What is a Product — A Strategy Perspective — V
Products provide us with economies of scale. We achieved these by spreading over the customers:
- Marketing and selling costs
- Product development costs
- When properly planned, even infrastructure costs
Are all the companies able to achieve similar levels of economies of scale? Why are some companies able to serve such a large-scale audience while others are struggling to scale up even with a moderate set of customers? There is the choice of business; equally significant is the choice of technology. We will conclude this series of articles with how technologies have achieved the modern scale for business.
A few students did the following:
- They crawled the web to find all the pages they could reach through cross-linking of the documents
- They put all the words in an electronic dictionary
- They gave the documents a higher rank if that documents had links from multiple other documents (PageRank).
The outcome is Google of 1998. Some of you may be imagining recreating this now out of your laptop. With all the new technology available, this product can take a week and a month put to actual use. Using today’s technologies, one can easily create the Google of that period. People who think AI/ML is a recent invention, let me reassure this was the AI of the late 90s. The amount of computing power needed to do this task would have been hard to conceive then. Google went on building its hardware to keep the costs to a bare minimum. Google continues to design its hardware like Tensor Processing Units (TPU) for Machine Learning (ML).
The March of Technology
Let’s see how good was the above system. Given a word, in the exact form, it returns all the documents where such a word exists. For such a system, encircle and circle are two completely different words and have no relevance to each other. Google did better than that as they had some more processing around such scenarios. Modern AI does not search for words. It searches for concepts. Google utilizes its mighty processing power to take all the web pages of the world; and tries to build a mathematical model for ideas or concepts. Instead of identifying named concepts, it creates unnamed ideas available in a latent sense in every sentence. In a document, the latent idea gets encoded as word order, character order, and sentence order. The modern AI networks model the ordering or word associations and learn from the billions of documents. At the end of the exercise, you get a concept model or knowledge space. Your search now places your query in the knowledge space and tries to find the closest match. You can ask questions and find answers as shown below:
Q. Where did I leave the phone?
A. You left the phone on the table.
A context is built into image and text algorithms today. I leave the rest to discerning readers to read more and learn from the research in the field. Processing power and choice of technology made these complex problems tractable in modern technology. We call all these a blanket AI-ML. Although some concepts and algorithms are from old research, they were not good enough to be tried due to a lack of hardware resources. The hardware resource requirements have gone up so significantly that the search for Artificial General Intelligence (AGI) as we know it today has become a purview of a very few large companies. An organization that utilizes superior technology providing the best value for resources can only achieve economies of scale.
As you can see, improvement in search needed the analysis of all the web pages. Search technology also requires a large number of search strings. People gravitate toward great products when they see the value of the outcome. Google provided just that. But it also collected information about documents the people accessed from the search results. Were they looking for specific websites? It helped them target advertisements to customers. And that became their business model for almost a decade initially. They needed more data and content. They provided avenues for data collection through products like Gmail, Blogger, YouTube, Google Sites, etc. Google encouraged customers to add trackers to their websites. In return, the customer accessed great products for free. Google was not the only company. Any company that provides customers free services looks to harvest customer data for information extraction and exploit the data to generate revenues. Here is a simple schematic representation of the same.
In short, companies employ very crafty means for collecting data. They encourage users to manage that data and prune it with relevant information. Let’s think of Google Maps. Google collects map data from users’ phones automatically when users start using Google maps. They suggest users act as local guides and submit reviews, update maps, and submit their photos. They also provide gamification to entice users and send gifts to add to those. The people who add AdWords scripts to their sites are rewarded with a small reward when their site generates a conversion lead. Essentially, Google encourages users to provide data and manage to improve its significance. Some users have made full-time careers as content creators and YouTubers in the social networking era. With well-managed content with clarity, Google can now provide additional services. For example, by analyzing the boarding pass documents in Gmail, Google provides a travel solution to customers. Since it could search airline, hotel, or travel booking sites, it could correlate and provide an end-to-end travel solution. Google tracks job search patterns and provides job postings as specialized services. Not just Google, any organization that has users’ data today is in the business of exploiting the data to offer additional value-added services. That’s how they plan to scale their offerings. In short, that is their definition of platforms.
The Modern Platforms
The modern platform is a crowd puller. Get the user to spend almost all the time on the platform and not look for any other solution. The best of the breed is almost over. Amazon started as a prime mover in the IaaS space, Google and Microsoft in the PaaS space. Initially, Amazon was lagging behind the infrastructure of platform services that Microsoft and Google provided. Today, all three cloud platforms AWS, Azure, and GCP provide almost similar one-to-one matching services. They want absolute vendor lock-in. Once a customer uses one of the technologies, they do not want the customer to shop around for the best of the breed. SalesForce is investing in Identity and Access Management to dissuade customers from looking at Oracle. With cloud-based services, the technology barrier to switching is slowly disappearing. Hence, the vendors are making sure there is something available for the customer whenever he needs a feature. Zoho provides a solution for most ERP and CRM processes. Adobe provides services for all publishing needs. In short, organizations are working closely with the customers to capture the best practices of their operations and provide services to the whole community of customers. These are the new-age platforms that can provide economies of scale.
We realized products provide the economies of scale that businesses require for profitability. Enhanced sales and marketing, engineering, infrastructure, and the right technology contribute to the overall success of a product. These can be considered the true pillars of building a profitable software product.
Note: Sambit Kumar Dash is a founding director of Lenatics Solutions Pvt Ltd, which provides product management services to businesses for Sustained Competitive Advantage. You can reach him at: email@example.com