Visitor #00001, we have been expecting you.

This is your profile, created by FROSMO based on your browsing behavior. Each time you interact with the website, we collect data to create a relevantly personalized experience for you.

The University of Helsinki ensures better customer experiences through AI-driven personalization

The University of Helsinki is Finland’s largest and oldest academic institution. Since 1640, it has contributed to the establishment of a fair and equal society that is considered the best in the world according to a number of indicators. The figures are impressive: 40,000 students and employees; 50,000 web pages; and 3M monthly page loads. 

In the international Webometrics ranking for online visibility, the university placed among the top 10 universities in Europe and the top 60 universities in the world. It’s great evidence of the data-driven approach and level of high ambition in creating exceptional customer experience across all their online services. 

Their website provides information on nearly 100 degree-programs as well as doctoral education and lifelong learning opportunities. The website offers content for many other groups also; international academia, researchers, alumni, university funders, business, and political decision-makers, journalists, and citizens who are interested in science. The pages offer research information, latest news, and cooperation opportunities for all friends of research and science.

The university website is run much like a big media house with tens of different interest groups and thousands of articles. The online communications team at the University understood that a static website that looks the same for everybody is not the modern way of serving different interest groups. They believe that website personalization is the way forward to create better customer experiences. 

Frosmo’s personalization software was chosen to replace the competitor technology during fall 2019. The university uses Frosmo personalization software to create customer journeys with personalized hero banners, A/B testing, and recommendations that use NLP (Natural Language Processing).  

Clear KPIs are set for personalization, increased click-through rate (CTR), better customer experience with better website usability, and personalized content as well as better content discoverability. A/B testing enables data-driven decision-making and it’s been part of the university’s way of doing for many years.

“We see website personalization as the way to provide the best possible web experience to all of the University of Helsinki target audiences. It also strengthens our efforts for true data-driven service development.”

Anni Aarinen
, Head of Development, Digital Content, University of Helsinki


AI-driven NLP for better content discoverability and great customer experiences

One of the most important functions of the university is to produce reliable information for the whole society and the news and articles have a wide audience among “regular” citizens. The core of good customer experience is great content discoverability. AI-driven NLP is the best option to cater to the needs of many target groups by recommending articles according to the interests of the reader.

NLP at Helsinki University website

The Frosmo NLP algorithm gathers insights from all the articles of the university by analyzing the text, titles, lead paragraph, and scoring the articles for similarity, and then ranks them in order of relevance.

Recommendations at Helsinki University website
Recommendations at Helsinki University website

The CTR is an important metric to indicate the relevancy of the articles that NLP recommends and how many clicks the recommended articles get. The results have been great and more than 25% of the users who have seen article recommendations have clicked them.

Semantic relationships of words are handled by Fasttext incorporated with Frosmo’s algorithms to give the best articles similarity.


Data-driven website management through A/B testing

The usability and whole customer experience on a website this big are really important. It’s next to impossible to make good decisions without testing and adopting the best performing versions. A/B testing is widely used in testing new web designs, the order of content listings, different titles, and texts.

It’s fascinating to test larger implications of the new designs on the website, for example, to test whether the size of a picture has an effect on the scroll depth. Frosmo A/B testing is able to give us more in-depth data on web behavior than we would get from Google Analytics. Testing is part of our core and will definitely play a big part in gathering data for our new website design project.

Anni Aarinen, Head of Development, Digital Content, University of Helsinki


Interested in website personalization and how to drive data-driven culture?
Request a demo

Recommended for you

Power delivers effective product recommendations in different phases of the customer journey

Power International AS is the leading electronics retailer in the Nordics since 2015. Their strategy of relying heavily on the omnichannel business has always been successful with a considerable market position and a turnover of 1.1B Eur. 

For a retailer like Power with 20,000 different items to sell, it’s crucial to have personalized and relevant product recommendations for every step in the customer journey. 

Creating an optimal customer journey with so many products is not an easy job. Customer journeys are far from linear but rather a maze of repeated visits, a variety of consumed content, and other actions. The final conversion is seldom gotten during the first website visit. Every customer comes from different circumstances and has different ways to discover products and make purchase decisions.  

All these mean that the whole customer journey needs to be personalized, not just the page where we think the conversion will happen. For example, it doesn’t make sense to show a static front page for the visitor who has previously engaged with the website and is likely to complete the purchase. With this complexity, Power wanted to have the best personalization software in the market. 

The Frosmo personalization software brings product recommendations to a whole new level by using different data points about the customer to select the perfect AI-based recommendation strategies to support the customer journey in the best possible way. These data points include the customer’s location on the site, the phase in the customer journey, segment information based on historical behavior, and customer affinity. Taking the whole customer journey into account makes Frosmo the most versatile personalization software in the market.

The cooperation with Frosmo and Power started when Frosmo got a chance to prove our product recommendation capabilities against a leading competitor, increasing the overall conversion which is their most important KPI to follow.

To support the whole customer journey, we implemented Frosmo’s AI-driven product recommendations strategies for the following pages in their online store: frontpage, category page, product detail page, add to cart pop-up, and search page results.


