Good afternoon, and welcome everyone to this webcast with a presentation of the Q1 2026 results from Columbus that were published this morning. In the first part of this presentation, you'll be in a listen-only mode, which will be followed by a Q and A session. During the Q and A session, if you have dialed in by phone, you'll be able to ask questions verbally by pressing pound or hashtag, followed by five on your telephone keypad. If you are watching this webcast online through a browser, you can ask your written questions in the chat below, and I will read them aloud after the dial-in. I'll do my best to translate questions from Danish to English. With that, I'll hand over to Columbus CEO, Søren Krogh Knudsen, and CFO, Brian Iversen. Please go ahead.
Thank you very much, Philip. Thank you. I will just switch for a brief moment here to the disclaimer for you to look through. As Philip was saying, I am joined by Brian here to assist me through this presentation today. I will start with some financial highlights. I'll follow up with a few, I would say, market and operational highlights, and a little glimpse into what I consider to be the next steps and the current steps in Columbus's AI journey. Then Brian will follow with more financial details or details, a detailed breakdown of the financials. We will then reiterate guidance, we'll conclude with Q and A. Financial highlights.
The revenue in Q1 declined by 4%. We still see some market conditions that are challenging, and I would say that goes especially for our Danish and Swedish operating companies. We have seen a very strong recovery in Norway, and we're starting to see some really significant signs of a recovery also in U.K. and U.S. It's also fair to say that they haven't been as hard hit, so there's not that much they need to regain in terms of the speed. Our EBITDA declined by 43% in Q1, reflecting mainly the impact of the weaker revenue. That leads, obviously, or that stems from a lower operational efficiency in Q1 2026 compared to what was a fairly strong Q1 in 2025.
Our efficiency, which is one of our main KPIs, if not the main KPI, ended at 62% for the quarter of Q1 2026. That is actually the same level as in 2025 for the same quarter. However, this is a very important thing to notice, we saw a gradual improvement of this efficiency number over the quarter, which means that January started lower, and February improved, and March improved again, which is more of a leading indicator of performance. We don't report these on a monthly basis, but I can say that the spread of the efficiency performance of these three months was 5 percentage points.
And that is a relatively fast development that we normally don't see as fast across that many countries and consultants. It developed for the better. Also, we continue to invest and develop our Data & AI business line. They did deliver a relatively strong growth of 23% in Q1 compared to the same quarter last year. They partly driven by very well-executed Agent in a Day events that we've been running in Denmark, Sweden, and Norway. It's essentially an event where we invite our customers for a full-day training event on how to set up autonomous AI agents and hand off workload to these agents, and they've been very successful.
Yes. We go to the market and operational highlights. Again, I will reiterate that some market caution persists. It's slightly different across markets, as I said in the beginning, we see mainly Denmark and Sweden as being the more sluggish markets. We see a reasonable market condition in both U.S., Germany, U.K., and also in Norway. In these numbers, in the efficiency numbers that you see, we also cater for, I would say, internal investments in retraining our workforce. There's a lot of new hires going into it. There's also a new tool set that we implement to stay at the forefront of consulting and to be able to be efficient in our deliveries.
The pipeline that we discussed earlier still remains strong, and a lot of it has been signed. Our attention right now is to turning these signed work orders into activities. It's not as straightforward as just signing the work order, and then you can show up with the full team. It's a process where we go through the initial steps of a big implementation program, and usually the first steps are slightly smaller than the subsequent phases. You can see here again, to your very right, the 62% that I mentioned before.
Same as Q4 last year, but with a very stark development within the quarter where we saw a setback in January compared to Q4 average, and then a relatively rapid improvement over the quarter, which we have also seen continue into April. Now, I will try to take you through this slide. Slightly busy. You will recognize the hype cycle, and then, you will recognize four e-evolution steps that we have focused on previously also with some of you to explain the journey of AI implementation.
If we start to the very right, a little bit counterintuitive, but if you can identify the one that's called AI personal boost, which has come all the way through the hype cycle, the peak of inflated expectations, then it dove a little bit due to hallucinations and others. Now I think I'm expecting everyone on the call to experience this on a daily basis, and it's fairly mature. It's getting better all the time. It's now a well-integrated tool in I believe everybody's daily both personal and professional life. We have the AI professional boost one tick to the left from that. We're not as high up the curve on that one, there's still ways to go.
