NOLL TILL 100 — ENGLISH TRANSCRIPT
Episode 104: Why 60–80% of your time should go on the product page, not the homepage #104

English translation of the full Swedish conversation. Original timestamps preserved.
Original audio: https://www.youtube.com/watch?v=pgYUhC6F9sg

[00:00:00] A very warm welcome to a new episode of Noll till 100. With me, as is now usual, is Fredrik Ankarsköld. My name is Fedja Porobic, and we have loads to get through today, Fredrik. There is so much happening in the world around us. Meta has rolled out a lot of new features. There are rumours that Meta will overtake Google in ad revenue, advertising turnover, for the first time in history. Pretty crazy, really. What are your thoughts?

[00:00:31] One of the interesting things is Meta overtaking Google for the first time. At the same time, despite its profits, it plans to lay off 8,000 staff in May so it can put that money into AI and model development instead. What do you think about that?

[00:00:49] I think that is the direction we are seeing a lot of tech companies move in. I heard Sebastian from Klarna on a podcast with Steven Bartlett, saying that they are no longer firing people; they are simply choosing not to hire replacements when people leave. So there is a natural drop-off of around 10 to 15% a year, and they become more efficient. AI is not replacing people's jobs; they are just choosing not to hire again. I think that approach is healthier and more sensible for a tech company over the long term. Then some may want to make bigger cuts, getting rid of some—I don't know—'dead weight', as they called it in one report.

[00:01:38] Mm. 'Dead weight' sounds very harsh. Yes, it did. I read that too. But it feels as though the stock market has really picked up on the news, and on how everything between Trump and Iran has developed.

[00:01:57] Mm. It has really gained momentum over the past ten days or so. Hopefully that lasts, because rate cuts are back on the agenda. We went from planning for three rate cuts this year to the Riksbank saying the outlook is neutral. The banks themselves say there will be three rate increases this year, and they chose to leave the rate unchanged at the last meeting. Now they are saying there may only be one more rate increase this year, but it will happen later, so we are buying ourselves some time.

[00:02:30] Mm. That changes how the whole macroeconomic picture moves, which means people's purchasing power might return this year after all, as we predicted at the end of last year. But we will see.

[00:02:44] We are keeping our fingers crossed, anyway. Fingers crossed. Things are much better when inflation is low, so we can bring it down, restore people's willingness to buy and get Swedish retail going again.

[00:02:56] Mm. But what I find strange is how fast things are developing at the bigger companies. We are seeing, as you say, Meta laying off 8,000 people.

[00:03:10] Mm. That is a lot of people. What are all those people going to do? I was just thinking about that. Where do they go? Another question is that if companies are not recruiting new staff, there is also—I have heard of some in tech who have struggled to recruit good people because everyone wanted to work for Meta, Google and companies like that. So this could also spread people out more. There are companies that perhaps needed, or still need, those employees. Can they recruit them instead? Yes, perhaps we can. I was talking to two senior developers.

[00:03:48] Mm. That is a very small data set, but both said it is really difficult to get a job in Sweden now. I imagine it is even harder in the US, where a lot of these tech jobs are disappearing.

[00:04:00] But I also think about India. Yes. India has a huge number of developers, and then you also have the Philippines, where a lot of customer service comes from.

[00:04:11] Mm. So that type of export will be hit hardest. Yes. I also think that is where we are heading. I don't know, Microsoft had a few hundred thousand developers just in Bangalore, in India. In one city, they have a hub.

[00:04:27] Mm. What happens to that hub? Yes, it is a little frightening. It is happening much faster at the big companies than we might have expected. Small companies and start-ups adopting new technology is one thing. There is no question that we get small improvements in results. We have felt that; we are quick to respond. But companies this big doing it to such an extent, so quickly, is still a little frightening. You don't really know where we are heading. Now there are also rumours that SpaceX, which Elon runs or owns, is going to buy the partly Swedish-owned Cursor.

[00:05:09] Mm. And that is because they have worked together before. Cursor has helped them with development and may now be acquired in a deal rumoured to value it at 60 billion dollars. That is completely crazy.

[00:05:25] A company co-founded by a 26-year-old Swede, started only a few years ago. Pretty crazy. So 60 billion dollars is some kind of fictional value. But I think the same about Base44, which is also a product similar to Lovable. It was sold after, I think, six months or something. Yes, it was under a year, if I remember correctly.

[00:05:49] And he also received 80 million dollars in cash, plus an agreement that if they reach the targets, he gets another 60 million dollars or so. Yes. A pretty good return. A good return in six months. It is. But AI is an enormous house of cards sometimes.

