Liip Blog Kirby Mon, 15 Jul 2019 00:00:00 +0200 Latest articles from the Liip Blog en SDS 2019: Business + Machine Learning = ❤️ Mon, 15 Jul 2019 00:00:00 +0200 The big over-arching theme of the conference was human-machine collaboration: How can the results of an AI system be best communicated to its users. This touches topics like the perception, trust, and ethics of such AI systems. I was happy to see these questions at the core of the conference since they have been at our heart too for almost two years: See

In comparison to the last years, I had the impression that more and more corporations are also hooking up with the formerly mostly academic data-science community. That is an impression I had based on the number of booths and talks at the conference.


Ken Hughes gave an amazingly well-rehearsed keynote, that made me think about how the development of technology transcends our every-day-businesses or how he put it: Where silicon meets soul.

One of his key insights was that it is not enough to just satisfy the customer needs these days, but if companies can manage to give their users a tribal sense of belonging and provide more than their customers expect they can truly empower the customer. It might sound cliché but generally shifting towards a high consumer-centricity seems to be a thing not only for Jeff Bezos anymore.


The SDS conference offered up to seven simultaneous breakout session tracks - see the program here - which made it almost impossible to attend all of the talks that I was interested in.

There were technical tracks on Deep Learning and NLP containing lessons from hands-on experience with the newest AI models but there were also business-oriented tracks offering insights on integrating machine learning models into production. I was happy to see that there was a Data Ethics track, which created an opportunity for interested data scientists to discuss and shape the impact of AI models on society. Bravo to this! To get an feeling of what the trends are I attended presentations in different tracks. Here are my musings on the ones I attended. Feel free to check out the now available slides and recordings online here.

Principles and Best Practices for Applied Machine Learning Models

At Swiss Re, showed how the collective expertise and experience of numerous expert practitioners and managers from data science and risk management can be harnessed to to create a definitive set of principles and best practices that guides all our data science activities. In this talk they presented and discussed these principles and emphasized the principles which need much more care by the Data Scientist in industrial applications than in education and research.

I liked the birds-eye view segmenting those principles into “data-related”, “model-related”, “user-related” and “governance-related” areas, which forces us to think about all of these aspects at the same time.

Revenue Forecasting and Store Location Planning at Migros

The data scientists at Migros have presented their own algorithm that provides realistic revenue forecasts even for complex scenarios such as multiple new stores and/or alterations of the competitor store network. This algorithm combined a heuristic simulation of consumer behavior with machine learning methods to deliver both accurate and interpretable results.

It was interesting to learn how their in-house solution was developed in close collaboration with key users at Migros’ ten regional cooperatives. It bacame clear that interpretability for the planning expert was one of the main features that drove the adoption of the tool. Conrats to Bojan Škerlak from Migros, who won the Best Presentation award!

I was also excited to see that in order to create their tool they made use of a lot of open data from the BFS and SBB, which was then combined with their transactional cumulus data. In order to arrive at the end result, they ended up combining their “gravity model” with business logic to make the results more interpretable to the end-users.

Do You Have to Read All Incoming Documents?

In the NLP track Mark Cieliebak showed how NLP solutions can be used to provide automatic classification of incoming documents into pre-defined classes (such as medical reports, prescriptions etc.) and then showed how to extract the relevant information for further processing. Using real-world examples, they have provided an overview of the potential applications, a realistic assessment of the effort and the resulting quality to be expected.

I particularly liked his assessment of the effort needed for data preparation and implementation in regards to the different project cases that they have encountered. Unsurprisingly when a business owner already has a huge corpus of annotated material that greatly reduces the data preparation part, and so allows the team to focus on an excellent implementation of the project.

Also combining supervised with unsupervised learning methods during training and data processing seemed to be an fruitful approach for classes where not enough data is available.

GPU Acceleration with RAPIDS for Traditional Big Data Analytics or Traditional Machine Learning

In the technical track René Müller gave a very interesting talk about how the RAPIDS suite provides the freedom to execute end-to-end data science and analytics pipelines entirely on GPUs. I liked how a simple python library can be used to accelerate common data preparation tasks for analytics and data science without running into typical serialization costs.

I was surprised how many classical algorithms already are implemented in their library (see screenshot below), yet many of those can only run on a single GPU (of course there is always (text: DASK link: . It is worth to note that when using a model that has been trained with RAPIDS one needs to also have a GPU to run it in inference mode, which makes them less portable.

If you are inclined to give it a try, you can either install it on your laptop (if it has a GPU) or simply try the fastest thing that works out of the box: This is a jupyter notebook in the google colaboratory wich even gives you T4 instances for free.

Creating Value for Clients through Data & Analytics

In this talk, Michel Neuhaus (Head of advanced analytics) and Daniel Perruchoud (Professor at FHNW) showed their journey towards a corporate environment focused on creating value through analytics for UBS and clients. They shared interesting insights from their track towards offering a collaborative work place for data science professionals.

As an illustrative example, they have shown a simple use case of segmenting clients into groups of individuals with similar needs allowing us to offer the right service to the right client. I liked how they emphasized that the lessons learned along the way were embraced to drive the design and operation of the solution. For them, this meant going from a use-case oriented view to a vertical use case built on top of horizontal capabilities (see screenshot).

In the sample case that they provided, I liked how they were honest about how they modeled temporal client behavior, by just computing some very simple statistics, such as the minimum, mean, median or the average trend of a customers bank balance in time.

Final Thoughts

Overall the SDS 2019 was an excellent conference showing how the business applications and machine learning technologies are growing more closely together. Having seen a number of talks I am convinced that the winning formula of tech + business = ❤️ needs two additional components though: data and user experience.

Only if you really have the right data - and also have the right to use it in the way you intend to - you have a solid foundation for everything that builds on it. Collecting the right customer data and doing so in a way that preserves the user's privacy remains one of the biggest challenges today.

Focusing strongly on user experience is an aspect that gets neglected the most. Users are not interested in technology but instead in improving their user experience, thus any new type of algorithm has only one goal: to improve the user’s experience. Only then is the final result perceived as useful can stand a chance against the existing status quo.