Great results in conversion rates and average order value

With product recommendation strategies implemented on the pages, the results have been proven strong and effective. Purchase conversion rates have gone up by 170% compared to the group who didn’t see recommendations. The average order value got up by 41% compared to the group who didn’t see the recommendations.

“We’re very satisfied with the increase in conversions and average order values that Frosmo brings us with AI-driven product recommendations. The overall experience of implementing the recommendation strategies and the cooperation with Frosmo has been straightforward and was surprisingly deployed as a turnkey project. I would highly recommend using Frosmo against other competitive solutions.”

Jarkko Lehtismäki, CTO, Power International AS


Different recommendations strategies for friction-free customer journey orchestration

Personalized and AI-driven product recommendations aim to perfect the customer experience by helping the visitors go through the customer journey as friction-free as possible. Here are the strategies implemented for different Power sites.

  1. The first strategy is to advance the customer journey towards better product discovery and conversion on the front page. Recommendations show the most popular products at the moment, based on most converted and most-visited products by all users. The second set of recommendations shows recently viewed products by the customer.
    Power recommendations- popular products
    Power recommendations- popular products


    Power recommendations- recently viewed products
    Power recommendations- recently viewed products


  2. In the category pages, the aim is to find the most relevant product to purchase. Two strategies were implemented to make the product discoverability easier. The first set of category page recommendations shows the trending products based on highly-converted and most popular products. To enforce product relevancy and increase average order value, behavioral data is also used to show recommendations based on particular user interests in a specific category and combine it with products viewed or bought together.
    Power recommendations- trending products
    Power recommendations- trending products


  3. The product page is all about making sure the conversion happens and that the average order value increases. Traditionally, customers put a lot of effort to choose the product that best suits their needs. To make the choice easier, Frosmo creates recommendation strategies based on relevant products viewed or bought by others and uses historical data of the visitor to recommend other related products.

    Power recommendation strategies based on relevant related products viewed
    Power recommendation strategies based on relevant related products viewed


  4. Basket pop-ups are being developed to include physical store data for recommendations used to raise the level of relevancy and effective cross-selling.
    Power basket pop-up recommendations
    Power basket pop-up recommendations


“We’ve skimmed the surface of Frosmo and have seen a great potential for our business in the near future. Some of the most interesting future aspirations include increasing AOV by tweaking the AI-recommendation strategies and gaining higher organic traffic by combining recommendations with content-crunching NLP”,  Jarkko Lehtismäki concludes.

Interested? Get started now!

Let’s get started

Recommended for you

Press release: Nurmijärvi aims for better website usability and targeted crisis communication

Finnish Municipality of Nurmijärvi takes part in a pilot program with the Prime Minister’s Office as Frosmo aims for better website usability and targeted crisis communication

Nurmijärvi participates in the pilot program where the aim is to enhance user experiences on their website. This is achieved by improving the accessibility of relevant information through recommending content according to the interest of the users. The pilot program will help Nurmijärvi to deliver easily accessible and personalized communication to all its citizens, for example, personalized content to different geographical locations.

Spring 2020 has proven the importance of trustworthy and timely communication. Cities and municipalities play an important role in making sure that their citizens have easy access to the information they need. Often, the challenge is to make sure that the right information is easily found by many different target groups that these organizations have.  

The pilot program involves the Prime Minister’s office, the Municipality of Nurmijärvi, and the technology company, Frosmo. 

It will test three different features to enhance the usability of the Nurmijärvi website. The first feature will include a popup with a link to the QA document that is located on the Prime Minister’s Office’s website. For example, if a visitor seeks information about how the COVID-19 pandemic has affected primary schools on the Nurmijärvi website, the visitor is recommended to visit the Prime Minister office’s website to read more about the topic. 

cris communicaiton communication popup


The second feature includes a crisis communication ribbon that is shown on all Nurmijärvi web pages immediately after it has deployed centrally. 

The third feature includes a personalized content block that delivers targeted and interesting content according to visitors’ interests. This AI-driven feature will recommend three categories to users: last viewed, others also viewed, and trending topics. 

These three features will bring better user experiences and better usability of the service by providing trustworthy and rapid crisis communication and relevant content based on the interest of the visitor. 


Personalized content block


“It is very important that all of our citizens find newsworthy and important information fast and in the format they can trust. Today, cities and municipalities also need to think of user experience and this pilot program will deliver features that bring the Nurmijärvi citizens and their needs to the center of the service. The services must also be easy to use and easily accessible”, says Nurmijärvi Communication Manager Teemu Siltanen.

“It’s a pleasure to be part of this project to develop web services that are more user-friendly and takes into account the different needs and interests of the users. It’s also important that all cities and municipalities have access to the latest web technologies that can be deployed fast and easy without big IT projects”, says Frosmo CEO Mikael Gummerus. 


Contact persons: 


Teemu Siltanen

Communication Manager

040 317 2020


Mikael Gummerus


050 323 4643


Nurmijärvi is  Finland’s largest rural municipality with 43 000 citizens. Prime Minister’s Office supports the Prime Minister and the Government in the planning of Government matters and related decision making.

Frosmo is a Finnish SaaS company with an AI-driven personalization software operating since 2008.  Frosmo helps companies that provide web services to serve their web visitors better with targeted content that is personalized according to the interests of the visitor. Frosmo is fully GDPR compliant.