[00:06:11] Mm. Because there are so many wrappers built on other AI technology. Take Lovable, perhaps one of Sweden's most well-written companies right now—or most written-about, and usually with a very good reputation. You think it is a fantastic tool. But fundamentally it is built on somebody else's technology. That technology could be pulled out from under its feet quite quickly if they make a few small changes, forcing big changes to the business model. Or you could get what is happening now, with Anthropic developing its service and Claude becoming so much better at building websites and apps that a wrapper around it may not be the way to go. I think one of the challenges—I work a lot with AI myself and build AI tools—is where you can help

[00:07:11] customers, because there is still friction. Even if you build something with Claude Code, it starts out as a localhost on your own device. Mm. But take what we do, for example: if you have a wrapper, it becomes more than a wrapper because you work more with cloud hosting and things like that. You might develop something and then just want to press Publish.

[00:07:31] But if you build it the way it has worked up to now—they might add a feature in future, we don't know—there is quite a lot of friction between having a prototype and having a finished site that is public, or that several people in the company can use.

[00:07:48] Yes. Something genuinely usable, where the application can actually be used. Yes. Because I know Lovable bought [unclear in the source: 'molnet'], which is one of those, to develop Lovable Cloud, from a—That is one of the reasons. Yes. Exactly, 100%. So we will see where all these AI wrapper companies end up if Claude gets stronger. It has undeniably become very good.

[00:08:14] So I don't know. We started experimenting with Claude at Porobic Group a few weeks ago and felt almost immediately that it was much better than ChatGPT. Mm. It is almost difficult to replace now, and it has only been a few weeks. They have something you can install, Claude Skills.

[00:08:32] Mm. It is a bit like installing a wrapper inside Claude, as though it has learnt something: marketing skills, for example, or copywriting, which we use, or an audit skill so it can do SEO audits. It can then produce a fairly well-written SEO report that might have cost 20,000 or 30,000 at an SEO agency. I am not joking. All you need to do is connect it to a few data points.

[00:09:01] Mm. It already has lots of other public data points. And because it has presumably been trained by experts in SEO, it has been fed lots of articles about how it works. It can produce a very good, practical, actionable report. So if you have not tried Claude as a retailer, it is time to do so. I also want to mention a mistake I made at the start, before we grew. I now have an organisation, and many people in the organisation use Claude.

[00:09:40] But I initially signed up with my personal account, and you cannot move or transfer that. You can transfer the information, of course. There is presumably some way to export chats and things like that.

[00:09:52] Yes. But get an organisation account from the start so you can assign roles to employees. Rather than everyone getting their own account, you have Claude as an organisation account and invite the employees. The advantage with Claude Skills and connectors is that, as an administrator, you can create skills and say that they are available and preinstalled for the whole organisation.

[00:10:23] For example, an onboarding handbook, if someone wonders [unclear in the source: 'höfte'], 'What is the company's holiday policy?' They can just ask the question. Yes. And you can train it to say, 'This is how we write our newsletters.' One thing I would add about creating skills is to discuss them with Claude so you do not have too many different ones. Say you have one skill for research and another for summarising the research, and you make them two separate skills. But in your message you write, 'I want you to research this and summarise it.' That can produce an error because it does not know

[00:11:08] which one to use. Yes. It needs to prioritise, and that can go wrong. So the way we have done it, as things stand on the day of recording, is that I have one skill that can both research and give me conclusions within the same skill.

[00:11:25] I think that is sensible. We have not got that far. We have different skills, so I just separate them. Mm.

[00:11:33] Across different chats. I work with one skill in one chat, then copy and paste into another chat to carry on from there. Mm. Then I take it into a third chat and continue working. I know it should be possible to connect things so it becomes much smarter.

[00:11:47] Mm. I have not had time to do that yet. We will get there soon. Soon, very soon. I was thinking about something else. With the big companies really getting rid of so many employees, what risks do you see?

[00:12:02] Well, the first thing you think about is obviously what people will do instead. What we are seeing in the US, at least, is more businesses being started. People are becoming more entrepreneurial; they need to be more inventive and come up with things to do. We have talked about how starting an ecommerce business has been getting easier for a long time, and I am seeing more ecommerce companies being started. More people are trying to start their own brand. Somewhere along the way, you romanticise how easy it is, don't you? You read about other people who have done it and think, 'How hard can it be?' It is not that difficult to get started today.

[00:12:46] It is much harder to become really good at it. But I do not know what the long-term effects will be. Elon Musk talks a lot about universal basic income. Mm. UBI has become a well-known term now. A lot of podcasters warn that this is the direction we are heading in. I don't know. I cannot speculate that far ahead, but initially it will mean weaker purchasing power for a large share of the population.