So coming back to the key-note this means that indeed successful projects are those that use the existing ML-technology to delight, engage and empower the users.

Swiss Venture Club – more than just a website Fri, 12 Jul 2019 00:00:00 +0200 Switzerland’s top business award

The Swiss Venture Club (SVC) serves SMEs and offers its 3000 members from all sectors and regions one of the largest and most important business networks in Switzerland.

At the 7 Prix SVC awards in all economic regions of Switzerland, a top-class jury selects the most successful companies in the region. The great and good of Switzerland’s business world all come to this awards ceremony – which has to take place in the large Hallenstadion stadium to fit the more than 3000 attendees. It is reported in the national press, and the companies involved enjoy nationwide publicity. The Swiss Venture Club is therefore a central driving force for the Swiss SME scene, setting an example of good practice. So, it was about time it set such an example for digital transformation too. In addition to the design of its digital presence, Swiss Venture Club’s entire event process has also been transformed.

From service design to implementation

A process that began with service design workshops ended with a comprehensive web platform. The service design process precisely analysed the needs of the Swiss Venture Club, its members and its sponsors in order to establish the basis for a successful web platform – because anyone paying a five-figure membership fee or sponsorship contribution is entitled to have their say on such a project. High quality services, user-friendly applications and informative content for about 3000 members as well as countless regional and national sponsors were in the foreground. Within the scope of the service design process, the following topics were dealt with:

  • Business requirements
  • Business and user objectives
  • Competitor analysis
  • Customer journeys for members and sponsors
  • Interviews with existing members and sponsors

The results of the service design process formed the basis for the next step, namely developing the UX concept and design.
The platform development process then built on these foundations. The CMS Drupal was used to implement the following core functionalities:

  • Website
  • Members’ and sponsors’ portal
  • Event management incl. ticketing

From the very first workshop, it became clear that we had begun an exciting and comprehensive collaboration. An innovative user experience and CRM integration were important in order to attract and inspire interested visitors, members, sponsors and also, of course, the national press.

Heralding digital transformation with good content

The project also included developing a content strategy. We defined messages, content formats and distribution channels. The new website comes with a few changes. We therefore started communicating with our stakeholders about the changes at a very early stage, and members and sponsors were regularly asked for feedback via our newsletter during the digital transformation process. Incorporating them into the change process in this way ensured a high degree of acceptance when the time came to launch the platform.

Compasses over maps

SEO and analytics help to measure, and thus optimise, the website’s performance, so we put them to good use for the SVC website: an audit and a needs analysis were used to define performance indicators and tracking requirements, which we then incorporated into the website. Target-oriented dashboards now provide regular information about the usage and performance of the new page. Key objectives from an SEO perspective were to comply with technical SEO best practices to attract visitors to the website, migrate content without any loss, and support the production of content geared towards users’ search needs.

User numbers and positive feedback show that the close interaction between concept, design, content, analytics and, of course, technical implementation has paid off. The Swiss Venture Club is ready for the future.

"The collaboration with Liip was intensive, geared to the needs of our target groups and resulted in a modern web platform. Our web platform was very well received by members, sponsors and the media. The Liip team skillfully moderated the agile development and supported us with high commitment and pragmatic solutions."
Alexander Saner, Head Services - Finance IT

The future is yours - congratulations to our apprentices! Mon, 08 Jul 2019 00:00:00 +0200 They learnt how to code and developed websites and mobile applications using technologies such as PHP, Magento, Angular and OctoberCMS. Our business administration apprentices gained knowledge in various fields like Finance, Human Resources and Administration. All of them developed their communication skills and their self-organising attitude.

We are very proud of them. They are the experts of tomorrow. By taking on apprentices, we are investing in the future of not only young people but also the community and therefore the society.

Get to know them better. They tell you all about their time as a Liipers.

Gian Zgraggen

Developer - Zurich - Ping me
Motivated, helpful, sarcastic, coffee lover, metal head, nerd.

Tell me about a project or a task that you handled during your apprenticeship and that you liked.

My first client project was, created with OctoberCMS. I really liked working on it since it was the first "real" project I was able to work on. Before working at Liip I was not able to work on any client project.

What do you like most about your job?

I really like working as a developer. I enjoy working as a dev at Liip, because I am able to work together with clients and directly take influence and help shaping a project. I enjoy the structure of the company and the possibilities it offers. The other employees are young, friendly, and easy to talk to, that way the whole atmosphere is way lighter than in other businesses.

What's your best memory with Liip?

Working on my first project for a client as well as both LiipConfs (our annual internal conference) I was able to attend.

What else do you want to learn?

I want to learn different technologies to create websites, increase my skills in mobile development as well as JavaScript. Additionally I would like to learn more about User Experience (UX).

Sonja Nydegger

Business administrator (Finance/Admin) - Fribourg - Ping me
Happy, motivated, ambitious, helpful, shy, unathletic, sensitive.

Tell me about a project or a task that you handled during your apprenticeship and that you liked.

I liked it very much to get an insight into the hiring process. I was able to document the applications after sending a confirmation of receipt. Handling the various applications not only gave me inspiration for my own application, but also gave me a lot of fun, as we also receive very creative applications. I was able to give my own opinion about each candidate.

What do you like most about your job?

I like the variety of my job. I get insights into the company and can carry very different tasks, which are associated with responsibilities (for example: cash management, applications recording and much more). I am convinced that this is a good apprenticeship and that it will bring a lot of opportunities.

What's your best memory with Liip?

Where to start? I have so many beautiful memories of my time at Liip. I love that all the Liipers are close, which results in a friendly and very nice working atmosphere. I will remember the opportunity I had to write my own blog post ;-) and the chance I had to take over Instagram for a week with other apprentices.

What else do you want to learn?

I would like to deepen my knowledge in the area of finance. I wish to pursue further education as a specialist in finance and accounting.

Sina Marty

Business administrator (Finance/HR/Admin) - Zurich - Ping me
Ambitious, motivated, helpful, funny, cheerful, athletic (gymnastics), impatient.