Nurmijärvi mukaan pilottiin valtioneuvoston kanslian ja Frosmon kanssa – tavoitteena sujuvampi verkkopalvelukokemus ja kohdennettu kriisiviestintä

Nurmijärvi lähti mukaan pilottihankkeeseen, jossa on tavoitteena sujuvoittaa kuntalaisten verkkopalvelukokemusta parantamalla sisältöjen löydettävyyttä käyttäjän omien kiinnostuskohteiden mukaan. Hanke parantaa kunnan mahdollisuutta ajantasaiseen ja räätälöityyn kriisiviestintään esimerkiksi kunnan eri maantieteellisille alueille.

Kevät 2020 on osoittanut ajantasaisen ja luotettavan tiedon jakamisen tärkeyden. Tiedon jakamisessa kunnilla ja kaupungeilla on erityisen suuri rooli. Usein haasteena on sisältöjen ajantasaisuus, löydettävyys sekä kohdistaminen oikealle kohderyhmälle. 

Pilottihankkeessa ovat mukana valtioneuvoston kanslia, Nurmijärven kunta sekä teknologiatoimittaja Frosmo. 

Pilottihankkeessa testataan kolmea Nurmijärven kunnan verkkosivustojen käytettävyyttä parantavaa toiminnallisuutta. Ensimmäisessä kokonaisuudessa Nurmijärven kunnan verkkosivuille tuodaan ponnahdusikkuna usein kysyttyihin kysymyksiin koronaviruksesta valtioneuvoston verkkosivustolla. Linkki ohjaa suoraan oikeaan paikkaan sivustolla. Esimerkiksi jos käyttäjä etsii tietoa koronan vaikutuksista peruskouluihin Nurmijärven sivuilla, hänelle tarjotaan mahdollisuutta siirtyä peruskouluja koskevaan kysymysosioonä.  

Toisessa kokonaisuudessa Nurmijärven kunnalle tarjotaan mahdollisuus tehdä vaaratiedote, joka näkyy nauhana sivun ylälaidassa kaikissa kunnan palvelun piirissä olevissa verkkosivustoissa välittömästi julkaisun jälkeen. 

Kolmanneksi Nurmijärven kunnan sivuille tuodaan koneoppimista hyödyntävä niin kutsuttu sisältöhissi, joka auttaa käyttäjää löytämään itselleen kiinnostavat ja ajankohtaiset sisällöt suosittelemalla kolmea kategoriaa:  viimeksi katsotut, tätä katsoneet katsoivat myös sekä nousussa oleva sisältö. 

Näillä toimenpiteillä tuodaan Nurmijärven kuntalaisille parempi verkkopalvelukokemus, jossa tärkeät kriisiviestit  ja kunnan yleinen verkkosisällön löydettävyys helpottuvat.  

“Meille on tärkeää, että kuntalaiset saavat ajankohtaisen tiedon varmasti, luotettavasti ja nopeasti. Tämä korostuu kriisitilanteissa. Tällä kotisivujen kehittämisellä varaudumme, mikäli koronatilanne huononee jälleen äkisti”, Nurmijärven kunnan viestintäpäällikkö Teemu Siltanen sanoo. ”Nykymaailmassa myös kuntien tulee olla asiakaslähtöisiä ja pilottihankkeessa tuodut ratkaisut tuovat kuntalaisten tarpeet, elämäntilanteet sekä mielenkiinnon kohteet keskiöön. Myös helppokäyttöisyyteen ja löydettävyyteen on kiinnitetty paljon huomiota”, Siltanen näkee.

“On ilo saada olla mukana kehittämästä kuntien verkkopalveluiden asiakas- ja tarvelähtöisyyttä. On myös tärkeää että kunnat pääsevät mukaan viimeisempien verkkopalveluiden kehitysmenetelmien piiriin helposti ilman isoa ja aikaavievää IT-projektia”, Frosmon CEO Mikael Gummerus iloitsee.




Teemu Siltanen

Nurmijärven kunnan viestintäpäällikkö

040 317 2020


Mikael Gummerus


050 323 4643


Nurmijärvi on yli 43 000 asukkaan kunta Uudellamaalla.

Valtioneuvoston kanslia on ministeriö, joka vastaa pääministerin johdolla hallitusohjelman toimeenpanon valvonnasta ja avustaa pääministeriä valtioneuvoston johtamisessa.

Frosmo on vuodesta  2008 toiminut, suomalainen pilvipalvelupohjainen ja tekoälyä hyödyntävä ohjelmistotalo. Frosmo auttaa verkkopalveluita tarjoavia organisaatioita palvelemaan verkkovierailijoita paremmin kiinnostuksen mukaan kohdennetulla, dynaamisella sisällöllä.  Frosmo täyttää täysin yleisen tietosuoja-asetuksen (GDPR) ehdot. 

Recommended for you

Add value with Frosmo Email recommendations

The customer journey doesn’t end when the customer leaves the website. You must continue delivering value after the visit and even after the purchase. Effective email recommendations complete customer journeys by connecting the last touchpoint, be it a Google search or a paid ad, to actual conversions. Relevant recommendations included in email marketing campaigns can increase customer engagement and encourage visitors to revisit a website again and to purchase products or services.

