MegaFon Big Data Challenge: big data against "deer", loneliness and much more
We, like any large telecom company, aggregate a lot of data on subscriber activities - starting with basic ones like a social profile and ending with more complex ones, for example, consumer interests. How can they be used to thrive users and businesses? Independent developers have a lot of ideas on this. To learn more about them, we organized the MegaFon Big Data Challenge.
From more than 300 applications, we selected 15, gathered these teams in the Moscow office and for two days told them about all sorts of useful things - business models, MVP, methods of market assessment, testing for consumers, mastery of presentations. And, of course, about big data. Although here we ourselves had something to listen to. According to the results of the hackathon among the three best, in our opinion, projects, we distributed a prize fund of one million rubles. Let's keep the intrigue up to the footer and tell about them at the end, and start with the other participants.
')
Reduce the number of car accidents by 30% using big data calculated by the Ride & Slide team. The initial data is taken from accident data, profiles in social networks mentioning an accident, GPS tracking information and other data. Based on this, a service is created that alerts users to dangerous drivers and areas nearby, allows you to add and verify them yourself.
This development may be interesting in B2B - for logistics, transport companies, taxis. Separately, insurers were mentioned in the presentation, which the “deer” detector would bring direct savings.
Insurance and pharmaceutical companies need to know more about the health of both current and potential customers. To help them in this, Pharma Disrupted plans to use geolocation data. How often does a person visit the hospital? How long is sitting on the hospital? This information will be used with the consent of users and linked to their profiles and activity in social networks. Thus, insurers will be able to improve scoring, and pharmacists - the range in pharmacies.
This project is aimed at improving advertising targeting. To do this, it is planned to use a large amount of data describing the user: real-time geo-targeting, a profile in MegaFon and Internet services (Mail.ru, Rambler & Co), cookies, as well as data from electronic cash offices and card transactions. The developers' goal is to occupy a 10% share of advertising programmatic intermediaries.
Big data comes to the rescue of socialization. The service will offer users the most suitable places, events and communities, based on information from profiles in MegaFon, in social networks, geolocation data and much more. Depending on the interests of users will be divided into groups. In these groups, for example, people with a mutual desire will find out that they can go to some event not alone - hello, Tinder.
This project has a lot in common with Smart Target. The main goal is also the personalization of promotional offers to increase their effectiveness. But here in the user processing system, one interesting feature is used on the verge of psychology - personalization based on personality types and human values:
The work of this system through VK profiles can be assessed using a chat bot .
When opening new points of offline retailers can also help big data. Using them, the Vanga Predictions team answers the most important questions that arise during startup. Based on the revenue forecast, it becomes clear where to sell. Based on the forecast by category - what to sell. And finally, by analyzing the movement of users - to whom to sell.
The work of the service is divided into four main stages. First, preferred stores are determined for each subscriber. Based on this, forecasts are made, users are clustered. At the final stage, revenues are broken down depending on the categories of goods. The project already has a development plan, where two versions with different functionalities are indicated.
The turn of Eycharov came. The Work2vec team will help network companies with a turnover of> 15% using vector analysis of online profiles to find more relevant employees who can gain a foothold in the state and reduce the cost of recruiting. This is true for large retail networks, the HoReCa segment and the service sector.
The project is similar to the previous one, but designed for the selection of middle and senior staff. Potential candidates are entered into the database, taking into account the history of work, offline activities, interests, psycho, attending professional events and even activity on Habré. When a company receives a request, it immediately receives options relevant to the job. In the near future, the system is planned to be deployed in MegaFon.
Here, using large data is planned to help outdoor advertising. With the help of data from MegaFon related to the activity of users in social networks, the coverage accuracy of billboards, screens and street promoters will be increased. All this will happen in real time. There is, for example, a group of female students from the university passing by some cafe, and a growth doll comes out of this cafe and distributes flyers for free coffee.
Medtracking based on the data of the transaction and user transactions is intended to make the range in pharmacies more relevant to the needs of customers. On the basis of Medtracking data, it will be possible to conduct marketing campaigns and understand why and why everyone goes not to one pharmacy, but to another. In the long run, developers are counting on their decision to raise pharmacy profits by 25%.
The efficiency of call centers today is on average 30-50%. At best, there is some kind of basic segmentation of the call to the obvious categories. Using the Smart Reach Concept product and MegaFon data on telephone activity, call centers could determine the best time for a successful call to a particular person. For this purpose, in addition to information about calls, you can take into account geolocation and subscriber traffic. All interaction is planned to organize in the form of a convenient API.