The easiest way to think about that is that in order for this one to work, it needs to have full access to the company's data sources so that you can truly be efficient. A lot of our customers have yet to reach that stage. But it's going fast. Then, we have the H 3 and the H 4, which I will just deal with combined because we see them collapsing into the same. This is the functional boost, which we've explained previously. This is where you have an AI function specific to a role in finance, in sales, in marketing, in HR, an assistant that really helps you with your specific daily workflows.
The difference between 3 and 4 is that in 4, it's truly autonomous, and it really deals with the workflow autonomously once it's been set up, fine-tuned, and reconfigured, then you enter into a different governance regime. This is yet to happen in large scale. We have done some here at Columbus, and we have done some for our customers. I'll show you a few on the following slide. By and large, the implementation remains. The limitation right now is less, much less technology than it was before. I would actually say the technology is fully ready for this. Especially big companies, and our customers are big customers, big companies with thousands of employees, they have some inertia. They have to consider a lot of things.
How they want to scale this organizationally, that differs a lot from a small startup company. They're much more concerned about security aspects, ethical aspects. They think about long-term governance. They think about the long-term cost implication of this. They also think about the dependency they create on suppliers. You need to go through all of that before they can truly commit to this, and it's how it should be. That's right where we are now, and that's where the Agent in a Day event fits in. They're starting to experiment. They're setting these up, now we're looking for 26 and 27 to be the years where we, where our customers truly scale this.
It means that most of them will have agents in the thousands that needs to be managed and governed, where we see our role as being their trusted advisor and partner in this. Four examples, I'll do them very briefly. Some of them may be rather simple to you, but they do release a lot of a lot of capacity. Top left is an invoicing agent that we built for a Norwegian customer, big company in the construction industry. They have had previously a very time-consuming and also error-prone invoicing process, picking up all the input elements that had to go on an invoice. That could be time, could be materials, could come from many different departments.
That process has then been set up in an agent that is able to generate and validate the invoices. We've not set it up autonomously yet, so it's currently running in a supervised mode. It's very feasible to get to that stage where it actually has a very low error percentage that it can run autonomously. Bottom left, slightly more technical use case, if you think about it like this, previously when you wanted to do some product exploration, research for holidays, buy something hardware, whatever, you would previously probably engage through Google, and a lot of that traffic is currently being moved or has been moved to the large language model.
ChatGPT, Claude, or whatever your preferred LLM is. That actually removes a lot of visibility for the companies that sell. They have previously been used to getting very exact information about who Googles what, which words do we need to use to effectively show in these searches, and what's the buying process. Again, it has been possible to set up an AI functionality that generates these metrics for the customers to fine-tune their marketing efforts on. We foresee this to be at the very early beginning because most of the traffic has actually still not moved globally from the traditional search engines to the LLMs.
You would think so perhaps, in this part of the world, but it's only a fraction that has been moved. This engine is very effective for the specific customer and something which we envisage can be used again. Completely different example on the top right, much more about construction and drawings, where a customer has very valuable, detailed drawings from many, many prior projects. They are in a format which doesn't really allow them to reuse it easily, and that means that they continue to redraw the same things again and again, without having sort of intelligent suggestions of, "Okay, I can see you're making a floor plan approximately like this.
Typically, that leads to maybe four archetypes of outlays for that plan. Which one of them is more suitable? Will you take it in its entirety, or do you want to adapt further? It's much more efficient workflow. The final case study relates to case handling. It's in the U.K., public services, and relates to a function of government that has thousands of calls with citizens that need to be carefully documented, and they need to be segmented and categorized afterwards, and some defined actions need to be taken. This has been very time-consuming post-call activity for the agents before.
Now we are able to deliver a lot of that grunt work to do it, and then it can be vetted by the case handler, and it still needs to be vetted, so it's not autonomous again. Some examples of where you see the AI technology finding its way into our customers. There are many more and it's really accelerating fast. I think we at Columbus, we feel like this time we're in right now is a very special era. Technology development is extremely fast and very complex to understand.
