Advanced Machine Learning, Data Mining, and Online Advertising Services
Here, I'm sharing some of my experience in the area of data science. In the last four years, I've been working for a tech few companies in SF, Victoria & Vancouver as a data scientist. My goal here is to discuss the type of services that data scientists can provide and a few go-to-market strategies to sell those services.
From my experience, a data scientist is a person who analyzes a company's data to increase their core product proposition value. This improvement should directly/indirectly impact the company's economy/revenue. Here, data can be any log collected from the company business activities. For instance, for a dating website the data is users activities on the website such as how user explores/find other people profiles, message them, repond to their requests and so on. As another example, for an advertising tech company the data is performance logs of how users responded to displayed ads. Or for a game copmany, the data is how players interact with game and progress in the game to the next level.
Therefore, the focus of a data scientist should be to increase the core product prop value. This requires collecting all necassary data for achieving that. A data scientist takes a scientific & data driven approach to achieve such a goal. For doing so a data scientist needs to deeply understand nature of the company buisiness by talking to C-level execs & product owners. A data scientist is a software engineer who takes data driven approach to tune/optimize the product. You can read here more on this topic: who's a data scientist?.
A data scientist needs to have a degree in computer science. They need to have a strong background in algorithm design and data structure since they have to push their predictive models to production and should be able to implement a scalable/efficient version of them. They also need to have a strong background in probability and statistics. One of the challenges I found was that most people have either a strong math background with less software engineering expertise or the other way around a strong software engineering background but not very strong in machine learning space. However, I beleive both skill sets are necessary to deliver world-class data science service.
I have seen CEOs/CTOs who view data science as a fancy role. Part of this is because the area is still new to the tech industry. It's odd to hear from C-level people from a tech company to say "we don't need software developers". This is because after a few decades we have a clear understanding of the software developer roles. However, this is not very much the case for the data science roles. Talking to C-levels, I often have noticed that they don't see having a data scientist in their team to be crucial. I have found this even with companies which can clearly benefit from such roles due to their products nature. Some of the reasons behind this are listed below:
The first issue makes sense as a startup first priority is to survive by solving the product-market fit serach problem. So, optimizing their product before solving prodcut-markey fit problem is an imnmature optimization. If you are providing a data science service as a company, it's really important to understand the right time to approach a company for selling such a service.
Addressing the second issue should be done through education. Data scientists need to pitch C-level execs and show them successful cases how a data driven approach added value to other companies core products.
To address the third point, I usually pitch companies where there is a close connection between product proposition value and machine learning 7 data mining areas. If a company's product is online fraud detection, obviously the CEO cannot ignore the importace of having in-house data scientists. This is a must as they need to keep inventing and improving their fraud detection algorithms in order to stay ahead of competition. Another good example for this is performance-based ad tech companies whose revenue is tightly coupled with how well their ad serving/buying algorithms perform. Thus, it's important to understand the nature of a company's product and see if/how machine learning and data mining techniques can bring value to the table.
Addressing the last issue can be done by having short and long term plans for data science projects. I elaborate this next.
Let's say you are pitching a CEO for a data science service you are providing. As the first step, you need to sit down with them and carefully listen to understand their product, their pain points and struggles, and find out low-hanging fruit areas that can be attacked in order to add values to their product. Next, you need to formulate a set of essential Key Performance Indicators (KPIs) which they need to watch regularly in order to move their ship in the right direction. This requires collectig data and computing those KPIs and show the KPI's on a dashboard such that exec people can watch how their company is doing in terms of product performance over time. Finally, over time as long term plan you can introduce machine learning and optimization algorithms to help them tune their product and increase their revenues.
So going back to the original questions. Yes, for some buisinesses the data science is nothing except a fancy role! Thus, it's important to think about company's business and core product and find out if/how machine learning and data mining services can increase their product ROI.
If you have any comments or story that you'd like to share, use the comment section below.
If your buisiness has a data problem and you are looking for an advanced data solution, contact us at info@AIOptify.com