[00:13:16] Yes. What is interesting—and I have reflected on this myself—is that people talk about a few companies owning all the power and controlling all of humanity. But that would not work either, because if they stop receiving money—if people lose their purchasing power—there is nobody to feed the machine that is supposed to be in control.

[00:13:36] Yes, I don't know. It is a good observation. But you are only speculating. Another speculation is that I think OpenAI has been overtaken a little.

[00:13:46] Mm. Claude is better at b—business cases, by far. Other services, such as Gemini, have caught up with ChatGPT on general-purpose work. And OpenAI seems to be panicking. How has Sora 2 development gone, or

[00:14:06] how has their AGI development gone? They have not released a new model in quite a while. There is not that much new happening at OpenAI. They cannot keep being this unprofitable, burning money every day and needing more and more power plants all the time.

[00:14:29] Mm. So I think one or more of these enormously valued AI companies will go under at some point. I am not saying it is a bubble, because there is underlying value. But you obviously cannot remain unprofitable for ever. They have tried advertising, but it has not really been rolled out widely. I don't know. It feels as though OpenAI will have a very tough time unless they manage a major change this year. Yes. It is funny, because just yesterday I was listening to someone who is one of Anthropic's investors.

[00:15:13] Mm. He has also invested in OpenAI. But he was talking about Anthropic, which owns Claude, saying they can become profitable whenever they choose. He argues that they are now making so much money, with so many business customers coming to them, that the risk and the bubble are no longer what they were before.

[00:15:37] He is invested, of course, so he has a bias. But it was quite interesting to hear. They were discussing how much money they were losing, and he said it was not as bad as many people think, at least for Claude. Claude is also used less widely by the general public.

[00:15:53] Mm. So it is very much business use, which is what it is good at. And when the general public uses ChatGPT, every prompt uses a little electricity each time.

[00:16:06] Have you tried ChatGPT Images 2.0, which came out recently? No, I didn't even know it had come out. When was it launched? Recently? Yes. What even is Images 2.0? They are just boasting about 'a new era of image generation'. You have to test it in real life, because there are different benchmarks. But we will try it before next time.

[00:16:30] We certainly will. A lot has happened on the Meta front too. Mm. Meta is rolling out updates non-stop. One of the bigger developments, I would say, is that they are following through on their plan to become more AI-driven. Manus, the AI agent they acquired as an app only six to eight weeks ago, has been integrated and is getting better and better. It is a little better now than when Jacob and I last talked about it, around three episodes ago. You can at least see a clear direction: they want us to move towards more autonomous media buying.

[00:17:14] Mm. As in: enter your budget, give us your goal, what do you want to do, what outcome do you want? And let us handle the rest. We are not there yet, but we could easily be there in a few years.

[00:17:27] Mm. I think, just do it. Yes, because at—what is it called?—[unclear in the source: 'First smoothie'], we are developing influencer marketing templates. We are researching different influencer platforms to see what is available. One thing I found interesting, which has been appearing more and more, is that you no longer just have a filter to narrow things down based on what you hope or assume makes the right influencer. Instead, you will be able to describe things in the influencer tool's chat: 'This is my company, this is what we sell, this is our target audience.' AI can then help find influencers based on a prompt rather than a filter search.

[00:18:03] How do you think you could take that further? Normally, with influencer marketing, you enter the criteria yourself: I want this many followers, I want this engagement rate. But follower count is not always what determines whether an influencer is right. It is about whether they have the right audience.

[00:18:22] Yes, of course. So you think you can get a bit of help? Yes, a little more help at the start. Yes, that is sensible. Another feature I saw Meta rolling out is Website Highlights becoming standard. They are starting to roll it out. What that means is that ads will preload part of the PDP. Usually you link to a PDP or the homepage, so the website will appear inside the ad. It takes the landing page you link to and shows that landing page in the ad. That puts even more weight on the UX needing to be really damn slick. Otherwise, you will get more expensive ads because you have a poor conversion rate, a poor click-through rate, a poor conversion rate. It all becomes even more connected as their beta starts being rolled out more

[00:19:20] widely. I have not seen it in our ad account yet, but I expect it very soon. Since we have seen very good data from American companies, I think this is definitely a feature worth keeping an eye on.

[00:19:39] But what do you do if, for example, you link to a listicle page instead? That will be shown in the ad. We usually send traffic to different landing pages, and sometimes the landing page is external. We might want to send traffic to—say, a Breakit article

[00:19:59] about Västerbottenssåpa, to give people positive top-of-funnel awareness of Västerbottenssåpa. Not because they will buy directly in that article, but they might go from the article to us, or from the article to Google and search for Västerbottenssåpa.