Frosmo Email recommendations allow marketers to extend the use of the Frosmo Platform beyond their websites. You can segment your visitors according to their interests, browsing behavior, recent purchases, visit source, and more. Getting recommendations for popular products, visitors are likely to convert and existing customers are more likely to purchase again. Effective email recommendations can significantly boost average order value and increase click-through-rate (CTR) by up to 300%.

How do Frosmo email recommendations work?

In the Frosmo Platform, an email recommendation is a piece of dynamically generated content delivered through an email which can display two to three products. The product displayed is technically a static image created by the Frosmo Platform and displayed in an email. The image features the product image as well as product information, such as name and price. When the recipient opens the message and clicks on one of the images, the link takes them to the corresponding product page on the retail site, where they can see more details about the product.

Add value with Frosmo Email recommendations

Frosmo Email recommendations can be easily implemented and require zero coding skills when there are recommendations already configured for the site. You can include email recommendations to any type of email you send to your customers. Here are some common examples:

  • Welcome email after a visitor has registered as a user
  • Newsletters or other regular emails
  • Order confirmation emails

Frosmo Email recommendations include built-in tracking that allows you to check the most important KPIs at a glance in the Frosmo Control Panel. Watch this video to learn more about how you can extend the use of the Frosmo Platform beyond your website, into the marketing emails you use to communicate with your customers.

Request a demo

Recommended for you

Delivering product recommendations that customers can trust

Helsinki is located on a peninsula and surrounded by the cold waters of the Baltic Sea. So half of the year means freezing wind and temperatures fluctuating between +5C and -5C. And now that winter is fast approaching, I have to make sure that I have the proper outdoor wear. 

This year, I was on my search for a new winter parka jacket as the current one has finally reached the end of its lifecycle.

I went through many options and utilized highly relevant product recommendations related to similar products by many retailers. In the process, I also came across a few irrelevant recommendation strategies.

The first interesting one was on a mobile page of one retailer, where they clearly had an issue regarding what level of product information is used for recommendations. The image shows the example of a green winter park size 38. The recommendation engine had decided that the top three recommendations for me would be:

  1.  A winter parka of the same brand, but three sizes bigger, the product itself was relevant, size not so much
  2. The exact same product, but two sizes smaller
  3. The exact same product, but five sizes bigger

product recommendations

What a shame. Mobile is a challenging user interface, as the screen is so small and often, the visitor browses quickly through pages so it would be extremely important to make the most of the real estate available.

After going through price and availability options, I decided where to buy the parka jacket. I would have been the perfect candidate for add-on products, such as gloves or scarves, in the last stage of my purchase journey, but guess what they recommended for me?

  1. Exactly the same jacket, but a different color
  2. Another jacket, less than half of the price

product recommendations

At Frosmo, we advise our customers not to make their visitors get confused and doubt their purchases during the whole customer journey. Instead, we consider offering complementary products all the way through the funnel as a form of good customer service. 

As I now wait for my order, I’m having second thoughts as to whether the green parka is better than the black one.

How to do smart and effective product recommendations

In the example mentioned, the company lost potential add-on from a customer who would have been willing to buy additional products. The recommendations failed to create a good customer experience that would help the customer trust her purchase and to find products that would nicely complement the jacket that’s already in the shopping basket. 

It’s crucial to pick the right kind of product recommendations to significantly increase average order values and effectively boost conversions. At the same time, smart and effective recommendations build and instill confidence in customers and ensure them that they made the right purchase. By showing the right content at the right time and in the right place, customers will also have unique customer experience tailored to their interests during the buying funnel. 

Read how international retailers Clas Ohlson and Reima use smart recommendations. 

Download the comprehensive guide to recommendations.

Comprehensive guide to recommendations

About the writer:

Maija Erkheikki has over 15 years of experience in the international software industry and is driven by a passion to use technology to improve the quality of life. This passion was best exemplified in her role as a consultant for healthcare companies, working on business processes and performance management. This led to a career in the partner-driven software business, where it is critical to build win-win-win models. Currently, she is fascinated by the digitalization of commerce and how artificial intelligence reshapes all industries.

Recommended for you

4 revenue-driving personalised customer experiences for iGaming

Brands need to stand out to be successful. Costs of acquisition are rising due to the ever-increasing competition. These factors, combined with the shifts in customer behaviour and the ease to move onto the next service, make providing a customer experience that stands out of paramount importance.

Your customers are spending the majority of their online time on personalisation supremos such as Amazon and Netflix. These powerhouses talk to their customers on a 1:1 level, which in turn sets your customers’ expectations at a much higher level and means that they now expect you to know their intentions and to serve highly relevant content each and every time they use your service. This bar raise in providing personalised customer experiences results in a squeeze on brands that are unable to compete.

The Frosmo SaaS platform provides igaming companies with complete freedom to offer any type of personalised customer experience, regardless of their existing technology stack and resource. The following four ideas are very easy to implement with Frosmo – and these brand-building ideas will have an instant impact on revenue.