The most popular area for bigdata startups has become advertising. Another representative of such services was the recommendation system from the GrowUP team. It collects data from MegaFon and partners, information about advertisers' products and as a result generates personal recommendations for users in real time.
We turn to the winners of the hackathon. A virtual mentor is a service developed by the K13 team to select ad campaigns based on big data. The user specifies the scope of the company, the preferred audience, then selects an advertising campaign on the partner site and geographically “tunes” it. In parallel, studying the trends and demand in its segment. Here is the prototype:
Provide a service planned for the subscription system. There was a specific figure - 15 000 rubles per month for full access. The developers have already concluded the first deals.
And again there is big data in telemarketing, but this time with a narrower scope, for construction developers and car dealers. In this area, for one target call, you can give up to 15,000 rubles, and as a result, the poor student hired for this purpose will call 500.
This is how Data4 works. The developer receives a call. Immediately a request is made to MegaFon for subscriber activity — geolocation, transaction, communication, etc. Data4 analyzes information and classifies the client. If he falls into the fraud group, then the call will be processed only after the fact or ignored altogether.
The winner, the Ingenix team, touched upon their project, perhaps, the most pressing problem in comparison with other participants. Based on an open protocol available for external audit, they plan to create a platform for regulating and monetizing the personal data market.
Through this platform, users will be able to regulate what personal data they provide. Of course, not free. Remuneration will be issued for consent to receive SMS offers or, for example, access to Facebook profile data. For a start, the platform itself will be the mediator here. Then, as users become aware of the value of their data, they will be able to work with advertisers directly and control who gets the information.
In turn, for business users the rating of the most interesting audience for their sphere will be formed. They will be able to acquire detailed information on such users - the information that they themselves provide.
Digital Fingerprint Workflow
Many projects seemed interesting to the jury. Why did the first prize get Digital Fingerprint? Perhaps the point is what place it assigns to users in the big data market, which are directly connected with these users. They become full participants in this market, due to the openness of the system, they understand what and how it works, and can be sure that confidentiality is maintained at the perimeter they set. Business clients also do not remain offended. They have at their disposal data from users with high conversion, who themselves agreed to provide their data. It looks like a win-win.
You can read about how we created a new line of tariffs with the help of big data in one of the previous posts .
From more than 300 applications, we selected 15, gathered these teams in the Moscow office and for two days told them about all sorts of useful things - business models, MVP, methods of market assessment, testing for consumers, mastery of presentations. And, of course, about big data. Although here we ourselves had something to listen to. According to the results of the hackathon among the three best, in our opinion, projects, we distributed a prize fund of one million rubles. Let's keep the intrigue up to the footer and tell about them at the end, and start with the other participants.
')
"Deer" on the road
Reduce the number of car accidents by 30% using big data calculated by the Ride & Slide team. The initial data is taken from accident data, profiles in social networks mentioning an accident, GPS tracking information and other data. Based on this, a service is created that alerts users to dangerous drivers and areas nearby, allows you to add and verify them yourself.
This development may be interesting in B2B - for logistics, transport companies, taxis. Separately, insurers were mentioned in the presentation, which the “deer” detector would bring direct savings.
Pharma Disrupted
Insurance and pharmaceutical companies need to know more about the health of both current and potential customers. To help them in this, Pharma Disrupted plans to use geolocation data. How often does a person visit the hospital? How long is sitting on the hospital? This information will be used with the consent of users and linked to their profiles and activity in social networks. Thus, insurers will be able to improve scoring, and pharmacists - the range in pharmacies.
Smart target
This project is aimed at improving advertising targeting. To do this, it is planned to use a large amount of data describing the user: real-time geo-targeting, a profile in MegaFon and Internet services (Mail.ru, Rambler & Co), cookies, as well as data from electronic cash offices and card transactions. The developers' goal is to occupy a 10% share of advertising programmatic intermediaries.
Megaphone. Entertainment
Big data comes to the rescue of socialization. The service will offer users the most suitable places, events and communities, based on information from profiles in MegaFon, in social networks, geolocation data and much more. Depending on the interests of users will be divided into groups. In these groups, for example, people with a mutual desire will find out that they can go to some event not alone - hello, Tinder.
uLime
This project has a lot in common with Smart Target. The main goal is also the personalization of promotional offers to increase their effectiveness. But here in the user processing system, one interesting feature is used on the verge of psychology - personalization based on personality types and human values:
The work of this system through VK profiles can be assessed using a chat bot .
