Tuesday, May 9, 2017

Future of Digital Analytics. The latest developments in digital analytics: top trends in 2018



Modern digital analytics requires knowledge of programming languages, understanding of deep and machine learning algorithms. There is a need for strategic thinking skills: business needs specialists able to understand the request for an analyst, see the processes systematically and confirm them in numbers.
Here are the top trends in digital analytics in 2018:
The market is moving towards an increase in the competence of both service providers and customers. Big advertisers do not want only the standard reports - Google Analytics, so they develop their own reports that map data from CRM or internal databases to data from web analytics systems.
Another trend is an increase in mobile traffic and the development of mobile analytics systems. In 2016, Google introduced Firebase Analytics, AppsFlyer and other similar systems. Working with the attribution of mobile traffic is somewhat more complicated than in the web: applications are installed usually through stores - Google Play or AppStore - and only they know the source that led the user to the installation. Therefore, data in mobile application analytics systems may be inaccurate. A feature of the mobile market is a wide variety of advertising networks and systems of analytics, and not all of them work correctly with each other.
Data collection is increasingly automated:
The demand of the digital analyst in the future will determine the knowledge of programming languages, machine learning, and the ability to work with big data.
However, it is necessary to check the consistency of the data, the results of the research - with this the machine cannot cope yet. That’s why the digital analyst becomes today the designer, the architect of the analytics system. And technology complements the human mind, creating a holistic picture.
 For example, the case of the US operator Sprint. The company used to terminate the service agreement with those customers whose cooperation with it seems unprofitable to it. One client often contacted the operator's support service with a request to understand the reasons for the connection failure. The quality of the connection was low due to poor coverage. Based on the machine analysis of the frequency of calls to the call center, the cost of processing and revenue, the company decided to break the contract with a non-profitable customer. At the same time, Sprint ignored the reason for the calls and did not solve the communication quality problem. After a while, according to the materials of the forum of its own support service, the company found out that the husband of this client, the owner of a small business, transferred all corporate numbers to another telecoms operator. Trusted by the decision of the machine, Sprint lost several customers at once.
There is a transition from descriptive analytics to predictive and prescriptive.
 In the nearest future the need for forecasts will become more massive. Therefore, analysts need knowledge of machine learning algorithms and programming basics that help analysts use a large array of data to make forecasts and improve their accuracy.
The approach to metrics is changing. There is a noticeable shift from the usual quantitative indicators (visitors, clicks, orders) to deeper, more calculated (user satisfaction, user experience). A few years ago, approximately companies did not analyze the buyer's way from the first contact with the product to the fact of the transaction, did not understand the conversion and therefore did not know what to focus on the promotion. Now it is important to measure not the sales but the behavior of the user.
Digital-marketing in the form of an unsystematic and unreasoned set of steps ceases to exist. Companies, where the analytical work is not adjusted and instead of the KPI system are considered separate indicators, will remain ever less. For analysts who do not act systematically, but by instinct, who understand only certain tools and do not know how to analyze complexly, there simply will not be room in the market. They will be replaced by people who have not only tried tools but understand the logic and interrelations have a systemic strategic approach.