 1. AI-powered recommendations:

Ensure that you’re harnessing all of the data available on every touchpoint of your service. With this data, you can leverage artificial intelligence to start predicting what slot/game/market your customer would like to see next. This combination of learning continuously from a customer’s product interest, their profiles, and the context of their visit, means you can offer highly targeted and relevant content.  These recommendations could reorder your casino lobby by creating an entirely new look and feel, or vary which sports markets are displayed to each individual user. Doing this provides a relevant and engaging experience that will increase retention, and as Accenture has stated, 91% of consumers will remember an ecommerce brand and be more likely to return when they provide relevant offers and recommendations.  

 2. Call out your brand’s USPs & promotions 

Boosting participation in your cleverly thought up promotions or rewards program is always challenging. The effectiveness of direct marketing is being eroded by increasingly noisy market places, compliance, and shifts in consumer behaviour. Using all of the web real estates at your disposal combined with the data available means you can start to offer and highlight to players relevant promotions and rewards, at the right moment in their journey. The most successful brands use unique reward programs to encourage repeat visits and loyalty. Effective personalisation includes maximising the value that these reward schemes drive by increasing customer participation with progress bars, reward callouts and engaging content prompts.  Ensure that you maximise the valuable time that you have with a customer when they’re already at your service. This will increase participation, engagement, and therefore brand loyalty.

3. Effectively cross-sell casino and sports products  

Customers expect a connected and consistent experience between various products and game lobbies. We have seen operators convert more players across products when they start to include recommended content at the right moment in a customer user journey. With personalised recommendations, an operator has experienced a 20% increase in CTR when reviewing against a control group that sees non-personalised content. This operator displays casino games that the player is most likely interested in, this content is shown in the most relevant area of the sports lobby, and when the visitor is most likely to want to engage. This approach provides a connected feel between your two main products, clearly increasing engagement. 

4. Make the front page relevant 

Regardless of whether you’re welcoming a customer to your brand for the first or 10th time, it’s vital that you show relevant and engaging content on the front page. One example includes ushering your roulette playing live casino customer back to your service and directly loading the live roulette game lobby. Another one is ensuring a Wolves football fan who bets routinely on his beloved Wolves is served a pop-up modal acknowledging this type of behaviour and welcomes him with Wolves-themed content and a relevant bet option.  This ability to show relevant content on the first page of your service can have a 4x impact on conversion. 

One marketing trend highlighted in igaming is one-to-one personalization. Read how 188BET is leading the way by being customer-centric.

Wrapping up

With markets becoming more competitive and various outside factors impacting the bottom line on your marketing spend, the rules to succeed are continuously evolving. With nearly all operators offering the same game content and payment options, the true winners will be the companies that realise being at the forefront in customer experience is the key to gaining a competitive edge. Surviving and winning in this customer focused market will depend on providing personalised customer experiences at every touchpoint of customer engagement.  If you would like to talk about how to increase your customer engagement and directly impact your revenues, contact me at

__________________________________________________________________________________________________ About the writer: Matt Buff is a digital customer engagement expert with specialised skills in sales and marketing. Over the last 10 years, Matt has used cutting edge technologies to maximise customer acquisition, retention, and revenues. He enjoys learning about brands and sharing ideas to improve user experience and CRO, and passionate about enabling teams to achieve their individual customer experience goals.

Recommended for you

Which product recommendations to use on different types of pages

Product recommendations used to mainly consist of highlighting the most popular products. But with the rise of AI and machine learning, recommendations in ecommerce have increased customer engagement by several steps.

Recommendation engines now start with understanding the customer’s intent by capturing the customer’s shopping patterns, behavior, and purchase history. With the data on hand, you can then find the best products to present to each customer accordingly. Other factors also come into play such as new trends in the market, bestsellers, discounted items, etc.

It’s crucial to pick the right kind of product recommendations to effectively boost conversions. By showing the right content at the right time and in the right place, customers will have unique experiences tailored to their interests across the buying process. Effective recommendations can significantly increase average order values and ultimately boost your profits.

Here are some of the product recommendation types used on different pages at different customer points. There are many variations depending on the customer behavior data and in most cases, it’s the combination of all the strategies that produce the best results.

1. Homepage

The homepage is to a website as a front-window display is to a physical store. Therefore showing the main product offerings is crucial as it enables users to start shopping and increase click-through rates. Use this space to insert personalized recommendations, what differentiates you from the competition and personalized messages of what you offer.

For first-time visitors, they sometimes don’t have anything specific in mind. Because it’s their first time, you know very little about them. The recommendations on the main page aim to inform these newcomers about your products and get them engaged, so recommend:

  • Hottest items or what stands out from the competition, items that have ranked best on your site in different categories based on popularity, purchases, CTR, views, and trends
  • Bestsellers, or those with the highest conversion rates
  • Newly added items
  • Emerging or trending items, or items that will be available soon and can be reserved early
  • Context-based recommendations such as location, weather, and device
  • Items on sale

For returning visitors, since you already have data about their recent purchases and browsing history, time to utilize personalized product recommendations. For example, if the customer previously bought a camera, you can show new arrivals of camera accessories. Try recommending:

  • Items that the visitor has previously viewed or purchased
  • Items related to the recent purchase
  • Discounts on recently viewed or purchased items
  • Items related to those abandoned in the basket
  • Popular or new items based on previously viewed or purchased items
product recommendations
Chopard uses a powerful image of a new and hot item for the homepage creating a visual and emotional connection with the readers.