[00:20:17] Mm. So it is one step removed. But credibility becomes so much higher if the source you link to is credible. That makes me think of what we discussed in an earlier episode: when you do PR for Twistboxes by pre-writing articles with AI and sending them out, is the idea also to link to those articles in your Facebook ads?

[00:20:43] We have done that, yes. Not specifically for Twistboxes, but for other brands, and it works really, really well. I think we will do it for Twistboxes too. For one thing, you get instant trust when a major newspaper writes positively about the company.

[00:21:04] And what percentage, roughly? You can still optimise for purchase conversions. Yes. How many different articles do you want a person to see before sending them to the actual product page, and how much of the budget goes to that? We usually put only a very small part of the budget towards an article, and we pick the best one. Then we agree with the journalist: 'We are going to run ads for your article.'

[00:21:28] 'That is good for your newspaper. You will get a boost in readers. Could you put a link to our site somewhere in the article? Just one link.' So instead of simply saying Västerbottenssåpa, the name is also highlighted in blue and linked to Västerbottenssåpa's website. That helps our SEO visibility, but it also helps the journalists if we start advertising the article.

[00:21:50] Yes, it is win-win. A real win-win. We do not always get—yes, most newspapers say, 'Unfortunately, we do not link out from our newspaper.' But some do. I know many tech publications find this very interesting.

[00:22:06] Mm. Niche publications usually say it is no problem for them to include a link. Mm. They do not really lose anything either. They gain traffic. They gain traffic and hopefully new readers. People discover the publication. And a lot of these longer articles are often behind paywalls.

[00:22:25] Mm. They might even get a few new paying subscribers from our ad spend. Win-win. We have more updates too. Things move fast at Meta. Meta now has a feature—or has had it in beta for a while—where they are rolling out more and more AI-generated content in video ads. It is a little frightening because it feels as though you lose control over what gets shown with your own money. Of course there is

[00:22:58] some choice here. You can turn that feature off. Do you usually have it on or off? It is on by default. I have seen random things appear that are completely off-brand, which the AI thinks it can generate for us. But

[00:23:14] I know it loves strange music over the images. Yes. It is far from perfect. I think it gets things wrong eight, nine or ten times out of ten at the moment. So it is not an amazing feature, but it shows where Meta is heading. What has been in beta for a while is now starting to roll out widely. There will be more and more AI-generated content in ads. We are also seeing more reports suggesting that this works. Most of Meta's ad spend comes from big companies, which generate a huge share of its revenue. Those companies may advertise with enormous budgets in several hundred countries and lots of different languages. For companies used to working fairly slowly with creative, this becomes

[00:24:08] a really good upgrade, I think. Don't you think there is a risk that those companies will be even more afraid of going off-brand? I think there is obviously a big risk, because they want to be 100% on-brand. But looking a few months ahead, and then a year ahead, I think the feature will improve. For those companies, it will become a big cash cow.

[00:24:32] Compared with their old ads, those were worse than the new AI-generated ones. So what this does is help the very worst advertisers, who are very good at running a few ads with enormous budgets, very high saturation, and rolling them out identically in every market. Mm.

[00:24:55] They will move up a level. Everything will become a bit more middle-of-the-road, so it will be easier to create decent content while still being very difficult to create authentic, highly engaging content. That does not change things for start-ups competing on those stronger qualities. But average content will become more or less evenly matched.

[00:25:27] That makes me think about tracking. They are also going to force—or not force, but it is almost a requirement—to use the Conversions API rather than just pixel tracking. Yes, but we have recommended that for years, so it is not strange that things are moving in that direction. If you still have not installed the Conversions API and moved to backend tracking, I don't know why you are listening to this podcast.

[00:25:53] Perhaps you should listen now to learn how. Yes, absolutely. We have mentioned it so many times that I hope the penny has dropped.

[00:26:02] For those using Shopify, at least, if you have not done it, go to Sales channels. Choose Facebook and Instagram, then Settings, then Data sharing settings, and change it to Maximum. You have then activated the Conversions API, which shares the most data with Meta.

[00:26:23] Yes. It is native, so it is fairly simple, and it does not cost anything. Other companies using platforms such as Askås and wanting to install it—nothing against Askås otherwise. Well, maybe a little. I do not like their business model. But those companies have to pay Askås quite a substantial amount to integrate with CAPI.

[00:26:48] Mm. Facebook's Conversions API. And then they want an ongoing fee for it. You think, 'Why? You have already done this for a hundred, probably two or three hundred customers.' Mm.

[00:27:01] Why do you want recurring revenue for it? 'Well, we have to maintain the API.' But come on, it is around a thousand kronor a month, or whatever it costs. In some cases that is as expensive as a fairly good premium app.

