Traffic acquisition costs

Google announced their Q2 results yesterday. I haven’t dug in too deep into them yet, but one figure that aroused my interest (and should probably interest everyone else in the online marketing and sales business) is Google’s traffic acquisition costs.

Q2 revenue for Google was $17,727M (that’s 17 billion dollars). Their traffic acquisition costs were $3,377M. More than three billion dollars to acquire traffic for one of the world’s most successful online companies. A quick search tells me that this is the money Google pays for partner websites in revenue share.

Google’s TAC is about 19% of its revenue for that quarter. One could also argue that this is Google’s cost of sales, but I guess some form of accounting laws require them to announce it like this.

Regardless – it’s a good reminder for everyone on the share of money to spend on marketing and driving your business. 20% may not seem like a lot, but not a lot of non-VC invested companies are pushing even that much into marketing.

LTE bands matter

This week I’ve learned first hand how much of a difference supported LTE bands can make between two internet connected devices. I’ve had an iPad Mini for a long time and use it quite often through USB tethering to give me connectivity for my MacBook Air.

Earlier this year I bought a OnePlus One – a superb Android phone. It was less than 300 euros and holds pretty superior specifications and technology, given how low cost it was. I’m really happy with the phone, but have realised when travelling outside the urban centers that it doesn’t quite well keep connectivity through LTE and keeps dropping to H+ or 3G.

On the other hand, the iPad Mini keeps perfect LTE connectivity and this led me to dig deeper why so. My iPad is model A1490 and it supports the following bands: 1, 2, 3, 4, 5, 7, 8, 13, 17, 18, 19, 20, 25 and 26. OnePlus One on the other hand supports “only” 1, 3, 4, 7, 17, 38 and 40.

I’m not an expert in radio technologies and how they enable internet connectivity. However, I believe the lack of bands does make a difference especially when outside of the larger cities. Pity, because the OnePlus One is a really good phone otherwise.

The VC point of view

One of the most interesting content producers as of lately are the different venture capitalists that are blogging, podcasting and creating content in other ways. The venture capital business has become extremely competitive and they need to be active towards startups even before the startups require funding. Building a brand some marketers might say.

One of the best ones in this game is Andreesen Horowitz. Their a16z podcast series is phenomenal. Another great addition to this is Benedict Evan’s weekly newsletter (also part of the A16Z team). It’s a headline based newsletter with short commentary on the most important items.

In addition to companies creating content, venture capitalists in general are quite active as well. Perhaps the one I mostly admire is Fred Wilson. Another one I occasionally follow is Mark Suster with Both Sides of the Table.

With the push of the different social media services, it is easy to forget that that the web is full of longer commentary in a more readable format. I’m personally trying to get back to both blogging and reading blogs. I’ve begun to setup a feedly account with my most followed blogs for a closer look.

Venture capitalists are a great source of commentary since they usually reflect their personal backgrounds quite a bit and not their company. They follow the technology industry from a very unique position and hence, it’s interesting to read how they experience it.

Any other good VC companies or personal bloggers out there one should get accustomed to?

Different kind of a summer holiday

Out with Sisu

This year, I’m taking a little bit of a different kind of summer holiday. My wife’s working in July and I’m roaming the country side with my son. In the picture above, we’re in Helsinki at the Havis Amanda statue. We spent that day at the UpCloud office, working a little bit and welcoming a new employee to the UpCloud family.

I’ve always loved traveling a little bit and it does feel a little bit like I’m missing out on something this summer. However, that is all made up by the fact that I get to spend so much time with my son and working a couple of hours each day as well. It does feel like we connect on a lot better level than before the holiday.

It does hit you quite hard to realise how little time you spend with your kids in the end, when you begin to calculate it. Then again, life is made up of choices and I do love my work as well.

Being able to take a few steps back from the normal routines and enjoy an ice cream if we feel like it – that’s summer. And I do hope in the end – he’ll hopefully remember these as fun summers, regardless if he remembers the exact details or not. And that is what counts in the end.

OnePlus One and Lollipop

I got tired of waiting around for the stable release of Lollipop from Cyanogenmod and decided to flash my OnePlus One with one of the CM12 nightly builds. I tried to update first to the CM12.1 Unofficial ROM, but this failed at flashing and I needed to try the official nightlies.

The official 3.4.2015 nightly build has been now running 24 hours or so without any issues and I’ve tried to utilise all the different functions of the phone. No problems whatsoever. A nice surprise was also that these nightlies support OTA updates, so if it keeps stable enough, you should be able to update your phone OTA without needing to install everything again.

Below are the intructions I used to update OnePlus One to Lollipop on a Mac OSX.