Tell me about a project or a task that you handled during your apprenticeship and that you liked.

During my three year apprenticeship, I liked the financial area the most. Most of all I liked to create the invoices. It was also exciting to experience the progress of the invoicing process. :-)

What do you like most about your job?

That despite an administrative job there is a lot of variety and contact with internal and external customers and partners.

What's your best memory with Liip?

I have many beautiful memories, but one of my most favorite is my first day at Liip. I was warmly welcomed and was able to work independently very quickly, which I really liked.

What else do you want to learn?

I would like to continue working and starting the vocational baccalaureate (Berufsmaturität) in order to study later.

Tim Keller

Backend & Frontend Developer - Zurich - Ping me
Eager to learn and discover, sarcastic, keen to debate, social, eloquent, helpful, tech-enthusiast.

Tell me about a project or a task that you handled during your apprenticeship and that you liked.

At the end of the IT apprenticeship, I was able to work on Kochoptik and I was part of a development team which integrated me fully.

What do you like most about your job?

The web development is incredibly varied and, depending on the complexity of the project, infinitely expandable and improvable. There are new challenges and problems every day, which have to be solved.

What's your best memory with Liip?

During every project I got to know and appreciate my colleagues better. I was able to work on exciting projects and get to know a lot of great people at the same time.

What else do you want to learn?

I want to expand my knowledge in backend and frontend development and explore the whole Linux world even better. Later I might want to continue my studies at vocational baccalaureate school (Berufsmaturitätsschule).

Rami Jumaah

Backend Developer - Zurich - Ping me
Motivated, helpful, funny, happy, fitness lover, impatient, tea lover.

Tell me about a project or a task that you handled during your apprenticeship and that you liked.

A website that I made for my IPA (Graduation Project) for school, because I built everything from scratch - UX till delivering. And every user who is gonna make use of the website have already liked the website.

What do you like most about your job?

The cooperation between the team members.

What's your best memory with Liip?

All the aperos that we have regularly.

What else do you want to learn?

Open for every piece of information that I can get from any person at the company.

Niclas Deplazes

Backend & Frontend Developer - St-Gallen
Patient, focused, eager to learn, hockey fan, hip hop enthusiast, kindle-worm, tea lover.

Tell me about a project or a task that you handled during your apprenticeship and that you liked.

I was able to work on a real customer project, right from the start of my internship. It was the Memberplus Portal of Raiffeisen Bank. I worked on the backend (Magento & PHP) as well as on the frontend (Angular).

What do you like most about your job?

Generally I like my daily work as an application developer because it's very varied and quite a challenge too. Furthermore, I learn something new every day, especially as a beginner. So I never get bored.

What's your best memory with Liip?

The highlight of my internship was the successful go-live of Memberplus, as I spent most of my internship (including my Graduation Project) on the project.

What else do you want to learn?

During my one-year training at Liip, I was able to deepen and expand my basic knowledge from three school years at IMS Frauenfeld. After the military service, I would like to continue working at Liip for a year in order to gain even more practical experience. Afterwards I will probably go to the ZHAW (Zurich University of Applied Sciences) to start a bachelor's degree in computer science.

The magic of t-SNE for visualizing your data features Fri, 05 Jul 2019 00:00:00 +0200 Typical Problems

In data science we are often dealing with classification problems: We want to predict a finite number of outcomes from our data.

Some typical problems are:

  • Distinguish fraud from valid transactions.
  • Identify customers that need attention asp.
  • Use skin images to detect skin cancer.
  • Pick out fake news.
  • Identify handwritten symbols.

The steps in data science

In most data science problems there are 3 steps:

  • Data Collection
  • Feature Engineering
  • Tuning an Algorithm

Let's take a look at those steps in more detail:

Data Collection

Often the data is given to you. At other times you decide yourself what data to collect.

Feature Engineering

This step prepares the data for the classification algorithm.

Here are some examples of what can be done in this step:

  • Drop fields that are not relevant.
  • Transform categorical fields into numeric values.
  • Normalize the data, for example scale them such that the variance is always one.
  • Orthogonalize the features to get them independent from each other.

These are relatively simple measures, but Feature Engineering can also be very complex.

Feature Engineering for image data

Image pixels are just not helpful as features. There is no way how you could use them directly for detecting skin cancer for example.

Therefore with image data you often try to derive better features from the images by running them through a deep neural network that has been pretrained with general image data. To do that you remove the classification layer of the pretrained network and replace it by your own classification layers. The output of the pretrained network with the classifiation layer removed can be regarded as high level features for your images.

Tuning an algorithm

While you can use pretrained algorithms for your feature engineering in the end you always have to train your own algorithm.

This step has a lot of insecurities:

  • You need to get the parameters right.
  • Intialization matters.
  • Batch size matters.
  • Choice of test versus training data matters.
  • There are a lot of different algorithms to chose from.

Since this final step takes so much effort, you normally want to make sure that your features are good enough before you start finetuning an algorithm.

The quality of features

So the question in feature engineering is always, when are we done with it. How can we be sure that our features have a good quality.

What does feature quality mean?

What we want to classify are objects. These objects are representated by their features. So what we hope for is that objects that belong to different classes also differ significantly regarding their features. Then we stand a fair chance of classifying them.

Example: Fruit market

Let us illustrate feature quality with a simple example. If the task would be to distinguish fruits on the fruit market:

  • the size might be a good feature to distinguish between grapes and apples, but not helpful to tell apples and oranges apart.
  • the color might be helpful to decide between apples and oranges.
  • the combination of size and color might be good enough to classify all the fruits on the photo above.

How can we judge our features?

For judging our features the easiest thing would be to visualize them. If we have only 2 features we can draw them on a map.

Example Iris data

The Iris data are a typical toy dataset. They come with 4 features and 3 species (see for example

You can easily plot 2 features on a map. But that way you can only judge on the combination of 2 features.

Here you immediately see that these sepal length and sepal would not suffice to tell all three iris species apart.

But what we are interested in is not the pairs of two features, but in the combination of all of them.

