

  • Sangeet Paul Choudary

    Seed quality is everything. You’re spot on.

    BTW, Jim. You totally get what I refer to as the Seed. I’m still in two minds about the best terminology for it. Is Seed the right name? Do you have any other words for it that would describe it better?


  • Sangeet Paul Choudary

    Thanks Harshit! Appreciate it.

  • Sangeet Paul Choudary

    Thanks Yury! More to come.

  • Harshit Taunk

    As usual, an awesome write up.. thanks Sangeet for this wonderful article.

  • про(:то CAR

    Thanks Sangeet for one more nice seed. I do like your aprouch. Economy runs on Platforms. Yury

  • Guest

    Thanks Sangeet for one more nice seed. I do like your aprouch. Economy runs on Platforms.

  • Jim Preston

    I quit Twitter several years ago because the seed content wasn’t valuable enough for the cost – my attention. As I think you’ve mentioned before the cost of consumption goes up as the curation effort goes down. Google and so many others struggle with this issue. From my experience with platform models seed quality is critical. Quality is defined by the market though. Craig’s List doesn’t have the curation quality of say Amazon Stores but is still successful. Seeds on Craig’s List are both ugly and often suspicious (theft, fraud).

  • Sangeet Paul Choudary

    Thanks Steve!

  • Phil Simon

    Haven’t read that book. Calling something a platform doesn’t make it so. Marketers have been bastardizing the term for years now.

  • Sangeet Paul Choudary

    I agree! And then we have this to add to the confusion:

  • Phil Simon

    The platform has no shortage of definitions, as you point out.

  • Sangeet Paul Choudary

    I think of platforms as multi-sided business models rather than the traditional view of technology platforms. The same governance and staging principles apply to platforms as business models whether you build apps on top of it (Android) or upload videos on top of it (YouTube) or create car aavilability information on top of it (Uber).

    Anyway, the point is not to get lost in terminology but to illustrate how information-based business models work. Frankly, there are no ends to definitions for platforms and there is no standard. Annabelle Gawer calls Intel a platform, David Evans calls a mall a platform, and Christensen thinks of any network-mediating system as a platform. So I’d rather avoid that debate and stick to explaining how information businesses work.

  • Phil Simon

    Hmm… are you calling Uber a platform? I don’t define it as one. What can developers build on top of it? Are there SDKs? APIs? Is there an ecosystem? It’s more of a marketplace than a platform.

  • Steve

    Nice one Sangeet. Keep them coming I cant wait for the next essays.