We consider ourselves very fortunate to be right in the middle of this, and having a strong position to advise our customers on the best way forward. Over to Brian for the financials.
Yes. Thank you, Søren. Let's have a brief look at Q1 financials. The three slides. First one is the revenue development, split into our business lines. As we mentioned in the annual report, we now have added our Enterprise Information Management System business line, EIM, the last one you can see. Just if you wonder why there's five and not four as we were reporting on during last year. If we start from the left, Dynamics, still by far the biggest business line. As Søren already mentioned, they are still facing some headwind, especially in the Danish and Swedish markets, to regain momentum.
We do start to see some light, but still 8% down in the quarter, which of course, impact the full company, with the 4% decline Q over Q. M3, roughly a flat development, which is, the good sign here is that we start to see some increased activity in Germany and Sweden, or especially in Sweden, which is by far the biggest, one we have. I mentioned that some quarters backwards, changed some bigger project, and are now slowly starting out on some new bigger project. In that process, there can be months where you have a few people on the bench, waiting, for the customer to be ready. Digital Commerce are still struggling to get back in positive growth.
Again, we start to see some positive signs. The quarter was 9% down. Again, they are big into retail, which have been hit, but there is also start to be increased activity out there. Then we have changed some of the product we have on the shelf within the business lines that we use to our clients and that also do change a bit in how we are getting out to the market. Data & AI, Søren mentioned that, strong growth.
I'm really happy to see that because we do some heavy investment in it, both in strong people that we get in, and also in involving many of our good customers in new project on that agenda. Good to see that they're growing. I would say, the smaller one, EIM, also solid growth, although it's on very limited or small numbers. They actually are moving into Germany, U.K., and U.S., where we already have a presence. It's good to see that they're coming in and are selling jointly with the sales force and consultants we already have in these countries.
If we have a look at the contribution margin, I think overall it is of course hit when you have a not fully satisfied efficiency level, it does hit your your contribution margin. Again, here you can easily see that on our Dynamics business line, which is down 6 percentage points quarter-over-quarter. M3 down to 22%. Here, especially for M3, they had a very good quarter, first quarter last year. Full year they ended at 20% and 25%. 22% is not approved, but it's also not a disaster. It's important to mention. Of course we would and need to get more out of that one as well. Digital Commerce, still around, very small double digits, 12%.
In a decreasing market where they have a top line decreasing 9%, it's good that they can keep the 12% margin here. Of course we would like more, and when the top line start to turn, we certainly expect that they also get up in the 20%s. Data & AI heavily impacted by intake of a lot of new strong consultants during the quarter, also in the beginning of the quarter, and they are starting to get out to customers during the coming quarters. Still generating growth, but of course, we need to see more. We have also invested both in training, as Søren said, Agent in a Day, et cetera, which is primarily driven by this business line.
EIM, they continue a very solid contribution margin, now in the 50% for the quarter, 52% actually. If you look at the real number, they're actually ending up earning DKK 6 million in round figures. Actually the third-largest business line on contribution margin this quarter. I think it's something that we have a strong focus on getting even more speed into that business line, as we see some good progress here. My last slide on the figures is the revenue per market unit or per country. We have been mentioned that earlier. Denmark and Sweden is the two countries where we still see that we need to work harder to regain the positive growth.
They are down with Sweden 7% and Denmark 15%. U.K. down 9% in the quarter. Adjusted for FX, it's actually 3%, close to a zero. They have had a very steep growth the past years, you might remember, and we still expect them the full year to be in the positive end of the growth end. They also had a strong or a difficult January when they came out into the new year. Norway, country have, as you might remember, have been declining the past years and is now really back in growth. They have had some good new major clients up there, and they continue to see momentum, so that's absolutely great.
We are happy to see that we also see a very solid efficiency in our Norwegian teams, that we kept a good team up there even though we went through some difficult years. We now start to see some payback on that one. Yeah, others, it's basically mainly Germany. That's why I would like to mention it this time. Germany see a strong development. We continue to gain new customers in that market. It's not something we talk much about, but they are actually delivering a solid business down there, and something we also would like to focus on going forward. Yeah. That was briefly the numbers from Q1.