Monday, May 8, 2017

Analytics for governments



 
Mostly business analytics is used to help corporation compete with each others. However, governments do not compete at all unless it is a war. The first examples of using analytics by politician were during national defense.  First computers were made in order to calculate the weaponize missiles trajectory. In the military field analytics are widely used by secret service including automated analysis of text and voice messages.
Nowadays local administrations, state administrations, and federal governments use analytics. It doesn't necessarily higher competitiveness but produces more effective results. Analysis of crime statistic in order to avoid any rule breaking was the most successful at the local level. In New York the program CompStat connects crimes with city’s districts and advises where to place more police departments.
There are many types of using analytics at the state level some of them cause significant saving of money. A few states, including Massachusetts, implement income optimization types which already earned hundreds of millions dollars. These are both taxable and non-taxable payments. Also states’ governments reveal fraud identification in order to lower percentage of illegal welfare collection, food stamps, and free medical insurance.
Departments of natural resources apply analytics for modeling and consuming resources such as gas, oil, minerals, and parks.
At the federal level one of the first examples of non military using analytics was monitoring citizens who try to avoid paying taxes. The campaign of measuring the honesty of the tax payer was launched in 1963 to analysis the kind of people who are most likely to hide their income in order to close the gap between the real and possible tax payment. The campaign was successful but the research proved to be very expensive. That is why the program was closed in 1988. However, IRS continues to practice analysis of law obedience and defining audit cost.
Healthcare is one of the most important fields for implementing analytics because it is the most costly expense after military defense. For example, Department of Veterans Affairs uses medical electronic records and its analysis. The article “The best medical assistance in the USA” from Business week says: “In the mid 90s doctor Kennet Kaizer implemented the biggest system of electronic medical records in the US. Beside this, he decentralized decision making, closed not fully occupied hospitals, and created reliable qualitative measurements. Hospitals of  Department of Veterans Affairs exercise analytical approaches such as forecasting modeling of chronicle diseases, scientifically defined clinical practice, automated decision about medical prescriptions. VA Department experience is one of the most powerful proof that analytics are able to influence government as a private sector.
Government all around the world more often implement forecasting analytics. For example, in Great Britain the analysis of criminal statistic showed that 50% of all felonies were committed by 5% of all offenders. The goal is to understand the crime logic instead of each individual investigation. It will grant a significant advantage to prevent crimes from happening.




Thomas Davenport, Jeanne Harris, "Competing On Analytics". 2010

More analytics thoughts you can find here: https://analyticsinbusinessworld.blogspot.com/2017/05/analyst-and-data-scientist-skill-sets.html

Tuesday, May 2, 2017

The difference between Google Analytics and Kissmetrics



The methods for identifying users in Kissmetrics and Google Analytics work differently. There are some differences below.
When tracking users, any analytics system should be guided by one of two assumptions:
·                     Each visit from one device is made by the same person.
·                     Each visit is considered a visit by a new user until he identifies himself.
Kissmetrics follows the first assumption. If a user visits your site from a computer, tablet or phone, Kissmetrics recognizes it only once - when a person signs up - and binds all the data from this device to his/her account.
Google Analytics, on the other hand, assumes that every visit to the site is made by a new user. The only way to get around this is to identify people during each session to see everything they do.
Kissmetrics is not a copy of Google Analytics. Although both tools are designed to collect analytics, the usefulness of using a particular service depends on company’s situation.
With Google Analytics, a company can understand how visitors interact with its site. Kissmetrics allows to see how customers use company’s product, identify the main channels for attracting users, document data on split tests and much more. That's why, in order to get a complete picture of the business, companies must combine the capabilities of both tools.
 Google Analytics is recommended if:
You want to track the number of visitors and visits. Kissmetrics can also track visitors, but using the service just for the sake of it does not make sense.
You want to track the bounce rate, the time spent on the page, and the outputs. For these metrics, Google Analytics is suitable, which can show where people leave the site - without the link to how this will affect your conversions. In Kissmetrics you can see from which part of the funnel they are leaving.
You want to track in general only the registration process or the order process. Google Analytics will give you the ability to track funnels (compound goals) by sessions, not real users.
Kissmetrics is recommended if:
Collect data on profits that help to make decisions.
Detect channels of attracting customers.
Track and optimize the sales funnel.
Get an understanding of how customers use the product.
Find points of multiple revenue growth.
Thus, Kissmetrics is more focused on studying customer behavior and optimizing the funnel. If you need an in-depth analysis and it's important to find a link between marketing and profit, the combination of Kissmetrics + Google Analytics will be the most appropriate option.

Tools like Kissmetrics provide user analytics — more of a UX function — how engaged are users on average, cohort-based retention analysis, what actions have specific users performed.
Google Analytics excels at web analytics — more of a marketing function — what pages were visited, from where in the world, what browser were they using.
While GA has some cohort-based visualization, they're not very useful to UX because they're analyzing visitors, not users — perhaps they have some value to marketing function.