Dimensionalty Reduction

So the problem of visualizing the features becomes one of dimensionality reduction: We want to map high dimensional feature spaces onto 2 or 3 dimension in order to visualize the distribution of the features regarding the classes.

Real world problems often have a lot of features

The data that we are given in real problems are often very complex. It is often unknown which features will be relevant.

In the case of the image data we have neural networks picking the features by themselves: So we don't even understand what the features mean since they have been picked by a machine.

So how can we put the features on a map?

Feature spaces are usually high dimensional vector spaces. So how can we map them onto 2 or 3 dimensions?

Linear Mappings

In vector spaces there is linear projection as one method to reduce dimensionality. But linear projections may map distant objects to nearby points, or even to the same point.

PCA stands for Principal Component Analysis. This method is a linear projection that optimizes the choice of the plane to project onto in such a way that it computes the 2 directions, that show the most feature variance.

Even though this sometimes may be a good enough, we can do better.

We can do that by just looking for a set of points in the plane that have a similar structure to the objects in features space: When the objects are close in feature space, the corresponding points in the plane should also be close to each other and vise versa. We no longer require a linear map from the feature space to the plane, in fact we do not require any such map at all.

t-SNE embedding

t-SNE stands for t-Distributed Stochastic Neighbor Embedding.

It constructs a probability distribution of the vectors in the feature space and finds a similar distribution for them as points on a map. That means that objects that are close to each other in feature space have a high probablity of ending up close to each other on the map and vise versa.

It contains a lot of randomness. If you perform the t-SNE twice the result will slightly differ each time. There are also parameters that go into it. If t-SNE does not give you a good result, it might just be that you got those parameters wrong.

Despite those inaccuracies the t-SNE visualization works surprisingly well in practice.

t-SNE was developped 2008 by Laurens van der Maaten and Geoffrey Hinton. Here is a link to page Laurens van der Maaten's own page on the t-SNE, where you can find further literature on it:


So far this might sound all very dry to you. When I first learned about t-SNE it was on
Kaggle, a place where data scientists compete against each other and practice their skills. I participated in a contest on insurance data, where it was not at all clear whether the features we were given were meaningful. Someone used t-SNE and when I learned about it,
I got hooked and it has now a fixed place in my datascience tool kit. I try it on almost every data science problem I encounter.

To fill t-SNE with practice, I will take you through some of my examples:


The first example is the Titanic. This is a well known beginner's task for future data scientists on Kaggle:

You get the historic passenger list of the Titanic and have to predict which passengers will survive the catastrophy. This shouldn't be 100% predictable, but there might be certain privileged or underprivileged passengers that have a higher or lower likelyhood to survive.

Here is the t-SNE visualization for this problem that I published on Kaggle


  • There are clusters where the majority of people survived. (Privileged passangers: women, children and 1st class?)
  • There are other clusters (red) where the majority of people died. (Under privileged passenger: men and 3rd class?)
  • Most clusters seem to be mixed to a certain degree, this is in accordance to the fact that fate will certainly have played a role in this.


This was another dataset on Kaggle:

  • The task was to tell seizures apart from normal brain activity.
  • It was supervised learning with two outcomes: seizure or normal.

Here you can see the t-SNE that I computed for this task after some basic normalization of the data:


  • the seizures where mostly detectable
  • the border between normal and seizure was fluent and not clear cut


Dog Breeds

I did this in a data science class, therefore I cannot publish my computations.

The task was to build a classifier for dog images, that would be able to distinguish between 137 different dogbreeds: If you show it a photo of a dog, it should be able to tell you the breed.

That is not an easy task, at least it would not be for me as a human.

For this task we were given pretrained image classifiers built by Google. These are publicly available. You can think of them as machines which were trained on a very general task of image classification. Then you take the classfication layer off and feed them with your own data. For the classification task we had to replace the original classification layer with our own classification layer.

We were given 4 pretrained Google image classifiers to chose from: So how would we know which one of these would give the best results?

This is again a question about feature quality. So I used t-SNE to visualize the feature quality of each pretrained classifier.

Below you can see the t-SNE visualization for two of pretrained classifiers VGG-19 and ResNet-50:

  • VGG-19 has 19 layers
  • ResNet-50 has 50 layers

Both are convolutional neural networks that were trained on more than one million images from the ImageNet database. There input are 224-by-224 images. They were pretrained to distinguish between 1000 object and animal categories.

In the picture I just highlighted 9 different dog breeds that I chose randomly in order to see how good the pretrained neural networks were separating the dog breeds:

Pretrained with VGG19

Pretrained with ResNet50


  • With ResNet-50 the clustering is a lot better regarding the dog breeds. There are almost no mixed clusters,
  • even though some dog breeds such as the Chihuahua are quite spread out.
  • With VGG-19 on the other hand only some breeds such as the Giant Schnauzer form clusters at all. Most breeds are quite spread out.
  • I went with ResNet-50 and got 82 % accuracy, meaning the dog breed was identified correctly in 82% of the cases.

t-SNE as recommender sytem for Freitag bags

Liip recently developed a recommender system as prototype for Freitag, a company that sells bags. The protype uses t-SNE to come up with a map that places similar bags next to each other. This enables the consumer to chose a bag that matches their taste by gradually moving through a plane of choices grouped by similarity.

For more details see on our Data Services Gallery.

Wingo – online and pragmatic Thu, 23 May 2019 00:00:00 +0200 A user-centered design

Wingo is the 100% online telecommunications brand for millennials. It is part of Swisscom's product portfolio. Wingo wanted to work with Liip to develop a new website focused on users. We are known for our user-centred approach and the quality of the delivered products.

The 5S model and Liip

We used the 5S model by Garrett to define the user experience (UX). 5S for Strategy, Scope, Structure, Skeleton and Surface. Systematically we went through these five steps. They form a logical sequence, starting from an abstract level and gradually focusing on the most concrete aspects. This ensures that we are able to handle all the elements needed to design an intuitive user experience. And to build solid foundations for the development of the website.