[00:27:12] Mm. Hello? Doesn't that ring a bell somewhere? Don't do this, because it causes problems for so many ecommerce businesses. I get annoyed with certain Swedish platforms that hold brands hostage. Every time I speak to a brand on Askås or something similar, I ask, 'Why are you still there?'

[00:27:34] Mm. Please tell me. It is so easy to move to Shopify or BigCommerce. I would choose Shopify, but still: why stay on an outdated platform that is technically worse, with higher risk—much higher risk—without access to the whole app store, and which also wants extra monthly fees for every little feature it implements?

[00:27:57] But I remember an ecommerce business I co-own. We were on another platform before moving to Shopify years ago. The reason was some simple, basic feature that an almost free Shopify plugin could do. They wanted absurd amounts of money to help us, and it would take bloody ages. When you move to Shopify, you just click to install the app and it is almost done.

[00:28:22] Yes, it is incredibly annoying. It is like having an iPhone or Android phone nowadays: you download an app from the app store, and it works out of the box.

[00:28:30] Mm. You are no longer surprised to have an app in your hand that works. But some ecommerce businesses are quite surprised: 'What, is it that easy to integrate tracking? Is it that easy to integrate analytics—third-party analytics?' You just had to install an app. Yes, you just had to install an app.

[00:28:50] Mm. You could change things in the checkout by installing an app. Yes, you can. Installing Swish was no harder than that. But I think back to the old days, because I have been doing this for so long. I remember having to understand HTML code to go in and do those things back then.

[00:29:07] Yes, 100%. I did that too. Ten or fifteen years ago, I was very nerdy about web development. But I have barely needed to touch that in the last ten years. Thinking about these brands, what do you think stops them? What mindset prevents many brands from switching platform? That is a difficult, broad question.

[00:29:30] Many people I speak to say, 'It is easier because we have Swedish support.' Mm. It is nice to have someone to call. I think, yes, perhaps it is. But Shopify sometimes has Swedish support too. They have a Swedish office in Stockholm, so I don't know. I have not needed it myself, because there are so many guides available, but I imagine they have good support too.

[00:29:51] Yes, their support, anyway. Have you received support in Swedish from Shopify? No, but I always choose English because it is faster, and I speak English, so for me it is not

[00:30:02] No, it is not enough. The cost of using something else compared with speaking English to support—the difference is too great. No. What else do people say? Many think there are ready-integrated plugins for the Swedish ecosystem on a Swedish platform.

[00:30:21] Mm. There are a few Swedish platforms I think are all right. Quickbutik gets a thumbs-up, at least. They have made it fairly similar to Shopify. It is a relatively recently developed platform, perhaps only six or seven years old, with a fairly modern technology stack compared with many of the others.

[00:30:40] The others really are legacy systems. They really are. You look at the interface when you log in and it is almost frightening. Clicking your way around takes two years. If you are used to things happening instantly, you get annoyed. Then tracking does not work, analytics sometimes does not work, the sales do not show up, and you just get generally irritated. Mm. But

[00:31:02] there are integrations with external platforms and apps, and AI development is moving incredibly fast in web development. That is not even happening on many of these Swedish ecommerce platforms. But one thing people mention is that it is easier with the Swedish ecosystem. You might have a local app you need for inventory management.

[00:31:27] You have nShift, and perhaps Ongoing integrated with—but they have Shopify plugins too, don't they? I know nShift does, at least. Yes. They are so big, but I don't think Ongoing does. I don't know, but there are some smaller local providers that do not—Ongoing has a Shopify one. Yes. Yes, they have—there you go. There is no reason not to. Most of the slightly larger providers in Sweden now develop their own Shopify apps. There was a reason two or three years ago.

[00:32:06] Mm. That reason is clearly starting to disappear when even Ongoing has a Shopify app. It almost sounds as if they sponsor us, but we are not being paid to say these things.

[00:32:14] No, why would we be? Sponsorship is a huge hassle, with reporting, administration, invoicing and everything else. Mm. It is not only because we do not want sponsorship. But it is true: we are not sponsored. We both just happen to use Shopify and have insight into all kinds of other platforms. I work with many different partners, and naturally they are on different platforms. We are migrating some of them to Shopify now. That process is usually fairly quick and painless. A lot of people raise their eyebrows and say, 'Was it that easy to learn a new system?' It is so intuitive. How would you handle this?

[00:32:56] For example—I won't name names—my girlfriend was a customer of a store. When they moved to a different ecommerce platform, she had an existing subscription, where you receive a refill every week—or every month. The problem was that she suddenly received an email saying, 'Your payment did not go through.' She thought, 'Didn't go through? Did I not have money in the account? But I do have money in the account.' The email she received was written almost as though the failed payment was her fault.

