Artificial Intelligence, Machine Learning and Big Data – how does this go togehter with marketing and sales?

E-Commerce and marketing generate enormous amounts of data, which can tell you a lot about user behaviour and preferences. Making this data work is the key objective of Adtelligence. For achieving this goal we use various Machine Learning algorithms, from Neural Networks to Markov Models. This high-level technology, e.g. AI, Big Data and Machine Learning, is normally used by IT departments, BI and Data Science.

AI and Big Data – Easy to use for sales

Our focus lies on making complex technologies usable for everyone. No programming, only configuration. We support your sales by increasing your conversions with AI- based data analysis and a personalized customer approach.
Our software analyzes customer and user data, segments target groups, automates and optimizes your channels based on all click and transaction data – all with help of advanced Machine Learning. Our algorithms are flexible and can be used for different purposes. We will help you to use them for your Business Case.


These Are Our Main Ingredients

Experience is the best teacher. And it is in your data. Descriptively, Machine Learning algorithms learn from past data to estimate future events. You want to know, what the website needs to look like for users to most probably convert? Or when you should send which email to whom, in order to make the biggest impact?Your experience and our expertise in digital sales will increase revenue – by using Machine Learning algorithm and different Use Cases along the entire customer lifecycle.

DSGVO is our standard

All data is being pseudonymized or at parts completely anonymized before they get used for optimization. You have full ownership of all data. In general, personalized data, such as a customers name, are not neccessary for optimizing behavior.

Transparent AI – Not a Black Box Anymore

Your website’s performance is usually influenced by a multitude of factors and initiatives. With Adtelligence, you receive a specificly designed comprehensive Performance Reporting of your actions. At a glance you can see, how different user segments react on different website versions and how effective algorithmic optimization worked. On demand we can connect the landing page results with defined context parameters to show even more granular effects.

We can use your exisiting data models

Over the years you might have had a large number of different AI and data models trained and refined. These do not get lost they should be integrated and used to achieve even better results.

What makes us special


We pursue a holistic approach. We use several complementary Machine Learning algorithms for different applications at differents points in order to be able to offer our customer comprehensive solutions.


Our software has a high degree of prefabrication. Our partner do not need to technically set up campaigns, instead they can simply configure the engine according to their requirements and start immediately. The usage is simple and intuitive.


Many of the steps that customers of other providers have to manually take are automated with us. This includes algorithms auditing data quality, segmentation of target groups, categorization of transaction clusters or valuation of customer activity.

Along the customer lifecycle there are many use cases which can achiev more revenue with smart data management and AI: From the automation of your sales processes, increase of conversion rates, personalized customer contact in the portal or app, until marketing automation and omnichannel campaigns.
Moreover the AI learns from the results and automatically optimizes target groups, placement, content or campaigns. Important: each area needs a different approach. In order to successfully improve your business, AI it needs to be used wisely.

Using Data to Optimize Content or Products

Not every website design works equally successful with every customer. We optimize the approach of new customers by showing each customer an individually tailored variant. For this kind of optimization, Adtelligence offers several options. Thus, a Neural Network can be used for optimization just as well as a Bayesian Bandit based on parameters. The algorithms autonomously and automaticaly recognize, for example, that on Mondays smartphone users from northern Germany prefer a different variant than desktop users on Fridays.

Furthermore, the algorithms iteratively continue learning and identifying changes in user preferences. They accomplish a significantly better targeted and continously optimized personalization compared to classic approaches such as A/B Testing, which assumes that preferences are static. This optimization can also be applied to the distribution of marketing and sales material to existing clients, in order to approach each customer individually. Application to new use cases is simple, quick and flexible.


Contextual Bayesian Bandit

Recognizing Trends Without Waiting For Statistic Significance of an A/B Test

In common A/B Test scenarios used in similar tools on the market, different variants get tested against each other and, in case of statistically significant differences, a preferred variant gets chosen. Whereas our Machine Learning Algorithm, an advancement of the Contextual Bayesian Bandit, starts optimizing on day one and continues long-term learning in an iterative process. Instead of excluding variants, traffic is continously being distributed and assigned according to the calculated best conversion probability.


Leveraging Raw Data for Marketing and Sales

For the first time it is possible for marketing and sales to have comprehensible and simple access to your data. Complex prefabricated models from Adtelligence allow for your raw data to be directly transformed in usable and understandable data.

That way transaction data can in real time be classified into, for example, online and offline transactions in order to determine online affinity, or single transactions can be divided into 50 understandable categories instead of more than 1,000.


Cloud, Data Lake and Big Data Analytics Infrstructure

You want to host your application in a private cloud environment with ISO certification?
Adtelligence is happy to advise you in your decision. Whether it be Amazon AWS, Google Cloud, Azure or rather a Private Cloud – a hotly debated topic. Adtelligence can call upon in-house competences. We find you the appropriate contact person amongst various providers and support you on your journey to the cloud.