Figure: On the Strategy plane, we only care about how the site will fit into our client’s strategy while meeting the needs of the users. On the Surface plane, we focus on the concrete details of the appearance of the site. Source: Jesse James Garrett

The 5S model and a strong branding

The collaboration started with Wingo’s branding guidelines. We interpreted them to make them compatible with UX best practices. A website has to meet a set of criteria to be accessible. All the needed information has to be found quickly so that the user wants to stay and come back.

Simplifying the information architecture was the starting point – a key step of the project. It relates to the first three S's of the 5S model: Strategy, Scope and Structure. Afterwards, we worked on the design elements related to the fifth S, Surface. We came back to the fourth S (Skeleton). The skeleton of each page was created using the branded components validated upfront. This small gap has made it possible to integrate certain graphic constraints and to manage the stakeholders regarding branding more effectively.

The ingredients for success

Wingo gave us the opportunity to demonstrate our UX skills. Two designers were integrated in the Scrum team and one of Wingo’s employers was fully dedicated to the project. This contributed greatly to the success of the project. Furthermore we encourage the client to make decisions quickly and regularly to ensure that deadlines were met. Knowing that it is always possible to iterate during the next sprint. The use of the 5S model is also one of the key elements of the project's success.

"Working with the 5S methodology allowed us to focus on the essentials at each stage of the project. Although our deadline was very demanding – thanks to Liip's expertise and the working methodologies they apply –, we were able to launch a new quality website on time."
Maëlle De Bernardini, Wingo

Practice over theory: test and iterate

We built on our knowledge – surpassed ourselves. We found solutions that we couldn't have imagined at first. The success of a team is when solutions are challenged and improved through the combination of everyone's skills. In other words, we are enabling clients to achieve their goals while meeting the users’ expectations.

Contact us to find out how to improve the user experience of your website or mobile application while ensuring a strong branding.

The new maBCF portal is online Mon, 20 May 2019 00:00:00 +0200 Practicing Scrum since the foundation of Liip, we lead complex projects that require frequent changes during their implementation. The maBCF portal is one of these projects. We mastered the infrastructure and security constraints imposed by the client's field of activity. And the integration of a User Experience (UX) designer into the Scrum team increased agility.

First of all, we defined and developed a minimum viable product (MVP). This is the version of the platform that meets the needs of users by minimizing the investment of time and money. This first version was released a few weeks ago - already working on an update. New features will be added soon.

Our solutions are designed to evolve, based on user feedback.

Agile and efficient

To meet BCF's request to create a customer space on its website, we gathered a Scrum team. A Scrum Master, a Product Owner, four developers, a UX designer and a content specialist have set to work.

We are convinced that it is necessary and beneficial to integrate the client into the project team. The constitution of a multidisciplinary team including the BCF team allowed a great efficiency during the 11 consecutive sprints (22 weeks). While respecting major constraints: the infrastructure and security required by the banking industry.

At Liip, we are driven by several principles. Our excellent collaboration with the BCF is perfectly illustrated by our principle "flexibility over strength". We are convinced that only an agile and iterative approach can meet these challenges related to the banking industry. This flexibility also makes it possible to deploy the customer's strategy while meeting user expectations.

Tailor made and user-centred

A client's product is unique. The solution we develop have to be just as distinctive. The UX designer participated in each of the 11 sprints, in order to make the creation process even more agile. In other words, to best meet the expectations of the users of the future maBCF portal while optimizing the work of the Scrum team.

Models created during the sprints

Often all the models are prepared by the UX designer before the first sprint begins. There is therefore a risk that the designs no longer meet the needs of developers, as the project scope evolves as development progresses.

During the maBCF project, we included the creation of these models into the sprints.

The UX designer was preparing the design elements for the N+1 sprint when the developers were working on the N sprint.

This approach allowed us to be more responsive to project developments. But it also presented a major challenge. Ensuring that the models needed by developers were available at the right time. So to ensure that the sprint runs smoothly. The planning of the UX designer's work was therefore crucial.

A styleguide as a compass

During the first workshop, we designed sketches of the future platform together with the client . The designer used them to create models, directly in the styleguide. This form of graphic charter is an essential tool for developers. It includes a library of styles and components. Besides, the client can visualize in the styleguide the page layout and the behavior of the different functionalities.

The BCF team was able to make decisions fast based on these 1:1 scale prototypes. Because the styles matches exactly the final product.

We are convinced that the scope of a project will and have to evolve during its development. To be as close as possible to the expectations of users.

This first experience was a success. We are excited about the rewarding collaboration we have created with the BCF. And we are proud to have put our principle of "flexibility over strength" into practice. We will systematically integrate the creation of designs during our next sprints on the maBCF project, but also during other projects.

Content is king, and AI will rule the world Thu, 16 May 2019 00:00:00 +0200 With more than 300 speakers and 50,000 visitors, the OMR conference is the biggest online communication event in the German speaking market, and a strong indicator of where the industry is headed. You can read all about our discoveries below.

Big topic #1: Content is replacing advertising

The topic of content was omnipresent throughout the conference. From general insights to full-on use cases, it was a hot topic that's sure to stick around for the foreseeable future.

Information and entertainment are one winning team

Image left: coffe roaster J.Hornig on chat bots – Image right: Facebook presenting user data

Despite the digitalization of our lives, humans will always be humans: We want to solve our daily problems and follow our interests. This is the point Ana from Austrian coffee roasters J. Hornig underlined with the quote from advertising legend Jean-Remy von Matt: “Good communication either entertains or supports.” And that’s why she and her team have chosen to use a chatty chat bot to help coffee lovers at home with any questions they may have. Nadine from Facebook pushed the point further with the platform’s statistics: Information (45%) and entertainment (46%) is what people seek out on Instagram.

Be patient & stay focused – success will follow

Left: Content Marketing Institute presents their magic formula – Right: Territory trusts in persistence

Brands have figured out that content is more important than advertising, and many have taken content production in-house. Some are even tracking the success of every bit and byte they post – like Joe Pulizzi’s Content Marketing Institute. His learning: Once you start putting out content regularly, it takes 12 to 18 month before you start seeing a return on your investment. Soheil Dastyari, CEO of content agency Territory, came to the same conclusion.