[00:33:34] Oh. When in fact they had changed platform, and she needed to enter her card again. Mm. Something must have happened in that migration, because you should not need to do that. You can transfer all the subscribers.

[00:33:48] They must have missed it somewhere. Either they had very few subscribers and it was a small part of the business that was not a priority, or they planned to implement the feature later: 'We just want to roll this out and go live quickly.'

[00:33:59] Mm. Or perhaps a third possibility is that it was technically impossible. But in that case they should at least have told those customers: 'Just so you know, we have changed platform. Please update your credit card details on our new platform if you want to keep using our service.' Yes, because I discovered it when I met the company that had built the new platform.

[00:34:24] Right. That was when the penny dropped: 'Oh, that is why it stopped.' But why did it stop on the new platform?

[00:34:33] I did not have time to ask them, but that was when the penny dropped for me. I see: you found out they had changed platform. Yes, because I had not known that before.

[00:34:41] Yes. And perhaps your girlfriend had not either. No. I think many companies assume all their customers will obviously notice the difference. Yes. But a lot of customers do not, especially if they have a subscription, because they might not have visited the website for a year.

[00:34:58] Yes. That is highly likely, and when you migrate you usually try to keep the same elements. Even if someone visited last month—people do not remember what they had for breakfast yesterday.

[00:35:11] Mm. Or lunch or dinner. You barely remember what has happened in your own life recently. How would you remember a website from more than a month ago? I know it myself when I receive newsletters. Perhaps they have not emailed me for a long enough time, and it was something I saw once and happened to sign up for. Usually not if I have bought something, but perhaps because I was interested in a PDF or wanted to download something. That is more likely.

[00:35:38] Yes. But they assume I remember who the company is. Yes, I know. So annoying. Whenever you create lead ads, the mistake is making too many assumptions. One is sending the email far too late.

[00:35:53] Mm. Send the email within 30 minutes of someone signing up, while it is still fresh in their mind. The first email should confirm that they signed up and share whatever you committed to sharing.

[00:36:08] Mm. Then explain at the end of the email what will happen next, so the person does not immediately unsubscribe. That is also a bad signal to [unclear in the source: 'ja seran plattformarna'], which in turn sends a bad signal to the inbox, whether that is Gmail, Hotmail or wherever they are.

[00:36:27] Mm. It is very easy to make these mistakes, and I see them again and again. Lead ads, by the way, are one of the most effective ways to advertise, especially slightly more expensive products. You move people from the ad into a large email database, where they receive continuous communication in a flow. Fantastic. You get another touchpoint you would not otherwise have had, at a fairly low CPL, cost per lead. What do you do when you build landing pages? You mentioned building your own landing pages for different products.

[00:37:03] Yes. When you are not sending ad traffic to a journalist's article, you send it to landing pages. Is the aim of those landing pages to collect people's email addresses, or what is the aim?

[00:37:16] It varies. Usually, when you work with products and D2C, the aim is to get a conversion on the landing page. You want that conversion to be as accessible as possible, at as high a rate as possible, with strong buying triggers. The secondary aim is that if people do not convert here and now, they leave a trace. That could be an email address, a telephone number or something else, with their permission for us to communicate with them later. Usually we have a pop-up. Ideally, it should not appear immediately, but after they have shown some interest: staying on the site for 20–30 seconds, adding something to

[00:38:01] the cart, or moving to another page on the same site, clicking between landing pages. Then a pop-up appears. We experiment with these a lot. It could be anything from a quiz to a downloadable PDF or something else in exchange for an email address. Or we ask when your birthday is: you enter it because you will receive a surprise. There are several different approaches. It is not that difficult, really. You can also test a direct discount. If you give, say, 15% off the whole purchase, try reducing it to 10%. I think there is almost no effect

[00:38:47] from that. That 5% goes straight to the bottom line. There are lots of things you can easily test to save money. It is also a bit better for the brand than discounting the first purchase. For one thing, it annoys existing, loyal customers: 'Why do new customers get 15% off? Where is my 15% off?' And it is bad for the brand if the first thing you see when you discover it is discounts being handed out.

[00:39:13] Mm. That is not good in the long term either, because you get stuck thinking, 'I got 15% off my first purchase.' That is the value of the product in your head. You are not willing to pay full price when it is time for a refill, a restock or a new product, unfortunately. You end up waiting for the next discount.

