5 Minutes with an Analyst: Misha Plotkine of MobyTrip.com
Misha Plotkine, founder of the algorithm-driven travel application MobyTrip.com, spoke with Upside recently about how analytics affects his company's growth.
- By James E. Powell
- October 6, 2016
Analytics is driving innovation in many industries today. Misha Plotkine is the founder of MobyTrip.com, a comprehensive end-to-end trip planning application. MobyTrip uses analytics to generate itineraries based on users' preferences, and it includes real-time collaboration, route optimization, and adaptive scheduling. Plotkine spoke with Upside recently about his work and how analytics affects his company's growth.
UPSIDE: What's a typical day like for you? Do you work mostly with a team or mostly alone? Which do you prefer?
Misha Plotkine: Half my day is spent coordinating and managing my team, and the other half is split between creating programs to gather (or parse) information and actually analyzing information. I prefer to split work that is easily segmented into team work, but I prefer to analyze information alone and regroup to discuss findings.
Is MobyTrip.com using analytics to determine the company's best course of action?
Absolutely. We're a software development company, and we primarily make business decisions by running analytics to test our hypotheses. These decisions can be anything -- cutting or adding features based on user behavior and flow, load testing our servers and software to determine fault points, or determining which advertising keywords to bid on by analyzing historical trends.
Do you have to merge data from more than one source?
Frequently we combine information from multiple sources in order to get a uniform picture. For example, in our marketing the actual purchasing of advertising keywords happens inside Google's system, but users drive traffic based on external travel trends.
The resulting behavior can be analyzed inside Google Analytics or on our back end. To get the clearest picture, we merge the data, maintaining separate key identifiers for each source, and create one normalized data set which we can query.
Do you use analytics to find new customers?
We use analytics to combine travel trends, our customer data, and historical data sets provided by Google to create a sophisticated advertising strategy targeting users traveling to specific destinations. This has led to an advanced ability to find new customers very inexpensively.
What's the most common roadblock you hit in your work? How do you deal with it?
Lack of data is the most common problem -- whether we mistakenly didn't log something or simply don't have it. I write programs to either recover this data or find equivalent data that we can use to make defendable assumptions.
Are you working on anything interesting right now? If not, what's your dream project?
We're always working on interesting things in the travel industry! One dream project for me is to build a system to analyze currency fluctuations based on economic seasonality in tourist markets to determine if there is a statistically cheaper time to visit certain locations.
What's your favorite part about being an analyst? Your least favorite part?
I love being able to combine large amounts of data into digestible pieces of information and use them to make informed decisions. My least favorite part is worrying that the numbers have lied. When you're combining huge data sets and running analytics on them, a small, simple logic mistake at the beginning can yield almost right numbers but drastically wrong conclusions.
If you could go back in time, what's the one thing you would tell yourself as a new analyst?
It is important to understand the logic of an analysis all the way through; otherwise you won't be able to tell if the results look logical or not. Just knowing how to program will lead to a data set but not a conclusion. The programming is easier to plan if you first really understand what you're trying to find.
What's your biggest pet peeve (abused buzzword, overhyped idea, etc.) and why?
I think the most abused word is "big data." Data analysis is about how you actually use the data, not how much there is. There is a technical requirement to use Hadoop, and you shouldn't just add it to your stack for the buzz.
We don't analyze big data yet -- right now we probably look at data sets across several millions of rows, which is trivial to do in-memory and easily transferable between production and replica databases. We are using analysis to find answers to specific questions. There is a huge demand and need for actual big data analysis, but I definitely find the term overused, especially in the start-up scene.
Whether it's the latest Python build or a 50-gallon drum of espresso, what's the one thing you can't do your job without?
Plain old SQL ... Python and PL/SQL are all great, but nothing beats the speed and simplicity of running a query to get an answer.
Where is data analytics/data science headed in the next few years?
I think that the development and proliferation of IoT technology will result in an amazing amount of data that companies can use to create or improve products. For example, insurers can analyze common household composition and behavior to build better life, home, or auto insurance policies.
This is already being done in auto insurance actually. By adding a driving tracker certain companies are providing insurers with better analytics for calculating premiums, and better/calmer drivers are rewarded with lower rates. This principle will definitely be extended to other aspects of our lives, for better or worse.
About the Author
James E. Powell is the editorial director of TDWI, including research reports, the Business Intelligence Journal, and Upside newsletter. You can contact him
via email here.