Big Data: Why the architecture of your tools is essential

Every second, your visitors interact with your site and leave tons of informations behind. You can use that information to offer them tailored user experiences, but you need to make sure that the technologies you use are able to keep up with this challenge and process this data volume accurately! Big Data is more than just a buzz word. Our architecture has been thought and constructed in order to be able to evolve along with the latest technologies and to easily process the enormous amount of data available. Today, we thought it was about time we let you take a peek under the wood and see for yourself Kameleoon’s technical foundations.

Smart Data is finally here

Today’s challenge for marketers is no longer to collect data, but to leverage it efficiently. Every month, you have hundreds of thousands, or even millions, of visitors on your website and collect at least as many details on their browsing behavior and customer profil. We want to help you make the most of this data and facilitate decision-making. That’s why Kameleoon concentrates since 2009 on the analysis and activation of behavioral, contextual, demographic data, as well as information coming from your DMP or CRM. That’s the core of our business, be it via A/B testing, personalization or our machine learning algorithms to identify the most promising visitor segments for your brand. Today, we offer the best solution on the market for even the most complex A/B tests and personalization campaigns, and we’re not the only ones to say so: Our customers made us Top Rated A/B testing and personalization tool on Trust Radius and in a benchmark by agency Cartelis, Kameleoon was the most performing among the four best A/B testing solutions.

Elasticsearch vs SQL, PHP: why our technological choices are important for you

Technology evolves really fast. I can already hear teeth gnashing, but some of the technologies that ruled the Internet in the beginning of the century have become near-obsolete today (such as SQL databases) or ill-adapted to real-time processing (such as PHP or other computer languages used on modern websites). The debate is not about questioning the quality of these pillars of the Internet, but about their place in nowadays’ corporate IT ecosystems. Companies face new challenges in data management which call for an upgrade of internal infrastructure. That’s why we took the decision to integrate more stable and adaptable technologies (unlike SQL and PHP technologies) and chose Elasticsearch and NoSQL (#nosql) to create a sound and lasting solution which is able to meet all our customers’ needs and grow with them. 

SQL vs. NoSQL : why is it important ?

SQL data bases are at the center of Internet developments since the 1970s. And major websites like Google, Facebook, Twitter and YouTube are still partially based on them. Unfortunately, these databases are not adapted to constraints brought on by real-time user experience personalization. When the volume of data to be processed becomes too important, this type of database slows down. When used in a tool for UX personalization, this directly influences the user’s experience. If you create advanced personalizations with cross-device user journeys and information uploads form your DMP or CRM, the promise of real-time just can’t be kept. There are two solutions to this problem: 1. Permanent upgrade of the SQL database, with hardware updates, but risking instability and jeopardizing data security. 2. Data distribution among different databases with specific attributes. That’s our choice, with a NoSQL structure. NoSQL (short for Not Only SQL) has been initiated by several of the big players mentioned above when they realized that better data allocation significantly improved the processing of this huge amount of information.

How does Kameleoon manage visitor data in practice?

1. A visitor lands on your site. All his data regarding the visit (behavior, context, etc.) is automatically saved locally, on his browser’s LocalStorage. This way, personalized experiences can be offered right way, without even using a distant server. 2. The data is also saved on one of Kameleoon’s servers. Once the visit is considered as ended (after 30 minutes of inactivity), the data is automatically sent to the storage infrastructure. Next step: data storage. 3. We set up two data storage platforms with NoSQL:

• Elasticsearch: This database has storage and search engine features and has been designed for real-time data processing. Reports can be sent to Kameleoon’s back-office for analyses and, if necessary, data can be retrieved from browsers. This is useful for cross-device visits, when the server needs to upload information from visits on different devices. It also is necessary when connecting to a platform pushing real-time events outside your website (#no, solutions such as Segment, call center phone call, etc.). These events are automatically linked to the current visit on the website and can be used to trigger personalizations.

• Cassandra: All visitor data is stored in this database. This enables us to perform in-depth analyses, without impacting the capacities of the other platform (ElasticSearch).

NoSQL architectures tend to replace their predecessors on the Web for two reasons:

• They are able to evolve and adapt to innovation.
• They manage a much higher data volume.

These technologies are a real advantage for us and for our customers.

The data volume we are processing is enormous. We didn’t want to be caught up in the permanent optimization of an obsolete system, so we chose right away to set up a more modern and sound architecture using NoSQL. This how we are able to process this huge amount of data while keeping the promise of real-time. -Jean Noël Rivasseau, Founder and CTO, Kameleoon.

Ask for your demo!

If for you, seeing is believing, check out our UX optimization solutions. We are happy to give you a detailed view of all the possibilities our tool offers.