[00:39:31] What is actually the difference between a landing page and a product page, then? You can complete the purchase on both? A landing page can be a product page. When people say landing page, many mean an external page somewhere out on a URL. But most commonly, your landing page is your PDP, your 'product data page'. The product page is your landing page. We make it as long and as clear as necessary so that, when we send traffic there, you get answers to all your questions. Many companies have very simple product pages. But if that is where we send the traffic we pay for, you are shooting yourself in the—if you have not developed a well-crafted product page, a PDP. So my strongest, strongest recommendation is to spend about 60 to 80% of the time you

[00:40:28] work on the website on the product pages, because that is where you need to convert the customer. I see again and again that this is not done in the industry. It is something very straightforward. Just look at how many of the best-performing companies in Sweden do it. Look at Västerbottenssåpa, which we discussed a lot in the previous episode. Look at another company working in your industry, and you will get lots of ideas for how your landing page should look.

[00:40:56] One idea for the future is to do a roast where we bring up different sites and give people concrete suggestions. Either listeners send in, 'Could you analyse my page and give me feedback?' or we pick random sites to review. That would give people specific suggestions for what they could do.

[00:41:20] Absolutely. Since this is a video podcast, we should also be able to share the screen while doing it. Yes. I know Jacob and I did that a long time ago, near the start of this podcast, in '24, in 2024. That episode was very popular because people received practical tips.

[00:41:38] Mm. Even if it was not their company we were discussing, lots of people commented afterwards: 'Oh, that is how to do it, or how to think about it.' This is still a very important part, and many companies still get it wrong. So it is [unclear in the source: 'små knapp gör stor succé här'].

[00:41:56] Mm. People tend to think that stripped-back, minimalist and simple looks good and will convert well. But it is the opposite. Longer, detailed, informative pages are usually what wins over the long term.

[00:42:15] Yes. If you worry there is too much information, just use a sticky add-to-cart. Once someone scrolls past the top of the page, the first thing they see, the cart button is still there further down. If they think, 'I have read enough now,' they can still complete the purchase.

[00:42:33] Yes. That is about the simplest thing. Another question: what do you think about Meta now receiving fines in the billions because it is considered to cause social media addiction, especially among children? Oh, I actually do not know about this, but I am glad it has. The whole shift was meant to make us more social. Their vision when they created the platform was 'connecting the people'. But the more connected we become online, the more disconnected we become

[00:43:09] offline. A fine in the billions is still a very small part of their total revenue—not even their annual revenue, but their monthly revenue. Mm. So I think it is good, and in all likelihood we will see greater restrictions on social media apps for children. Meta was fairly early in introducing some restrictions. You are not supposed to be able to register on Facebook unless you are thirteen.

[00:43:40] Mm. I do not know the exact age. But they introduced that sort of restriction years ago. What people do is simply enter a different age, and Meta has not done its best to verify ages. It has not even tried, because active users have been what matters. That is not surprising when the whole business makes money from ads. So, at first glance, I think it is good. I know that in Sweden, around ten years ago, many school rules changed: we were going to become more digital, with more digital teaching and much more digital everything. Textbooks started being replaced by digital booklets, digital teaching—everything was going digital. They bought

[00:44:31] iPads for nurseries, and I think a lot of parents recognise that their nursery also has iPads. Children scroll and get hooked on the rapid images, bright light, good music and sounds. Even with developmental, educational apps—EdTech, as it is called—the EdTech app's goal is to keep people scrolling and staying in the app, so it has a reason to exist. Now the Public Health—was it the Swedish Public Health Agency that issued new guidelines last autumn about screen time? Schools then started updating things and reducing screen time too. Swedish schools are moving back towards textbooks and physical text on paper, with a little less digital content, which I think is very, very good. So a major reason

[00:45:32] for the addiction is social media. Unfortunately, it has also been a major contributing cause of higher rates of depression and suicide worldwide. A lot happens because people compare themselves with others, and it mainly affects young girls.

[00:45:52] Mm. Yes, that is tough. How do you regulate your children's screen time? We try not to have screens at all. We were worse at it when they were small, precisely when you should be doing it best. I was not sufficiently well-informed either, I can admit, especially in my daughter's first two or three years. Then we removed the iPads completely. At first, she could sit with an iPad if we were at a restaurant, so we could get a little peace and quiet.

[00:46:28] You thought, 'What is wrong with that? Give me a break. Life is hard enough, and I have so damn much to do.' But you noticed that, even going into the supermarket, she would quickly become irritated, lose concentration and struggle to play by herself. Many of those things improved as soon as the screens disappeared. Now we regulate it more. We might watch an actual film on a Friday evening, have a cosy family night and enjoy it together, instead of coming home and scrolling on a screen. That happened many years ago now. There were about two or three weeks of real detox.

