16th May 2022No Comments


  • Understanding machine learning, including the basics of how it works and the different types of it.
  • Collaborating with data scientists, understanding their process, and facilitating collaboration as a product designer.
  • Translating algorithm outputs to create user experiences, and the different considerations and tradeoffs to be aware of.

Understanding machine learning

First off and I guess the most obvious thing to learn is the basics of how machine learning works. I highly recommend a course by Andrew Ng founder of deeplearning.ai called ‘AI for everyone’ — It’s very comprehensive, easy to watch, and will give you all the context you need to get started. Also, it’s worth keeping this glossary at hand in case any of the terms don’t make sense.

Graphic illustration of a diagram showing machine learning proccess

As you can see from above, a key part of any machine learning model is the data that powers it. As a designer, it’s important to understand what types of data are out there and how data gets sourced and used. This should give you enough context to be involved in conversations around data and watch out for bias in the model. Bias in AI is a hot topic and I’m sure there are many articles about it, however, one thing to call out is that typically bias happens from poor or incomplete data that amplifies biases in the real world.

Finally it’s worth learning about concepts related to machine learning model outputs as you will be using these outputs to create experiences for users. Particularly concepts like confusion matrix, accuracy, explainability, precision, and recall. The impact of these concepts will be covered in the last part of this series, but it’s good to familiarise beforehand.

Collaborating with data scientist

Having foundation knowledge of how machine learning works will definitely propel you towards success, however in my opinion knowing how to effectively collaborate with data scientists is one of the most important skills for a product designer when working on machine learning-driven products or features.

Graphic illustration of a diagram showing machine learning process

Translating algorithm output to create experiences

Final skill to highlight is the ability to translate machine learning algorithm output into user experiences. Translate is a useful word here because, just like languages, understanding one is necessary in order to construct a sentence in another. Similarly, in order to provide a seamless experience for users, you need to understand how to use the outputs of machine learning algorithms. Understanding the fundamentals of how machine learning works, as well as working collaboratively with data scientists, will definitely help in understanding the output — however, translating it into something intuitive and easy to understand for users is a skill on its own.

19th January 2021No Comments


If you’re a user experience or a product designer, you will be very familiar with the concept of human centred design —a problem solving approach that involves people in every step of the process. It is woven into everything we do as designers and we often champion it in our lines of work. However, the design landscape is always changing and some could say even faster than ever, especially with emerging technologies like machine learning and AI.

Over the last couple of years I had the opportunity to work as a product designer exclusively on machine learning driven features and products. Over time I noticed that there are multiple significant considerations and tradeoffs that need to be taken into account. Not being aware of these considerations leaves a lot of room to damage the experience of your products and sometimes even the overall perception of the brand.

That’s why I decided to capture what I learned in an article series — ‘Machine Centred Design’, where I will be covering core skills and best practices when designing machine learning driven features and products. The article series was written with user experience and product designers in mind, but it’s a useful resource for anyone working in product or tech.

It’s worth noting that this space is relatively new and I’m still learning, so there might be some things that I have not captured. Also, there are multiple companies that have dedicated teams researching this, so I will be linking out to their resources where appropriate. Lastly, no prerequisite knowledge about machine learning is needed, but having a general understanding would help to understand a few concepts.

The series will consist of three parts:

Part 1 — Will give an overview of core skills needed when designing machine learning driven products and features. It will cover understanding machine learning, working with data scientists and translating algorithm outputs to create experiences for users.

Part 2 — Will dive deeper into collaboration with data science and where I will explain why it matters, how it differs from other stakeholders and expand further on the actual process.

Part 3 — The final part of the series will dive deeper into understanding algorithm outputs and translating them to create experiences for users. Covering concepts like explicit vs. implicit data capture, precision vs. recall and many more.

I’m excited to be sharing this article series with you and explore the intersect between user experience and AI. Follow me for updates and make sure to like the article if you found it interesting. The next part in the series is coming out very shortly.

5th February 2019Comments are off for this post.


This blog post was written as part of FutureLearn Medium blog, follow the link below to read the original article:


5th February 2019Comments are off for this post.


If you’re a product person you’ll know that user interviews are great, they give a substantial amount of qualitative insights that can be used as foundations for a product idea or to validate one. Not only it provides a lens through your target user group, it is also a great activity to involve and bring on board different parts of the business. User quotes can be a quite effective way to demonstrate and justify product decisions.