2. Category page

The category page drives discovery to the online page, hence product categories need to be clear and quickly differentiated. Relevant products are stored in the category pages giving the shopper the chance to narrow down to a specific subcategory to find exactly what they’re looking for. Navigation should be effortless with proper and easy-to-understand labels.

For first-time visitors, get them to browse your category pages by recommending:

  • Bestsellers
  • Last week’s or last month’s popular and trending items
  • Newly added products
  • Category variety or subcategories
  • Items that are high in stock

Offering products to returning visitor needs fine-tuning. You can use the gathered data to trigger personalized offers in the category page. The offers can include seasonal, daily deals, rating-based recommendations, and others such as:

  • Recently viewed or purchased items
  • Popular items from the user’s favorite subcategories
  • Items related to recent purchases
  • Deals and promotions in the category
  • Abandoned items in the basket
product recommendations
Ikea’s category page is outlined clearly, with images for every subcategory, a highlighted item, and some recommended products as well.

3. Product page

The product information page is where visitors find detailed information about the product. The goal is to keep the visitors in the sales funnel and for them to add the items to the cart. Recommendations on product pages aim to increase average order values. Often, visitors have one thing in mind when shopping but end up buying more, so you’ve got to take advantage of this and stretch their buying eagerness by recommending relevant items and launching upselling efforts.

Customers usually spend the most time on product pages, browsing, and choosing the products they want to purchase, so take this opportunity to offer:

  • Complementary items from the same category
  • Items to upsell, for example, those that are better in quality or have higher prices
  • Alternative relevant products
  • Associated items, for example, batteries for an item that needs them
  • Most purchased or viewed items from the same category
  • Other items based on recent purchases or views
  • Promotional items
product recommendations
Jewelry maker Pandora provides comprehensive information about the product, reviews, and does a great job at suggesting related products for its customers.

4. Shopping basket page

When customers are in the last stage of the shopping funnel, it’s important for them to complete the purchase. Get them to do it without so much distraction, but grab this chance as well to get higher average order values. When they are about to finish the purchase, customers are usually willing to add more items, or even pay for a slightly higher priced product, if the recommendations are done effectively. The checkout page is an excellent place to recommend because you already have their preferences. Recommend:

  • Items often purchased together with the item in the shopping basket, such as add-ons and accessories
  • Items related to the purchase that would be a perfect add-on to get free shipping
  • Popular items from the recently visited categories
  • Items that are not commonly purchased with the product in their baskets, but have been previously bought together, say a dress with colors that match a bike
  • Items based on the visitor’s affinity and preferences
  • Items that have relatively low prices and are easy to buy
product recommendations
Amazon’s cart page provides recommendations with prices, reviews, ratings, and the opportunity to compare the item on the cart with similar items.


Ecommerce stores are now expected to deliver more personalized shopping experiences. Product recommendations require smart execution to produce the most effective results. Consumers are more likely to spend money and return to a site that makes them feel valued with personalized offers that are tailored to their needs. It’s all about the perfect timing – recommending items that your customers want at the time when they’re most engaged can enhance user experiences hence improving conversion rates.

Read on to find out how recommendations nowadays learn about customers in real-time and personalize all pages they visit. Start delivering different recommendation types based on how they behave and where they are on the customer journey.







About the writer:

Katja Kokko is passionate about strategically developing digital businesses and creating memorable digital customer experiences on a tactical level— what she gets to do as Customer Success Director with her clients and her team. She is a growth hacker by heart and with her exceptional interpersonal and leadership skills, she thrives in challenging business cases as these challenge her to think differently and out of the box.


Recommended for you

4 most powerful phenomena in ecommerce marketing that every CMO should start adopting now

The first 3 phenomena personalized customer journeys, AI, and experimentation have been around for a while now. It’s good to see that after years of talking about them, we’re finally starting to see great results from brands that are embracing AI and personalization tools and using them as part of their core business models.  

An interesting new trend that marketers should start looking into is Progressive Web Applications (PWAs), as they bring many advantages to marketers who are looking for better reach and coherent omnichannel customer experience.

It all comes down to the fact that brands have now realized that putting the customer first and embracing their differences by creating personalized customer journeys is the only way to keep up with the competition and grow.

1. Personalized customer journeys

It’s very difficult to create superior customer journeys if you still think of segments as large groups based on demographics, location or some other loose definitions. Every buyer is different and you need to be able to delight every individual. Designing and implementing online customer journeys improves customer satisfaction, lowers customer acquisition costs, and makes customers loyal.

Janne Timmerbacka, ICT director of online grocery store Kauppahalli, sums it up nicely:” The best customer service isn’t based on profiling customer sets any longer. Every single customer is different with their needs and preferences. In an online grocery store, great customer experience is made from knowing the customer so well that you are able to delight the customer by recommending products of their interest”.

To get to that individual level of highly personalized customer journeys, real-time behavioral data of the visitors is needed. What they click, how deep they scroll, what they read, watch, and download will give you a lot of insights on how to direct them to find more relevant info for their individual needs.