On the outlook, as Søren also mentioned, we maintain our outlook for the full year, where we look into a organic growth of 0%-5% and an EBITDA margin between 8%-10%. Good. We move over to questions, and I assume the moderator will handle that.
Great.
Thank you for the presentation. I'll take over now. We're now ready for the Q and A session. To repeat earlier, you can get in line to ask questions by phone by pushing the pound key or hashtag followed by five on the telephone keypad. Should you wish to withdraw from the line, you can push pound key or hashtag followed by six. Please limit yourselves to two questions at a time before joining the back of the queue. If you're watching this by webcast online through a browser, you can ask your written questions in the chat below. I will read them aloud to management after the dial-in section is complete. I will do my best to translate Danish questions to English. The first question from the dial-in part comes from Michael Friis from HCA. The line is now open. Please go ahead.
Regarding these AI agent, just for our understanding, is it your partners, the M3, the Microsoft that you're building in together with, is the Anthropic and the big large language model or a combination? Just to have an understanding on how you are building those AI agents and what is the, you might say, the model behind it.
Yeah. First of all, Michael, the simple answer is yes. It is with our large partners that we're building these agents. Slightly more detailed answer, if we take the world of Dynamics, Microsoft and Copilot, as it's known, there's a belief or a notion that that's probably OpenAI based. Actually, you are able to configure in Copilot Studio, you have a wide selection of LLMs to choose from, which will include those for from Anthropic, for instance. You can configure them that way.
You have different things to be mindful of in terms of data, you know, which data you want to limit to yourselves, GDPR legislation and, so you need to know the specifications of each model there.
Yeah, actually, having it through your partners or someone like Microsoft, I guess some of the hesitation on this one has been, you know, the security, the, the control of the agents and so on. Does that help when you're speaking to clients that this is actually Microsoft who has a, you might say a whole system running in Copilot and other where you already run a lot of your systems and are comfortable with doing that? Would that actually help you in the future, you know, to, you know, to get this off the ramp and get it implemented out in the companies?
Yeah. Okay. Michael, the line is breaking up a little bit. I think I only caught the first part of your question, which was, is it of help, you know, with some of these security concerns that it's a big and reputable company that stands behind it?
Yeah
I would say certainly on the security aspect, I think it's been a big help because I think Microsoft has paid attention to the setup in corporate so that you can basically switch between work and work mode, and you can ensure that your the way your data is handled. I think it's slightly different when it comes to the cost consideration. It's much more difficult for our customers and even for us to foresee the full cost impact with a full visioned rollout for one of our big customers. We're working on that, and Microsoft is working on that, getting that transparency in there.
What does it actually cost to run this agent versus what does it? It is one of our main working areas to be a sort of a guarantor of the business case for the customer and overseeing the charges that you trigger and incur when you set up these agents. Because this will be a, for me, like a lifecycle management, like with everything else, is that some agents you will want to stand down again, they will be replaced by others. You can't just leave it running because then it continues to incur cost. It may also cause issues with conflicting with other workflows.
I think the best way is to think of it is like a lifecycle management where some of them will have to retire at some point, some of them will have to be updated and work in different ways. You can't just set them up and leave them forever. That's not the way it works.
My second question, I hope the line is better, is maybe for Brian. 62% in efficiency. What should you run at? What is the targets for you? What is a comfortable level for you to run at your e-efficiency grade?
Yeah. That's a good question, Michael. Now Søren is sitting here next to me as well. I would say between 65% and 70% would be a strong, healthy level. Behind the 62%, you will also see a lot of pockets that is running on 80% for months. If we have the right clients and the right setup at the right people. Then we might have some corners that is running with 50% because they are waiting a month for the right clients or working on sales or whatever.
Yeah.
On average, now Søren correct me, 65%-70% is definitely where we have to be.
Yeah.
Agree.
Perfect. I have some questions, but I will join the queue and jump in if no one else has.
Philip, is it okay if I just transfer the, or do you have more from the telephone?
There's one more in the queue.
Right. Yeah
Yiwei Zhou from SEB. Your line is now open, so please go ahead.
It's Yiwei from SEB. Thank you for taking my question. Also two from my side. Firstly, on the margin pressure. I mean, I can see that the efficiency has been more or less in line with Q1 last year. What explain the margin decline? Is it price pressure or is it increased investments? Any structure change here in this quarter? I mean, last year the growth, I mean your margin was quite resilient, but what happens here in Q1?