Involve your audience

Left: Investor David R. Bell believes in storytelling – Right: OMR believes in story sharing

The most powerful brand message is the one delivered by the brand’s users. That’s why investor David R. Bell puts his money into start-ups that have a story people want to share. “Orators not customers” is how he summarizes this trend. OMR conference founder Philipp Westermeyer presents the agrees that sharability is the new brand currency. His example to illustrate this trend is Balenciaga who created an extremely expensive Ikea bag replica just to create buzz.

Voice & audio - the revolution has begun

Left: "Gimlet's" podcasts give brands a voice – Right: "OMR" observes investments in voice recognition

Matt Lieber, cofounder of the podcasting platform Gimlet, knows that the second revolution of voice is upon us. The company has just been bought by Spotify and is now reaching millions more listeners. He presented some great insights: Gimlet identified that people have around 2.5 hours per day where they seek entertainment yet cannot look at a screen, like during their commute or when cooking. This is all potential time for a podcast, and the perfect place for a brand to be present for their users and deliver valuable content. An even bigger shift is predicted in voice command. Philipp Westermeyer underlines that with the example of how Amazon is investing in voice recognition. He explained that Amazon currently employs 10,000 people supporting its Alexa project and he showed us a page from Amazon’s job platform with more than 2000 open jobs in the area of voice regocnition at the moment.

Big topic #2: Artificial intelligence and use case “chat bots”

Yuval Harari’s vision for a world with AI

Image: Yuval Harari ponders all the (unexpected) issues AI will raise.

AI will soon know us better than we know ourselves. “What will happen to us when Coca-Cola finds out I’m gay before even I know?” Yuval Harari began his speech with a personal story: He didn’t realize that he’s gay until the age of 21. Yet in retrospect, he says, the signs were obvious. So a machine tracking his eye movement would have noted that he glances at the man in a picture instead of the woman. With that information in hand, Coca-Cola would then present him their product in conjunction with his deep and unconscious desires. And he would always opt for that brand because unconsciously he feels they understand him best. He has plenty more examples of how AI will intrude in our lives: What will happen to relationships when the refrigerator delivers on our needs us more precisely than our spouse? How will society change? His scenarios go even darker: What will happen when North Korea forces its citizens to wear bio-sensors that detect anger when looking at Dear Leader? His advice: Technology takes no side, so we must choose how we use it wisely.

China is way ahead of us

Image: OMR observes heavy investments in china

The many speeches about the Chinese market (speech 1, speech 2, speech 3) made it clear that we’ve got to keep watching out for what’s coming out of China. Predictions are that what we see today is just the very beginning, as huge investments are being made overseas, especially in AI. Philipp Westermeyer's example shows that a rather small city in China invests almost 10x more into AI than the entire European Union.

Chat bots become a commodity

Image: AI agency knowhere categorizes AI in three use cases

Frederik from AI agency Knowhere presented several use cases for chat bots. The ones we liked were the ChatYourself chat bot, which asks early Alzheimers patients questions about their life in order to support their memory later on. Another useful one for daily life is the Novi chat bot, which presents news in a user-friendly way. And then there’s the golden state warriors chat bot that let's you discuss sports results. Toothpaste brand Signal’s chat bot supports kids when brushing their teeth, and the alcoholics anonymouns's chat bot let's you discuss taboo topics in a private way. Medical device producer Dräger was also on stage to present their process to install a chat bot. After using the bot for about 6 months, they can already say it has reduced the amount of incoming questions by around 45%.

Chat bots win over a website and an app

Image: coffee roaster J. Hornig's presents their arguments for a chat bot

The coffee roaster J. Hornig wanted to create a “barista in your pocket.” They were looking for a digital solution for anyone with a coffee-related question. When exploring possible implementations, a chat bot won. Easy accessible via voice, conversational in tone, and with the free chat bot solution provided by Facebook Messenger, a single employee at the roastery could create a proof-of-concept within a month.

Consultancy firms have become the new marketeers

Image: Accenture presenting their view on AI - with the help of Arnold Schwarzengger

Interesting things are happening in the world of communication. Is it a good or a bad sign when Accenture and McKinsey have several slots in a marketing conference? Either way, it’s a solid indicator that data is taking over. Just as The Terminator predicted.

Side note: Agencies are dead - long live agencies

Image: Content agency wldmr illustrates how agencies and clients become one team

Platforms like Instagram, YouTube, Facebook, Snapchat, and Tiktok all held several speeches and masterclasses. Data agencies were quite present, as were several brands presenting their cases. What was missing? All the big creative agencies. If an agency was on stage, the focus was always on how they form a team with the client to yield the best results. The days where the agency was boss and told everybody how things are done are long gone. And we think that’s for the better, as brands know their own story best and we all need to find the user’s need in a joint effort.

This is actually what we at Liip believe in too. And visiting the OMR conference has underlined how important that is. We also believe in data and data analysis. The only thing that we don’t do: Work with influencers, a topic that was also quite present at OMR. Something to focus on next yea!


Big Topic #1: Content is replacing advertising

  • Information + Entertainment = A winning team
  • Be patient & stay focused – success will come
  • Involve your audience
  • Voice & audio - the revolution has begun

Big topic #2: Artificial intelligence and use case “chat bots”

  • China is way ahead of us
  • Chat bots become a commodity
  • Chat bots win over a website and an app
  • Consultancy firms become the new marketeers

Sources: header image

The experts behind this article

Thanks to Lena, Jenny and Daniel for content and copy cleverness. This article would not have been possible without you!

Cognitive UXD: Motivation & Flow Wed, 15 May 2019 00:00:00 +0200 In my first blog post Cognitive User Experience Design - Where Psychology meets UX Design I gave an overview about combining Psychology and Design. As mentioned in my second one, Cognitive UXD: Motivation, the topic Motivation is very comprehensive and has a valuable impact on design. Therefore, I take you on a short tour and describe the role of flow in design. In short, knowing the Flow Theory can help that the user comes or maintains in the flow zone. The aim is to “Go with the Flow!”