2. AI

To go deeper with personalization, AI is needed.  Forbes just concluded a study that   AI-powered marketing campaigns increase customer engagement by 7x and revenue  3x by helping marketers become more targeted, more relevant in their content, and more effective in how they engage with each individual customer.

So with that in mind,  an AI-powered recommendation engine is definitely something that every ecommerce business should invest in this year. Think of it as a great way of serving your customers better, as they can find new interesting products or content to consume. In an article, Netflix said that its AI-powered personalized recommendations save them $1 billion in revenue annually by avoiding canceled subscriptions.

Kristina Lagerroos, Development Manager for the global leading brand in functional kids’ wear Reima, says, “Frosmo enables us to create winning user experiences by providing a recommendations engine that brings value to our customers. They help us to understand our customers’ preferences better and make data-driven decisions”.  

Natural Language Processing (NLP) is also a very interesting phenomenon in that regard. NLP is a subfield of AI that focuses on enabling computers to understand and process human languages or at least get a tiny bit closer to a human-level understanding of language. It ’s cutting edge and a great way to show your visitors more relevant content.

For example, it can be used as the basis for an algorithm that recognizes words in the site content and makes recommendations based on similarities. If a visitor reads an article about a specific topic, the algorithm recognizes the relevant keywords and crawls other articles to find the same keywords. These articles are then recommended to the visitor. This method offers great user experiences also for those first time visitors whom we wouldn’t otherwise be able to serve according to their interests.

3. Experimentation for data-driven decisions

At times, the most challenging part of running an ecommerce business is to get people to work together and agree to website changes. Thus, validating customer journeys through experimentation is the key to gather data and make more insightful data-driven decisions that everybody can agree on.

Multi-armed bandit (MAB) is a powerful AI-powered experimentation tool. With MAB, you test different variations on your website and direct traffic towards those variations that are performing the best. Small changes might turn out to be big revenue generators. That’s the reason why testing is essential and why testing with AI offers a powerful advantage.

Clas Ohlson uses multi-armed bandit to optimize the most promising product recommendations in different markets and in doing so, enabled them to increase their average order value by 8% and the product rows by 27%.

4. Progressive Web Applications

Progressive web applications (PWAs) are web applications that load like regular websites. It allows end users to have a mobile app like experience when visiting the main website, including a home screen and push notifications. From a marketing point of view, one of the biggest advantages is that customers get the same user experience as on the main website without being tied to any app store. These are the reasons we’ll see many ecommerce sites turning into PWAs during this year.

Let’s start creating superior digital experiences with personalized customer journeys!

Request a demo


About the writer:

Tanja Säde is the Chief of Marketing at Frosmo. With 20 years of experience in technology marketing, she is always curious about the latest technologies and marketing trends. She is enthusiastic about creating superior digital customer experiences through data, personalization, and AI. Outside of work, she is a yoga addict.

Recommended for you

Clas Ohlson gains great results with validated personalized customer journeys with SAP Hybris and Frosmo

  • Continuous and fast front-end development
  • Smart product recommendations
  • AI-powered multi-armed bandit for data-driven decisions

Clas Ohlson creating superior customer experiences with Frosmo

Clas Ohlson celebrated their 100th anniversary in 2018, which makes them one of the oldest online stores in Sweden. When it comes to online retailing, Clas Ohlson considers itself a startup despite being one of the strongest retail brands in the Nordics and their long history with brick and mortar stores.

It shouldn’t come as a surprise then that they’re heavily invested in figuring out the ideal online customer journeys on their way to create flexible and inspiring shopping experiences.

Continuous and fast development

To meet customer expectations, continuous development of the front end is needed. Clas Ohlson uses SAP Hybris, which was tailored to meet their business requirements eight years ago. Back then, the online store was built to show what products were on the stock and the focus was to get people to buy from the store after browsing the products online.

The ecommerce platform was also custom-made so it was inflexible with too many code dependencies to other systems. The front-end development started to lag and it could no longer match the modern online retail experiences that customers expected.

Clas Ohlson wanted to partner up with a company who could help them with the continuous development of the front end and offer a vast variety of features to create optimized and personalized customer journeys.  

“Fast time-to-market is essential in a fast-paced ecommerce business. With Frosmo, all the changes to the front end can be implemented around 30% faster,” says Deneb Pettersson, Online Operations Manager, Clas Ohlson.

Smart product recommendations

When many of your products are similar, it’s essential to help customers pick the right product for their needs. Frosmo helps you enrich the product data so that when a customer tells a bit more about their requirements, the store can recommend the right product or selection of products.

Recommendations bring value to the customers. On the front page of Clas Ohlson’s webshop, the Frosmo Platform is used to show the most viewed and most bought products. When buying a product, there are many different ways of recommending other products to accompany the initial item of interest. For example, if you buy a robot lawn mower, your recommendations will show the right add-ons for your lawn mower such as cables and tools to complement the product; extras such as a storage unit, accessories; and finally, what others have bought with the lawn mower.

With add-ons, upselling, and recommendations, Clas Ohlson sees an 8% increase in average order value and 27% increase in bought articles.