Yeah. It's a good question, Yiwei. We of course also looked at that one. Efficiency has a big part of how the margin is moving, definitely. It's also about which kind of consultant is on the job. Is it more India? Is it more local? Is it high-end? How do we use subcontractors in which we do to mitigate some different push or knowledge maybe. Also, which customers are we taking in?
No doubt that new customers, I wouldn't say it's not a healthy margin, but we sometimes see that when you start up a new customer, it takes some time to get into the right level of margin that we are working on. It's a lot of small factors. It's unfortunately not just one big chunk that I can say that's where it went wrong in Q1, Yiwei. Yeah. That's the closest I can get to that.
Yeah. I think it's a fair point. A lot of what Brian's saying is business as usual for us to manage. I think the one that stands out for me here is the subcontractors. Due to that very fast increase we saw in activity in Norway, we did have to use more subcontractors than what we normally do whilst ramping up our own organization. Which is, I know when you sit there and you say, "Well, efficiency is not that high. How come you couldn't cater for it yourselves immediately?" It's dependent on in which geographies people sit and which specialty areas they possess.
Okay. Can I just follow up? I mean, you have to increase your margin, and it I can see that it's reflecting on target. If you have Q1 decline, I mean, you have to do a big lift in the remaining of the year. I mean, what is the visibility and, what give you the confidence and can you talk about the main margin drivers?
I can start with that, then you can follow up, Brian. I think, Yiwei, What gives us the most confidence, and that also relates to, I think one of the upcoming questions from the interface here, is this, the big increase that we have seen already in the efficiency. That is the most important thing that we have and the one we focus on. There are additional drivers that we work on, the one I just mentioned is already happening, is where we've seen a quick increase in demand for subcontractors we gradually phase in Columbus employees that drive a higher margin.
Third factor has been that I explained already that efficiency was very poor in January, much better in February, and then again better in March. Some of the margin mix is actually that India was one of the countries early on in the year that was doing quite well, and even though we have a good percentage margin from our Indian consultants, the fee rates are lower, so it doesn't drive the same absolute contribution as when our in-country teams are busy, which they are now. What gives us the confidence margin-wise? Brian just said to be in a healthy area, we need to be between 65% and 70%, which I believe is broadly right.
We are already performing now in that range at the lower end, and that's what mainly gives us that confidence. I will say in terms of achieving the long-term target, which we've communicated to you before, of 15% EBITDA, we need to reach the levels of 70%. The way consulting works, as you probably know, is that up until a certain point of efficiency, you're just covering cost, and then once you get above that point, it falls pretty straight to the bottom line with some modifications to that statement.
Okay. Thank you. My next question is regarding the demand outlook. You talked about some improvements here. Can you talk market by market? I mean, where you are building up a pipeline or where you are already receiving large orders?
Okay. I'll try to cover this one, Brian. As you've already seen, and we communicated back in, I believe it was in November, we had this very large contract coming into Norway. We also took over the consultant of another company that gave us a boost. We continue actually to see even more potential from Norway. I would say that is one that I would highlight as there could be more large contracts coming from that direction. Also on the positive side, there's a good strong possibility that we will have some very significant contracts starting up in this quarter from our U.K. subsidiary and also from our U.S. subsidiary.
That's sort of in terms of large contracts that, where you can sort of say you have a single customer impact on overall performance. That being said, what we also saw in the improvement during the course of Q1 was actually just almost like the tide lifting all boats. That we've seen a good improvement in Commerce, as Brian was saying before. They moved quite a lot. There's a question here about the Data & AI where we're growing a lot and where the margin is dropping. We really recruited a lot of people, and that does drive a drop in the margin, which we saw in January, and they have already reverted to +60% also.
We're actually also seeing just a general improvement from existing customers, from smaller customers, and then it's sort of sprinkled with these very, very large contracts. Now I would say Norway, the U.K., U.S. as the first ones. Thank you, Yiwei. We move on. Philip?
Yes.
Yeah.