What is the Flow zone?

According to the psychologist Mihaly Csíkszentmihályi an excessive challenge leads to feelings of overload, frustration and anxiety. Contrary, excessive skills can cause feelings of underload, routine and boredom. Matching challenges to skills is therefore considered a key factor for flow.

Source: Concept by Mihaly Csíkszentmihályi; modified by Viktoria Kluckner

The state when challenges and skills are in the right balance are known as to be “in the flow”. Within this state, the person is fully immersed in a feeling of energized focus, full involvement and enjoyment of the moment. Being within this flow zone, the person loses track of time and/ or space, performs the task for a sheer pleasure, is intrinsically motivated and much more productive and happy.

How can we ensure to bring or to remain the user in the flow zone?

First, you need to know who your clients and your users are. What are their needs, desires, expectations, approach f.i. during the purchase process. The easiest way is to ask questions.
Here is a general selection that you can ask the client:

  • Who is your ‘typical’ user?
  • Why does the user buy the product?
  • Where will they use the product?
  • What are the most important tasks the user should fulfill when buying the product?
  • With which other products will yours be compared with?
  • What should the user think, feel or say?
  • What happens after the user bought the product?

Next, ask the user:

  • Why do you want to buy the product?
  • What are the most important tasks?
  • What do you think, feel or say?
  • What helps you with the purchase decision?
  • What happens after you bought the product?

With these first questions you get a first overview of the purchase process. Of course you need to dig deeper to get a more detailed understanding. You can ask for the reasons to buy one specific product:

  • What specific feature was the most attractive one for you?
  • What makes this feature so attractive for you?
  • How can you touch the user emotionally to make him a loyal user or that he remains a loyal user?

As you can see, a look from different perspectives is necessary. The more detailed and holistic you understand and involve your users, the better you can create experiences for them that will carry them along or make them to remain in the flow zone.

In this phase you have already achieved a lot. But don't forget to test and improve.
By measuring, you can see where the user flow is working and where it is not. Find out where a smooth process is interrupted and where the user falls out of the flow zone. Take a look at the dropouts and do a Usability Testing. Especially in Usability Testings you can directly observe the facial expression of the user and the behavior during the execution of tasks, which gives you even more information. You can see whether the user is in the flow zone, i.e. whether he is fully immersed in a feeling of energized focus, fully involved and enjoying the moment. Take the time to test. It's worth it and the key to make it right. You will receive clear indications of starting points to improve the experience.

Which recommendations can help?

As UX designers, we should ensure that users enter and stay in the flow zone to experience joy. These recommendations can help to increase flow in the design:

  1. Guide the users by providing information on what to do and how to do it.
  2. Ensure that the users understand where they are and what tasks they need to perform. Create f.i. a clear and intuitive navigation structure.
  3. Find out what disrupts the user’s workflow. Reduce confusion and interruptions such as notifications or bugs.
  4. Be close to the user’s semantic to reduce the cognitive burden. Think about how the user naturally wants to interact.
  5. Think about what is really needed and delete unnecessary interactions, content, etc.
  6. Ensure that there are no annoying, repetitive tasks or functions. This can be reduced with Usability Testings. Think about how you can make the experience smooth.
  7. Set clear and achievable goals such as the note to check the 4 meetings you have today in your calendar.
  8. Divide a big goal into several smaller ones. Let the user achieve more quick results while driving the big goal forward.
  9. Think about how the communication be improved to achieve the goal.
  10. Provide accurate and immediate feedback on actions. Visible, auditory or tactile feedback can give confidence that a process is running. If you press a calendar entry for a longer time, it gets a little bigger and you can move it to an earlier time.
  11. Give some rewards such as a free downloadable e-books or something similar that makes the user happy.
  12. Provide challenges that match the skills of the users. It is very important to have a good understanding of the users and their skills and to challenge them to stay interesting.

I hope that I could give you some interesting questions and recommendations. I wish you a lot of fun to create inspiring designs in order to bring or to remain the user in the flow zone. At the end of the day, our aim is to satisfy users.

In my next blog post I focus on Cognition and describe its role in design based on the Cognitive Load Theory.

Love is in the air Tue, 14 May 2019 00:00:00 +0200 The increasing use of social media, artificial intelligence and personal assistants could make us forget that we are human beings. Living beings. Loving beings. The motto of the 2019 edition of TEDxFribourg is LOVE. This 5th edition brought together enthusiasts who remind us that love is everywhere. In every action, every project, every adventure, every story we live.

Our crushes: Lovely ideas worth spreading

The topic of TEDx Fribourg 2019 was love. Eleven speakers made us travel for more than two hours. You will find a selection of my personal highlights below. And here are the videos of the presentations and the photos of the evening for you.

Johana Dayer and the wine, a vintage of love

What is love? Whether it is a passion, love for others, love for our planet, love for humanity, its strength lies in sharing. Johanna Dayer thus evokes wine as an experience that we savour together. An emotional story, actually.

Perfect ikigaï for the founders of The Green Drop

Estelle and Eléonore from The Green Drop found their ikigaï. The ikigaï in Japanese is the perfect alignment between passion, mission, vocation and profession. We recognize ourselves in this quest for meaning. We are convinced that an action makes sense when we are personally driven and when it is linked to our values.

Azadeh Tadjar: love has the power to provide solutions

A lot of people share the idea that a meaningful approach comes from the heart. Problem-solvers are all over the world. They are so in love with what they do that it is hard to not be inspired. Azadeh Tajdar is convinced by the contagious power of love. And so are we. Let’s share these ideas and spread the love.

These experiences of life shared by the speakers of TEDxFribourg reminded me of several projects we led. Projects are nothing without people, whether users, customers or Liipers. Like One.Thing.Less which enables individuals to regain control over their personal data. Or like Urban Connect and Smart Energy Link, two projects born from the commitment of start-ups for a more sustainable future. Or Freitag City Guidelines, which give a voice to local heros who are passionate about their cities.

