Our customer meetings in 2025 are almost always about artificial intelligence in some way. Some people still wonder if AI makes a difference for Norwegian online stores, or whether it's just hype.
After following the development as an advisor, I find AI gradually becoming a concrete tool that several online stores actually use. At the same time, it's not like AI just does everything completely without it. Fortunately, for us people.
Let's look at some concrete examples, results, pitfalls and what we should think about, as we go to Black Friday and Christmas in 2025.
Norwegian examples of AI in customer service and product search
Development is taking place rapidly and it is fun to see that Norwegian enterprises are involved in the change.
In Norway we now see several concrete AI bets in online commerce. One example is Mimir, who has built an AI-based customer service for online shops. Mimir connects to the online store and can answer customers automatically 24/7. You as a shopkeeper go in and improve continuously, and get the agent to learn from their own faults.
The bike shop Bikeshop used Mimir in a pilot project, where the AI chatbot took over 90% of the enquiries one month. For an online store, it could mean that 9 out of 10 questions are resolved automatically, which frees time for your people and ensures that customers are helped even outside normal trading hours. Money saved, satisfied customers.
AI is not used only for customer service. Known online stores also start to take artificial intelligence into both product search and personalisation.
Maxbo recently upgraded his search function with artificial intelligence (Google Vertex AI Search). The aim was to help customers find the right product among over 20,000 goods.
After launch, Maxbo saw, among other things, a 40% increase in average order value, 15% more goods per order and 70% higher click rate on the search. At the same time, the number of searches without matching decreased by 30%, and customers found it easy to do less.
If the numbers are correct, it could mean that AI search and product recommendations can give customers a better trading experience. For the stores, it means more money in cash and more money again for your marketing crowns.
Norwegian suppliers like Digitroll have started integrating AI into their online store solutions. Among other things, we see AI-based product search, personalised recommendations (e.g. with Hello Retail) and automated content to make the trading experience more tailored for each customer.
Effects on conversion, customer service and competitiveness
The above examples indicate that AI can actually raise important key numbers in the online store. Maxbo's higher order value and customers who put more products in the basket are a concrete proof of increased sales per customer
Personal recommendations and smarter searches make it easier for customers to detect relevant accessories and additional goods. This was often difficult with traditional filtering solutions in all online store resolutions.
AI can understand the intention behind the customer's search and web behavior better, and therefore suggests products that the customer actually needs more than usual solutions. This improves the use of traffic already in place and increases the conversion rate from search to purchase.
One important assumption is that good results require good groundwork: good product data, proper implementation and time to get the machine learning model real user data to chew on. Are we good enough on this in Norway?
At customer service, the gain is mainly related to accessibility and efficiency. Chatbot solutions like Mimir can be in the entire day without any corresponding increase in compensation of employees. Staff can focus on the most complex issues and proactive customer service. More rapid answers often give more satisfied customers and increase conversion
But people have not become redundant either. A KI agent must be trained and monitored and continuously improved. It still answers wrong, it misunderstands questions, and it depends on a good basis. Those who have tested this say that human touch is still critical.
Real FOMO
There is little doubt that AI has become a factor in competition between online stores in record time. It has something to say: Those who use AI tools quickly get a head start. Those who don't do it can miss something.
Google launched Google Automated Discounts (GAD) this year, which can automatically provide discounts in Google Shopping ads. Lyko used this to hijack several sales, while Bedakker chose to stand outside. Lyko's products were given more favourable prices in Google, which gave them a competitive advantage. Binders say they missed up to $300 million in turnover.
It is part of the story that this practice was stopped for a period because the Consumer Protection Authority believed it was contrary to Norwegian legislation, but the example shows how AI can change its competitive position in a short time. Whatever industry, if any players succeed with AI, it forces competitors to assess the same thing in order not to fall behind.
There are exciting times when there are many opportunities for future online shops! But not quite without risk.
Fall mines
Although the possibilities are great, AI also has its pitfalls.
The case of Google Automated Discounts shows how fast AI can cause problems. The service automatically dropped prices in ads, often lower than in the online store itself. For customers, this could be seen as price discrimination, and the Consumer Protection Authority considered the scheme to be in breach of the pricing legislation. Google changed the solution later, but the Belander case reminds us that AI tools must always be adapted to Norwegian rules. Otherwise, both fines and customer trust are at risk.
Personally I also wonder at Lyko's margins--but it must be up to them to worry about.
The Google case may be a single case, but AI can create other challenges as well. Recommendation systems may propose sold out or incorrect products if the data base is weak. Chatbots may misunderstand questions, or in worst case find answers that damage trust. We have also seen that advertising tools can use budget for wrong types of sales, like cheap goods with discounts, instead of the most profitable products.
Just because of such examples, AI must always be monitored and controlled by humans. Without clear frameworks and follow-up, gains can quickly disappear.
Search is modified
AI also affects how customers find online stores. Google still dominates Norway by around 90 percent of the search market, while Bing has only a few percent. Chatgpt has grown rapidly globally with over 100 million users, but is still far smaller than Google.