Great. We have no more people in the dial-in queue. We move over to the written questions. I can see we have answered some of them along the way. There are two questions here on guidance. I think you've addressed some of it already. I'll try and lump it together. There's a question that asks, in relation to the current guidance, how much of that guidance is in projects that you have already signed, and how much is relating to potential new wins that you need to reach in the market?
That is, I don't have an exact figure for this. I will start by saying this, but just to have some exact figures, is that we have an 85% carryover revenue year by year in Columbus, pretty consistently. That means that we have 85% of every year comes from customers that we had in the prior year, and we don't see any development in that. That doesn't mean that 85% is fully covered by the order book or the backlog, or whatever we call it, because some of these customers work on multiple work orders that are renewed and expanded, and some things will conclude and then we move on to other things. Do you have a better way of?
Yeah. No, no, I'm also thinking while you're talking, Søren. It's, I mean, to be fair, any customer can stop the contract next month if they suddenly are hit by whatever is going on around the world. Other customers is increasing suddenly over time. I would also say, it's difficult. No, I don't want to put a number on. That because we normally don't count it like that. I understand the question, it's, I would also say, like Søren, 85% of our revenue is with our strong customer base.
Within visibility.
Yeah.
Yeah.
Yeah.
Yeah.
That's great.
We can work on that, Henrik, for next time if we can give a better, sort of have a better number for that. I think the best way to express it, if we have high visibility of these 85% plus, which is typically a forte of Columbus. What we, because we have these very loyal customers, what we really need to drive the top line growth is expand into new customers, so completely new names. We have a very low percentage of customers that leave us. Because these customers are very big companies, it's not like when the project ends, our relationship ends. Typically, we just continue to the next subsidiary, to the next country, to another data platform that needs something.
They tend to be in a sort of continuous relationship with us. It can vary in activity level, but it's always there.
Great. We have a question about where you see increased activity levels. You mentioned a little on the regions. Is there a particular type of customer that you also see increased activity levels with?
I go back a little bit to the answer we gave to Yiwei. My starting point is that we're really happy about this improvement that we see almost for every business unit. We do see it for every global business unit. We see a very stark improvement in the efficiency. When you break that down to business unit per country, there are still some business unit, particularly Dynamics Sweden and Dynamics Denmark, that we want to see improvement in as the most important ones. Otherwise, that is, gives us a lot of confidence that it's not just one corner of Columbus that has to carry the weight of the guidance. We have it's well sort of spread the load across the units.
Then in terms of the type of customers where we see increased activity, well, that's a response to where we spend time chasing it, of course. That continues the path that we've told about for the last couple of years. We chase the A and B segments of our ICP customer profile, that means they're large international customers, usually with several thousand employees, often more than that. They're almost or always represented in multiple countries, some of them are even sort of true global conglomerates. We don't have a lot of new customers within the lower customer segments, we're not chasing it either.
Right.
That's in terms of type of customer. I would also say we stay pretty strictly within our four verticals, our industry verticals. However, as you could see from my examples, previously, some of them, there was a public sector client in the U.K. We do take the opportunity right now to see and have some engagement also with other sectors.
That's great. We have one-
Yeah, the vast majority-
Great
still within the four that we've had, so far.
We have one minute and one question left. I know you already partly answered it, but it's a question on the Data & AI section where there's high growth in the quarter, but the contribution margin fell sharply. The question is, do you expect this margin to rebound quickly? Is that a result of new hires that you can't invoice for yet in Q1?
The simple answer to that one is yes. It's caused by very big intake of new people. As I said before, the efficiency has already returned. If we do nothing further, the margin will return, has returned. I don't think it's at a stage where we just want to stabilize it. As soon as we see we have a proven business model, they deliver in a new way, they have new products, I would say, that we deliver and the more we get evidence for that's what the customers want, we will hire a lot more.
Right.
It may bounce back and forward a little bit over the next years. We continuously return to wanting to see it work. Once it works again, we take another big step.
That's great. Well, we're out of questions and also out of time. I will hand back to management for closing remarks.
Yeah. Well, I think we just thank you all for attending. It's very, very interesting times. I think we have put enough emphasis on that we are not happy, of course, about the 4% decline in quarter, but that the quarter month-by-month have looked very differently to us, and that April looks a lot more like March, and we continue from here. Thank you.
Thank you.