Experimenting is key

Sometimes, the most challenging part of running an ecommerce business is to get people to work together and agree to the changes in the front end. Thus, experimenting is the key to gather data that shows what should be developed next. And because there’s evidence that backs up the decision, it’ll be a direction that everyone can agree on.

Clas Ohlson uses AI-powered multi-armed bandit to optimize the most promising product recommendations in different markets.

Small changes might turn out to be big revenue generators. That’s the reason why testing is essential. Clas Ohlson grew conversions by 13%, session value by 14%, and visit durations by 16%,  just by adding a sticky bar with the search to the navigation.

Focusing on services

In the future, Deneb sees the company moving towards offering more services. Customers buy products to get something done and Clas Ohlson wants their customers to achieve what they set out to do, whether that’s installing a TV on the wall or connecting the lamp to the ceiling. Renting equipment is also an upcoming trend where people no longer want to own all the tools and equipment themselves but would rather rent them from a trustworthy place.

Details matter

To create winning user experiences, it’s the details that matter. Having a solution that enables experimentation with such details will prove to be invaluable. It’s the key to continuous development and the perfect way to turn a good SAP Hybris site into a great one.

Want to know how Frosmo can help you with your SAP Hybris website?

Watch the 21-min recorded presentation where Deneb Pettersson from Clas Ohlson talks about how to create superior experiences with SAP Hybris and the Frosmo Platform.

Read more about how multi-armed bandit optimization makes testing faster and smarter with machine learning.

Request a demo or contact sales directly.

Recommended for you

Reima is creating superior customer experiences with SAP Hybris and Frosmo

  • Increased conversions with product recommendations
  • Customized user experience
  • A/B testing for data-driven decisions

Reima is the global leading brand in functional kids’ wear. Its mission is to encourage children to discover the joy of movement that stays true to its business idea of providing a good quality, year-around, tip-to-toe wardrobe for active children, ages 0 to 12. Their products also include the Finnish baby box, a starter pack to parenthood that holds essentials from hygiene products to indoor and outdoor clothes for the baby’s first year.

Digitization is in the core of Reima’s strategy. The net sales via their own digital channels accounted for about 13% of total sales in 2017. Digital is not only about sales, but it also touches the whole value chain and allows Reima to build new business models to serve their customers even better. One example of digital innovation is ReimaGo, an activity tracker that adds a new dimension of joy to kids’ activities and lets parents see how much their kids are moving.

Today, 80% of net sales come from international markets, from over 70 countries across three continents.

Reima ecommerce in short:

  • 8 different stores, 7 languages.
  • Shipping to approximately 40 countries.
  • An international team, but operates mostly in Helsinki.
  • Delivery of 285 000 items annually with 110 000 transactions.
  • Different marketplace platforms that have been integrated to Reima’s systems.

Reima thinks of ecommerce holistically. The core is an omnichannel experience where the customer service has to exceed the customer expectations and the design drivers easy, intuitive, relevant, and fun need to be recognized throughout the digital journey.

We have to be relevant and need-based”, says Kristina Lagerroos, Development Manager at Reima. “Weather is a big sales driver. Snowfall always takes parents by surprise and they hurry to update their kids’ winter gear at the same time”, she adds.

Reima wanted an independent and secure ecommerce platform that offers a vast variety of features and capabilities so they chose SAP Hybris in 2011. Since then, customer expectations have changed drastically and the need for more personalized buyer journeys and product recommendations have become a core of superior digital user experiences. That is where Frosmo comes in.

Better conversion with product recommendations and delivery calculator

With the Frosmo Platform, Reima is boosting sales by providing their customers with more personalized product recommendations to find the most suitable items each customer’s individual needs. The recommendations include the following elements:

  • Product recommendations
    • Most bought
    • Most viewed
  • Highlighted product elements
  • ReimaGO recommendations in the checkout
  • Tags on category pages

The delivery calculator has helped Reima boost sales at check-out by reminding customers how much more they need to buy to take advantage of free shipping.  

Data-driven decisions

With customers’ expectations increasing, Reima wants them to experience positive surprises when shopping online. For example, Frosmo is helping Reima to explain the uniqueness of the clothes by highlighting hero products and bringing extra stories to support faster decision-making.

This involves continuous smaller improvements to the site and thus A/B testing has increased its importance. Continuous analysis of the test results running on Reima’s sites has been crucial to making data-driven decisions. Frosmo plays an important role as it offers the testing capabilities and allows fast time-to-market for visitors to see continuous new elements on the site.

Automating traditional A/B tests and switching to multi-armed bandit for optimizing tests will be the next milestone.

“Frosmo enables us to create winning user experiences by providing a recommendations engine that brings value to our customers. They help us to understand our customers’ preferences better and make data-driven decisions”,  Kristina Lagerroos concludes.

What does the future hold for ecommerce?

Reima wants to be the forerunner of creating superior customer experiences and to continue offering features and services that people would not expect from them.  These would include functionalities to help customers choose the best possible product for their particular needs.

Openness is one big trend in the future. Open and honest communication during the product life-cycle is essential. Gamification will play an important role in future product development where Reima wants to involve customers and build communities around its products and brand.

Request a demo

Recommended for you