Vanga Predictions
When opening new points of offline retailers can also help big data. Using them, the Vanga Predictions team answers the most important questions that arise during startup. Based on the revenue forecast, it becomes clear where to sell. Based on the forecast by category - what to sell. And finally, by analyzing the movement of users - to whom to sell.
The work of the service is divided into four main stages. First, preferred stores are determined for each subscriber. Based on this, forecasts are made, users are clustered. At the final stage, revenues are broken down depending on the categories of goods. The project already has a development plan, where two versions with different functionalities are indicated.
Employee with warranty
The turn of Eycharov came. The Work2vec team will help network companies with a turnover of> 15% using vector analysis of online profiles to find more relevant employees who can gain a foothold in the state and reduce the cost of recruiting. This is true for large retail networks, the HoReCa segment and the service sector.
MegaHR
The project is similar to the previous one, but designed for the selection of middle and senior staff. Potential candidates are entered into the database, taking into account the history of work, offline activities, interests, psycho, attending professional events and even activity on Habré. When a company receives a request, it immediately receives options relevant to the job. In the near future, the system is planned to be deployed in MegaFon.
Targetoid
Here, using large data is planned to help outdoor advertising. With the help of data from MegaFon related to the activity of users in social networks, the coverage accuracy of billboards, screens and street promoters will be increased. All this will happen in real time. There is, for example, a group of female students from the university passing by some cafe, and a growth doll comes out of this cafe and distributes flyers for free coffee.
Medtracking
Medtracking based on the data of the transaction and user transactions is intended to make the range in pharmacies more relevant to the needs of customers. On the basis of Medtracking data, it will be possible to conduct marketing campaigns and understand why and why everyone goes not to one pharmacy, but to another. In the long run, developers are counting on their decision to raise pharmacy profits by 25%.
Smart Reach Concept
The efficiency of call centers today is on average 30-50%. At best, there is some kind of basic segmentation of the call to the obvious categories. Using the Smart Reach Concept product and MegaFon data on telephone activity, call centers could determine the best time for a successful call to a particular person. For this purpose, in addition to information about calls, you can take into account geolocation and subscriber traffic. All interaction is planned to organize in the form of a convenient API.
Growup
The most popular area for bigdata startups has become advertising. Another representative of such services was the recommendation system from the GrowUP team. It collects data from MegaFon and partners, information about advertisers' products and as a result generates personal recommendations for users in real time.
Virtual Mentor - 3 place
We turn to the winners of the hackathon. A virtual mentor is a service developed by the K13 team to select ad campaigns based on big data. The user specifies the scope of the company, the preferred audience, then selects an advertising campaign on the partner site and geographically “tunes” it. In parallel, studying the trends and demand in its segment. Here is the prototype:
Provide a service planned for the subscription system. There was a specific figure - 15 000 rubles per month for full access. The developers have already concluded the first deals.
Data4 - 2 place
And again there is big data in telemarketing, but this time with a narrower scope, for construction developers and car dealers. In this area, for one target call, you can give up to 15,000 rubles, and as a result, the poor student hired for this purpose will call 500.
This is how Data4 works. The developer receives a call. Immediately a request is made to MegaFon for subscriber activity — geolocation, transaction, communication, etc. Data4 analyzes information and classifies the client. If he falls into the fraud group, then the call will be processed only after the fact or ignored altogether.
Digital Fingerprint - 1 place
The winner, the Ingenix team, touched upon their project, perhaps, the most pressing problem in comparison with other participants. Based on an open protocol available for external audit, they plan to create a platform for regulating and monetizing the personal data market.
Through this platform, users will be able to regulate what personal data they provide. Of course, not free. Remuneration will be issued for consent to receive SMS offers or, for example, access to Facebook profile data. For a start, the platform itself will be the mediator here. Then, as users become aware of the value of their data, they will be able to work with advertisers directly and control who gets the information.
In turn, for business users the rating of the most interesting audience for their sphere will be formed. They will be able to acquire detailed information on such users - the information that they themselves provide.
Digital Fingerprint Workflow
Many projects seemed interesting to the jury. Why did the first prize get Digital Fingerprint? Perhaps the point is what place it assigns to users in the big data market, which are directly connected with these users. They become full participants in this market, due to the openness of the system, they understand what and how it works, and can be sure that confidentiality is maintained at the perimeter they set. Business clients also do not remain offended. They have at their disposal data from users with high conversion, who themselves agreed to provide their data. It looks like a win-win.
You can read about how we created a new line of tariffs with the help of big data in one of the previous posts .
Source: https://habr.com/ru/post/345892/
All Articles