There is no way around a modern IT infrastructure. But do you really need a Data Lake, special applicatons or an AI framework? And how can transformation proceed? Together with our partners we happily assist you in finding the right answers to these questions.

Female and Male IT Engineers Discussing Technical Details in a Working Data Center/ Server Room with Internet Connection Visualization.

Automatic Data Quality Analysis

The groundwork for learning from data and successfully implementing Machine Learning is a clean data base. The Adtelligence software automatically examines the quality of your data, openly displays them and offers recommended actions to improve data quality.

During each data integration, data is automatically cleaned and utilized for your specific Machine Learning use cases.


HCB Scan And Fuzzy Clustering

Automatic Formation of Target Groups

For the messages to meet the right target group, we have developed a solution with which we can apply different automated clustering methods to your customer data. We, for example, use frequency based techniques such as HDBSCAN or Fuzzy Clustering in order to generate reasonable and usable customer segments from your customer data.

With K-Means, K-Medoids or EM-Clustering methods, you can seperate your customer base into arbitrary subgroups to, for example, carve out the primary distinguishing features of your customer base.

Adtelligence has the experience to determine which method works best for each of your use cases.

Mit K-Means, K-Medoids oder EM-Clustering Verfahren können Sie Ihren Kundenstamm in beliebige Teilgruppen unterteilen, um bspw. herauszufinden, welche die primären Unterscheidungsmerkmale Ihres Kundenstammes sind.

Für jeden Ihrer Use Cases hat Adtelligence die Erfahrung welche Methode die beste ist.

Learning With Machine Learning: Which Content Works Best With Your Target Group

Email Automation is not new. Our software enables you to start rule-based automated email campaigns including an automated optimization with the Bayesian Bandit. The Bayesian Bandit independently learns, which email variant and content achieves the highest conversion rate with which customer.

Hidden Markov Model

Waking Up Sleeping Dogs

For finding the best performing message to customers we need to understand their behaviour. Is someone inactive? Do they only purchase certain product groups? Or do they change their behaviour over time? We analyze your customer’s activities, classify them and identify behavioural patterns, for example with the Hidden Markov Model. We can react, whenever a customer becomes inactive and we learn which trigger is most successful to activate him again.



From Robots and Mutants to Neural Networks


Automatically Counterbalancing Seasonal Effects

We have improved the well-known Bayesian Bandit approach to enable recognition and consideration of short and long term trends and sustainable opimization.

In common A/B Test scenarios used in similar tools on the market, different variants get tested against each other and, in case of statistically significant differences, a prefered variant gets chosen. Whereas our Machine Learning Algorithm, an advancement of the Contextual Bayesian Bandit, starts optimizing on day one and continues long-term learning in an iterative process. Instead of excluding variants, traffic is continously being distributed and assigned according to the calculated best conversion probability.

Furthermore, the Machine Learning algorithm recognizes seasonal effects and trends. It can be configurated to “forget” learned knowledge for a defined timer period for reacting to current situations or events.

happy woman gambling at casino playing slot machine having lots of fun



The ideal application of Neural Networks for optimization is always based on the individual configuration of a network. All parameters need to be adapted to the respective business case (number of visits, number of available variants, type of optimization goal,…). The so calles Hyper Paramater of a Neural Network determines its capability to deliver best results. The configuration of Hyper Parameters is a task which is performed by experienced Data Scientists.

Adtelligence offers you Neural Networks with an autopilot. With just on click, Neural Networks can automatically and continously be adjusted to your individual business case without requiring expert knowledge. Experienced Data Scientists can, of course, make individual adjustments when needed. This makes AI usable for everyone!

Adtelligence bietet Ihnen Neuronale Netze mit Autopilot. Mit nur einem Klick passen sich die Hyper Parameter eines Neuronalen Netzes automatisch und fortlaufend an den individuellen Business Case an, ohne dass Expertenwissen vorausgesetzt wird. Erfahrene Data Scientists können natürlich jederzeit ihre individuellen Anpassungen vornehmen. Dies macht KI für Jedermann erlebbar und nutzbar!

The Formula 1 of Converion Optimization – Teasing Out the Last 5% of Conversions With Neural Networks

Long time ago since you passed through A/B Tests? Automized Traffic allocation sounds ordinary for you? How about a Neural Network? If implemented correctly it enables you to tease out the last 5% of your conversion rate. This is the Formular 1 car for your increase of conversions: difficult to drive but extremely performative.

Digital x-ray human brain on blue background 3D rendering

You Have an Especially Complex Use Case?

Upon request we apply all steps of Big Data Analytics processes to your specific use cases.

For example a typical call center case: Who do I call first?
We startet a smart merger of data and different classification processes, which will enable you to identify purchase probabilities of your customers and prioritize the calls according to sales potentials.

Recognize Trends Before They Are Becoming Trends
By merging product configuration data with sales data and accordingly applying different predictive analytics methods, we can find out for you, which product development will follow a trend.

Smiling handsome customer support operator with headset working in call center.


Start risk free with a proof of value project and we specificly show you the potential that lies in your data.