Recently I’ve been involved in a round of user interviews where I decided to take a little bit different, yet common approach, which is more hypothesis lead.

This idea is nothing new or radical it’s been around for a while, on the most basic level it’s going into a research activity with an already formed assumption that you might want to validate, and then learning from user responses and reactions to that assumption.

Including it in the digital product environment brings some interesting benefits and challenges. In my view, there are three main benefits: team buy-in, efficiency, space for bold changes. And a couple of challenges like designing upfront and bringing stakeholders on board.

To go a little bit more in detail on the benefits and challenges I will briefly discuss them below:


Team buy-in, hypothesis led research allows you to collate all the different opinions from the team and put them to test. This way all members of the team get to be part of the research, feel involved in it, and be more receptive of the outcomes of research. It also validates or disapproves any opinions that the team may have, reducing the risk of unconscious bias towards one or the other idea.

Efficiency, this links to the benefit above, with hypothesis led approach you can really focus on delivering best user experience possible without any internal voices in the company cherry picking insights, it also involves team members throughout the process meaning less reporting and reviewing.

Space for bold changes, this slightly goes against an agile iterative approach, but going in with a hypothesis allows you to test bigger changes in a controlled environment. You might not naturally get to do this, with an iterative approach, which can be especially useful within the discovery stage of the design thinking process.


Designing upfront, this approach usually will mean more design work to create different prototypes based on various assumptions. And as result discarding a lot of these designs in the process.

Bringing stakeholder on board, showing big bold changes can be discomforting to stakeholders and equally there’s a tendency for people, not familiar with the process, to think that the proposed designs are “final”. It’s important to emphasise, that it’s only an assumption that can be discarded and that the work attached to it will even out with more efficient research process.

Interview insights

In this round of user interviews, where I decided to take a more hypothesis driven approach, I also had the opportunity to conduct one of the interviews. It brought some interesting challenges and discoveries:

One of the key challenges was asking unbiased, non direct or leading questions. This can be really tricky as a product designer in a hypothesis led environment, as in the back of your mind you really want to prove or disapprove any assumptions. Thinking a few steps ahead helped with this, knowing where the conversation might go prevented me from guiding it in a specific direction.

Another, more general, challenge was the ability to control the pace and time. It comes with practice, but getting answers to all the question you have in mind without sounding like a robot and at the same time resolving any tangents can prove to be difficult. Again, thinking on your feet and changing the plan as you go along helps with this.

Personally, the most interesting discovery, out of this particular piece of research was when a user — shown exactly the same copy paragraph twice, just within a different colour background context, thought that the copy was completely different in each scenario. Even though there were more substantial and useful product discoveries during the research, this one proved again, that people don’t read, but scan. It also gave me a new found respect for UI/Visual design when it comes to UX.

In the end I would say that, hypothesis led research is something that I will try to keep doing forward, even though it can bring some challenges - it also brings some really great added benefits like team buy-in, efficiency and scope for bigger changes. I would definitely recommend trying this kind of an approach, especially if it’s during the discovery phase of the project.

11th March 2018Comments are off for this post.


Not that long ago I went for a hike to Yorkshire dales. Now I’m not the biggest hiking person but I did enjoy the views and physical activity that comes with it. However this post is not about the hike itself, rather about something that made it more enjoyable.

So as mentioned before I am no expert in hiking and I needed something to guide along this specific scenic route that I chose. Of course me being a millennial living in London, I started looking for an app that could guide me. I found a few, nothing too magical, but I think it would have done the job. However a person I was with insisted on using this old website that she has found, on the outside it looked like it’s still living in the 90s. But I thought why not it seems intriguing to use something like that. All it was a simple one page static website with bullet points interlinked with small paragraphs of text. It began by giving simple instructions how to reach the beginning of the route. Then on, it described visual clues that you can navigate by to find your way forward and the bigger paragraphs gave more information if you came across a point of interest.

This is where I found magic in its simplicity - by just following visual clues and brief descriptions there’s an added level of excitement, to figure it out how to get from one point to another. You feel a sense of accomplishment that you can navigate purely from a single bullet point – it was basically like solving a micro puzzle every time. The other great thing was that by following a website like that there was room for error, you would from time to time get a bit lost, which when you are on a hike adds a dimension of adventure - or you could say variable reward in product terms.
Now if you compare this to an app that was available - it would have given me a geo location based route drawn out on a map perhaps a guiding arrow to direct me even further along the route. That kind of a product would just get me from A to B and would get me there more efficiently than the simple website. However it takes away those two added dimensions, making the route less enjoyable.