Our contribution to TEDxFribourg: Make a wish

If we participate in a TEDx event, we foster participants to make a wish. More than 30 wishes were hung on our Make a Wish tree. The majority of authors want more love, harmony in the family, happiness and prosperity in a world that better protects the environment. A few wishes were related to a personal project or the search for meaning in daily work. Not to mention the wishes related to new forms of work organisation.

Our commitment: Cheers to love

Love is also about human relationships, about what unites us. What if it starts over a glass of cider? This year, we served apple and pear juice and cider at Liip Bar. We spoke to Jacques Perritaz from the Cidrerie du Vulcain. An enthusiast who is committed to the cultivation of old varieties of apples, pears and quinces in the countryside of Fribourg.

We like these events where ideas pop up, during a conversation, a drink or on a Make a wish card. And what we like even more is to develop these ideas with you.

To wrap it up: Cheers to love!

Pourquoi j'aime être une apprentie informaticienne chez Liip Mon, 13 May 2019 00:00:00 +0200 Pourquoi l’informatique ? Pourquoi ce choix d’apprentissage ?

Parce que j'apprécie l’informatique depuis que je suis petite. Et je me suis dit “pourquoi ne pas essayer de faire quelque chose que j’apprécie ?”.

Comment c’est l’informatique pour les filles ? Comment est-ce que tu perçois l’informatique en tant que fille ?

J’ai l’impression qu’il y a de plus en plus de filles qui veulent s’engager dans un métier qui touche à l’informatique. Je trouve que c’est bien parce que les métiers ne doivent pas être réservés à un genre en particulier. Il n’y a pas encore beaucoup de filles qui font l’apprentissage de l'informatique dans mon école. Par contre, au bureau de Liip à Fribourg, nous sommes quatre apprenti·e·s en informatique, dont deux filles.

Qu’est-ce que cela fait de travailler dans un métier occupé principalement par des hommes ?

Pas grand chose, parce que les femmes sont très bien intégrées dans l'entreprise et il n’y a pas de différence. Chez Liip, un homme ou une femme aura le même salaire et c’est pareil pour les apprenti·e·s. La seule vraie différence est à l’école. Dans ma classe, on n'est que 2 filles sur 21 élèves.

Pourquoi est-ce que des filles devraient faire ce métier ?

Parce que ce n’est pas un métier que pour les hommes et qu’on a le droit de faire ce qui nous plait sans contrainte.

Qu’est-ce que cela fait de travailler chez Liip ? D’avoir Liip comme entreprise formatrice ?

C’est génial parce qu Liip ne fixe pas beaucoup de règles. Cela me demande d’être autonome et responsable. En plus, il y a vraiment une bonne ambiance dans l’entreprise. Tout le monde se donne bien. Les apprenti·e·s sont considéré·e·s comme des employé·e·s à part entière. Et cela nous permet de bien nous intégrer à l’entreprise.

Quelle est l’ambiance au bureau ? Quelles sont les activités organisées chez Liip ?

Liip organise beaucoup d’activités, autant pour les apprenti·e·s que pour les employé·e·s. Par exemple, une fois par an Liip organise et finance une journée de conférences durant laquelle les cinq bureaux se rassemblent. Il y a aussi des journées dédiées aux apprenti·e·s. Durant celles-ci, les apprenti·e·s des cinq bureaux de Liip se rassemblent pour parler de leurs projets.
Nous organisons aussi des journées portes ouvertes à Fribourg pour toutes les personnes qui aimeraient découvrir le métier d’informaticien·ne de développement. Cette journée est entièrement organisée et animée par les apprenti·e·s de Fribourg. Ou encore, on partage nos repas de midi ensemble, on fait du sport et plein d’autres activités.
En ce qui concerne l’ambiance, c’est génial car tout le monde s’entend bien.

La pause parfaite selon les apprenties : Sofia et Ksénia jouent à Mario Kart

Qu’est-ce que tu apprends ? Parle-moi d’un des projets sur lesquels tu travailles.

En ce moment, je travaille sur un site web au sein du bureau de Fribourg. J’ai aussi développé un petit jeu pour apprendre les bases de la programmation. J’ai également appris à gérer un serveur de gestion de publicités en ligne. J’apprends aussi beaucoup sur les autres domaines de l’informatique à l’école.

Quelles sont les difficultés que tu rencontres ?

En soit, c’est un métier assez difficile. Il y a beaucoup de notions à connaître. Par exemple pour faire de la programmation, je dois aussi connaître des notions des autres domaines de l'informatique. Il y en a trois: informatique d’entreprise, développement d’applications et techniques des systèmes.

Quelle est la journée type d’une apprentie informaticienne ?

Il n’y en a pas vraiment. Mes journées se ressemblent mais ne sont jamais pareilles. Je ne vais pas faire exactement la même chose que le jour d’avant ou que la semaine précédente.
En plus, Liip est une entreprise qui favorise la “self-organisation”, c’est-à-dire que les employé·e·s s’organisent tou·te·s seul·e·s. Par exemple en tant qu’apprenti·e·s, on doit faire des journées de 8 heures mais les horaires sont flexibles. Une de mes seules contraintes est d’être présente lors du daily le matin. Je peux aussi prendre une pause de midi plus ou moins grande.
Il y a aussi des périodes entièrement consacrées à l’école. C’est-à-dire que pendant plusieurs mois, je vais tous les jours à l’école. Le reste du temps, je le passe en entreprise. Au final, je ne trouve pas qu’il y ait une journée “type” pour un·e apprenti·e chez Liip.

Début mars, j'ai préparé du contenu pour le compte instagram de Liip afin de présenter à quoi ressemble mon apprentissage d'informaticienne.

Qu’est-ce que tu changerais dans ton apprentissage ? Et pourquoi ?

Personnellement, je ne changerais rien. Parce que je fais quelque chose que j’aime. Parce que je travaille avec des gens que j’apprécie. Bref, je travaille dans une entrepris juste parfaite.

N'oublie pas de lire les réponses de Ksénia !!!