[00:47:10] Mm. It is horrible to say, but it was like a real addiction to hard drugs. They could be very, very sensitive. Crying, irritated, screaming, all sorts of things during those first two or three weeks. You could see it getting gradually better. After three weeks, it was as though they had never used an iPad. So it is possible to wean them off fairly quickly, because children's brains are presumably so flexible and develop so fast.

[00:47:37] Yes. I have heard the same thing when children get used to unhealthy food: they may even refuse to eat for a few days. Mm. But fairly soon afterwards, if you do it the right way, the children adapt too.

[00:47:52] Yes, I think so too. Society is moving in the right direction there, at least. I am fairly convinced of that. Soon we ought to see regulation around AI and its use, because we do not yet know the consequences and effects. But we also know AI development is moving so quickly, especially in educational settings such as schools. I want to share a tip here, because I had problems with motivation for a while. One of the things I discovered was—I was seeing a psychotherapist four hours a week. I had a mental coach. I was doing, what is it called, self-hypnosis.

[00:48:34] This was every week, and my motivation started declining. I wondered, 'How can it be going down? I do not know anyone who does this much. I do mental training morning and evening, I have a coach and a psychotherapist, and my motivation is going downhill.' Then I discovered that I was looking for motivational videos on YouTube, or educational—I thought I was watching educational YouTube videos to learn something new, or listening to a podcast. I did not understand that it had become a vicious circle: my brain wanted the kick of feeling it was learning something new. Then I redefined learning as applying something in real life.

[00:49:21] That gave me a different perspective. I found an app, unfortunately only for Apple at the moment, I think, called [unclear in the source: 'Yomo'], meaning Joy of Missing Out. I have set it up to block things. They have some kind of cooperation with Apple, with the privacy settings, so you cannot unlock it.

[00:49:40] Oh. I can set it up so that in the morning, or between certain times, I cannot access things such as email. Yes, exactly. For example, I have blocked Instagram on my phone. I have blocked LinkedIn. They are completely blocked; I cannot even unlock them. For email, I have put in a forty-second time lock.

[00:50:06] Otherwise, I would casually check my email until I realised I had checked it fifteen times. How many times am I going to check it? This removes those little impulses, the things you use to avoid being bored. A forty-second delay is still enough to make you ask, 'Do I really need to check this now?' I get access to Facebook for one minute a day to look

[00:50:30] whether anything has happened in a—Then, if someone has written something, I can contact them on Messenger afterwards, because that app is unlocked. YouTube is completely blocked on my phone.

[00:50:41] Mm-hm. So I do not do that idle—But I have YouTube Music, so I can still listen to podcasts, and Spotify. Mm. It is not the same stimulation as sitting and watching something. I cannot actively sit looking at the screen. If I want to learn something new, I have to sit down and engage with the video, take notes and perhaps put them into AI.

[00:51:05] I do not always manage that, but it is what I aim for: to get a checklist of actions I can take from it. There must be a similar app for Android. Android is a fairly open operating system compared with Apple's closed ecosystem, so there ought to be something similar. It sounds like an interesting thing to try.

[00:51:23] I have had so much more to do—I mean, got so much more done. I feel more present and enjoy everyday life more. I have a thousand questions about this, so we should have lunch and you can tell me more. I also want to learn more about self-hypnosis: how it works and what it is supposed to lead to. Yes.

[00:51:42] Before we wrap up the podcast, I want to thank all the listeners. This is our third episode together now, Fredrik, and we have had good feedback. Just yesterday, two completely new people wrote to me on LinkedIn: 'How nice to see Fredrik in front of the camera and listen. For the first ten minutes, I did not know what to expect, but the episode turned out really well.'

[00:52:17] So I was not a complete idiot. No. I think people do not know—yes, it was very good. Mm. If you have listened to lots of episodes with two people, you get a bit sad when one changes, because our brains work like that. If something changes at the bank, you also think, 'Damn, I want to go back to the old version.'

[00:52:47] Mm. You almost get annoyed, because we are creatures of habit. We like things the way they have always been. Mm. Even if something gets better, you get annoyed and want it to be the way it was. Why has it changed? People fundamentally do not like change. With an updated website, for example, people may think it is much better. But if you have a lot of visitors returning quite often, those ordinary people get quite irritated because they can no longer find their way around. Even if ten clicks have become three, they think, 'I do not know these three new clicks. Give me my ten clicks back.'

[00:53:30] Conversion then drops a little before eventually coming back, and people are delighted. There is always that transition period, and you and I are in that period now, Fredrik. So, if you have feedback on the podcast,

[00:53:44] please write to us, because that is how we develop it. Yes, write to me on LinkedIn too. Links to both of us are in the show notes. Absolutely. All right, we will see you again next Thursday, as usual. I hope you have a lovely weekend coming up. Bye.

[00:54:05] Bye. Yeah.