That brings me to my point about what user needs and what user wants. If you would put the problem on a blank piece of paper it makes sense to provide users what they need - a clear route which they can follow and eventually complete. Yet that is not exactly what they really want, which is a memorable and adventures journey. I think all of this comes down to truly empathizing with the user you design for and understanding what is the core behavior problem rather than a surface problem. Different methodologies can help to identify the core problem, yet there’s always a danger to make an assumption and draw out a loosely defined persona, that can lead you into inherently biased decision on the solution. Real empathy on qualitative research, can help to steer away from solutions like that.

Now it's worth noting that all those apps, that provide direction and routes, have high rating on the App Store and not surprising as they serve a real function. However they don’t necessary bring value as an actual consumer product, they appear to be more like a tool that can be used once or twice and not like an app that you would actually encourage you to go on a hike.

Therefore it’s important to understand what the users really want, put yourself in the shoes of the user and most importantly use qualitative research and really look for an insight. Otherwise there’s a danger to make an assumptions on what the user needs and build something that resembles a tool and not a consumer product, which can extend beyond the function and grow user retainment, provide opportunities for monetization and scalability.

15th October 2017Comments are off for this post.


AI is becoming the buzzword of the year, we can hear it almost in any advertising article. All of the big brands are pushing AI as an element of their brand to appear innovative and up to date. Whether it’s just a PR stunt to start people talking or an actual integration in the brand system - AI is becoming a big part of digital marketing. And it’s true a really clever integration that enhances the brand message can be fascinating more so now than in the future, purely because we will get more and more of it. And there’s nothing bad about that, don’t get me wrong, I think it should be promoted even more. However, in the current world of digital advertising, there are a couple of areas where I see AI applications spiraling in the wrong direction.

First one being the classic riding the wave application which in a nutshell just takes a marketing or advertising execution and wraps it in an AI foil. The application itself could very well live on its own without the need of having AI hacked into it or could use another more tradition application to elevate it further. And I'm not talking about a one off PR stunt which in my eyes is only a natural occurrence when there is so much hype built around it. It's more about prolonged and actual long term marketing strategies. For instance, a brand could say they are utilizing AI to deliver personalized greetings on twitter, however saying that they are all personally written by a brand advocate can be in some cases more useful to deliver a brand message. And of course in both cases, they will be generated automatically but the technology behind it is not truly an AI and neither did the brand advocate write it himself, it just changes the message the tone of the communication and it's easy to assume that whacking an AI into it will make it more effective. Of course at the current climate, they may very well be true, but emerging technologies tend to blend with the existing ones and become a part of our daily lives and then the choice becomes even more important.

This links to the next, even more, future looking, instance where I see AI applications taking a wrong turn. We all remember or look back to with mixed emotions at the days of mass media advertising, it wasn’t that long ago but due to consumer behaviour changing within the digital age, social media was quick to take its place. Some principles for that change can be attributed to oversaturation and the change of preferred communication channels. One of them I think can be very true in utilising AI for marketing purposes. If more and more brands take on AI and use it throughout their services, which will probably happen, not only to change brand perception but because of the monetary value as well. We can end up in a world that is predominantly focused on AI interactions, most of us have seen videos of YouTube imagining a hyper realistic future where everything is controlled by AI and the city is overwhelmed by VR displays. Of course, the future will probably not look the same, because as someone said - if you can imagine how future will look like, it’s probably already the past. However some elements of this AI take over could be true and probably will happen, and for some aspects, it will be really great and exciting. However, I think as it happened with mass media advertising where authentic stories became a rare commodity, equally in the age of AI advertising human contact and interaction could become a rare commodity. With all the cashier desks changed to an automated service, with everything being integrated into IOT, what might happens is that people will start missing face to face interactions just like they started to miss authenticity and realness during mass media days. In that case, we might see brand cleverly hiring back cashier as brand strategy, having real assistants or gym trainers might become a luxury. To not reach to that point we should realise the value of humans as an asset, not as an element that can be substituted with AI chasing an up to date brand image or monetary value.

And I know right now AI doesn’t look like an advertisement channel or medium, but as we are changing from long advertising videos on TV to snackable content across different social channels the same way in my opinion product and service interactions will become a means of delivering a brand message and that very well might be a process automated by AI.

You made it :)
2022 © Benas Skripka