Chatgpt handles about a billion questions daily, while Google is over 13 billion. This means Chatgpt has taken some attention, but Google remains the clear key entry point for customers to discover and evaluate products. And it's going to be exciting to check these numbers in 1-2 years.
What happens now, however, is that the market leader Google needs to change how search results are displayed. They've launched an AI function that gives a complete answer on the top of the page, instead of just links. For users, this is fast and practical, but for online shops it may be critical. Tests from the United States and the UK show that the first place click rate can fall by over 30 percent when the AI responses appear and some websites have seen traffic halving. Many fear that if people get used to getting the answer directly at Google, fewer will click on to the shops.
Bing tried to challenge Google early by building Chatgpt into his search engine in 2023. This did not produce huge growth in users, and the market share in Norway remains at just a few percent. Yet, this pressure on Google to develop its own AI faster. Bing Chat can already respond with recommendations and product links directly in the conversation, making visibility in AI responses to a new competitive area.
So online stores need to think not only of SEO, but also of creating content that AI tools pull in and show on. That means that guides, tests and thorough descriptions are becoming even more important to be mentioned in these answers. And we can expect prices to increase, when it might be less search and click to pick up.
In brief, Google remains the most recent traffic driver for online shops, but the way it operates is changing. AI in search may mean fewer visits, but the visitors who are coming may be even more qualified because they selected to click on after getting a good AI response. This sets new requirements for online shops in the form of SEO strategy and how to measure the value of traffic in the future.
Black Friday and Christmas Trade Council 2025
We are approaching the most important sales periods of the year. Here are some tips – based on what we now know about AI – for what online stores should look forward to Black Friday and Christmas 2025.
My first advice is to bring the seminars from us in INEVO, we want all of you to be best rigged into this season, and we're happy to share our experiences. We work closely at Norwegian online stores, and so we always have to be at the ball in terms of, inter alia, AI.
Compliance with the law and price strategy. Make sure that your AI tools don't break Norwegian rules, it can quickly get expensive, and sometimes eat up the entire profits for the year. Avoid situations that Google Automated Discounts created, double check that all discounts KI provides follow the rules for price labelling and do not damage the trust of customers. If unsure, please insert minimum margins and accept any price changes manually.
Train AI on current information for your online store. If you are going to use a chatbot or AI assistant in the high season, it must be up to date on promotions, stock status, delivery times, terms, etc. Nothing is worse than a smart chatbot that provides out-of-date answers about your Black Friday offer or delivery times. Please take an extra round internally so that your AI assistant can answer the best possible questions you think come from customers.
Stress test your solution. Black Friday means too much traffic and a lot of customers at once. If you have implemented AI-based search, recommendations or chat, test the scalability already. Does your search engine take ten times normal traffic when everyone is going to find offers at the same time? What if the chatbot gets a thousand questions in one night? For that matter, can you deliver what you sell in a shorter time? Check that AI solutions are robust when pressure comes, or that they are in the worst case in an elegant way, e.g. “I’m sorry! We have a big boost, you're switching to support...".
Measure the effect - not just assume. Do not allow AI to run on autopilot during the entire sales period without monitoring. Configure the points of measurement for conversion and sale where AI is involved. For example, if you activate a new AI recommendation engine before Christmas, please follow if actual sales increase for the products recommended. Make A/B testing where possible to see if it has a positive effect. Adjust or disable if the numbers indicate that AI does not deliver as you wanted.
Remember human contact where applicable. In Christmasstria, customers can be extra stressed and expectant. An AI that responds lynracically gives value to your customers, but don't hide the possibility of talking to a real man when needed. Make sure that customers can reach customer support in a traditional way if the AI assistant does not understand them or they have a special case. AI can handle the volume, but human empathy can be crucial in some situations, especially when gifts do not reach out, or something goes wrong with the order just before Christmas. There's a lot of feelings in games then, take your customers seriously.
Conclusion
Until 2023, it was mostly about interest and hype around artificial intelligence, but in 2024 both small and large online stores have started using AI solutions that actually work. It can be anything from a chatbot that takes away customer service, a search that gives higher sales or a smarter (and perhaps sometimes a little too smart) marketing technique like GAD. Yet AI is not necessarily a magic shortcut.
My claim is that those who succeed do so because they use technology strategically and follow up with human governance.
My impression now is that AI is not a gimmick and neither is a guarantee. It's a tool. And like all tools: some build works of art, others knock on their finger. The 2025 will certainly show us both.
Sources
- StatCounter (August 2025) figures – market shares for search engines in Norway: gs.statcounter.com
- Knowit/Maxbo case study (2023/24) – AI-based product search increases order value and click rate: knowwit.no
- Web commerce.no (11. October 2024) – Bikeshop automates 90 % of customer service with AI chatbot: online commerce.no
- Shifter (22. July 2025) – Mimir AI agent put into service at 15 Norwegian online stores (customer service): shift.no
- INEVO Blog (9. March 2025) – Debate on Google Automated Discounts and Norwegian Law: inevo.noinevo.no
- Webvise/E24 (10. July 2025, Stay with us. Lyko in the GAD case (lost sales and legality assessment): ntavien.no
- Digitroll (August 2025) – Impact of Google AI search (CTR drop and traffic drop): digitroll.no