  • Data cable
  • Phone battery should be at least 80% full, just to be safe.
  • Mac OS X Toolkit (download through one of the links)
  • One of the CM12 Bacon (device name for OPO) nightlies from the Cyanogenmod Downloads.
  • You’ll need either GAPPS or GAPPS minimal to get the basic Google apps on your phone (I went with minimal).
  • Important: Have backups of all your files as your phone will be COMPLETELY WIPED/ERASED.

And here are the steps I went through:

  1. Fire up the Toolkit on your Mac.
  2. Connect your phone to your Mac with the data cable.
  3. First you need to select the “Unlock the bootloader of my OnePlus One” from the Toolkit. Follow those instructions carefully.
  4. Once you have unlocked the bootloader, do the next step “Flash a recovery”. I chose TWRP from Teamwin. Do follow the Toolkit instructions – they are simple and work well.
  5. Once you did this – root your OnePlus One with the next step in the Mac OS X Toolkit. Again, follow the steps of the Toolkit – they’re really sufficient and well done.
  6. Once you have done this step – add the nightly file to the phone. Also add the right GAPPS version to your phone. Make sure you’re adding just the zip-files. There’s no need to unzip them.
  7. Once the files are on your phone, make sure you have the “Update CM recovery” option de-selected in the Developer options menu under Settings.
  8. Shut off your phone and once shut off, press volume down and the power button combined. This should take you to recovery mode where you can install the nightly.
  9. Go to install and select the nightly file from your phone. Install that and once this has been installed, install the GAPPS-version you went with. Once these are done, reboot the phone and you should enter the Lollipop installation/configuration process.
  10. Now – go configure you’re phone. You are running Lollipop on your OnePlus One.

I did mess up on my own update process at some point and I needed to install the TWRP recovery app again after I had added the files to the phone. This is, because I entered Cyanogenmod recovery mode with the “volume down + power buttons” when I was trying to flash the files. CM recovery does not enable you to do this.

I’ve been super impressed with the nightly I’m on. Again, do all of this at your own risk.

Seven years on

Kiva sent me an e-mail this week thanking me for being a member of the service for seven years. Seven years is a long time in internet years, but Kiva is a truly world changing service that I wholeheartedly support. I can’t remember how or through whom I discovered the service back in the day, but I remember loving the idea from first visit.

The fantastic thing about Kiva is that it takes the internet and gives all the entrepreneurs who would not have the opportunity to leverage the wealth of the world to support them. It is the classic matchmaking service that internet works so well with.

Some of my achievements on Kiva include:

  • $8.17 on credit at the moment
  • $48.62 in outstanding loans
  • $76.50 in total deposists
  • $550 total amount lent
  • 22 loans in total
  • 32 loans by people I invited
  • Lent to 14 of 84 countries on Kiva

One of my “investing” strategies has been to lend primarily to women and mostly in the retail industry. The reason I support women is due to the fact that I believe the success of the world really lies with the requirement of women to become equal members of our society. Education is important, but those who are able to bring food and financial aid to their families will most certainly have better chances at life.

This is also one of the reasons I really love Kiva. You can really help the people in need – all around the world.

Cancelling New York Times

I cancelled my subscription of the New York Times today. This was not a result of there being something wrong with the product, but the amount of time I have available for the service itself. New York Times is still one of my favorite publications along with The Economist. I paid $3.75 a week through monthly payments. Not a lot, but it adds up to yet another $15 a month. These little payments here and there tend to add up – which is the whole point of enabling customers to consume multiple services for small payments.

However, since I didn’t have enough time to read as much as I liked – I decided to cancel my subscription. This wasn’t as easy as I anticipated though. I had to call a US landline to do this. Before getting through I had to wait approximately 13-15 minutes. The call itself was over in about 3-5 minutes. I called through my Skype so the cost wasn’t very high for the call, but it is really a burden for international customers.

But the hassle of calling a landline from overseas to cancel your subscription? Not that useful, despite the obvious increases in continuation of subscriptions this has. I believe in offering the most effortless service to your paid customers – even be it that of them leaving. Because if the processes are really effortless, they might come back again sometime later. The whole point of digital services being automated is the fact that signing customers up isn’t costing you a penny (or for that matter – them leaving you). Now I’m going to a couple of times before I restart my subscription, if I ever do, because I don’t want to spend another 20 minutes on the phone calling a landline in the US.

Raspberry Pi 2 and the coming of servers

The Raspberry Pi 2 was announced today and while it is a great improvement, the bigger news in this BBC news article were the sales figures. So far, 4.5 million Raspberry Pis have been sold and they are selling them at a pace of 200 000 a month. While many talk about the Pi as a computer, I see it more as a server.

The Raspberry Pi has become an inexpensive computer that is able to run lots of different applications, but I personally believe it is at its best in a server-like setup where it is constantly kept on and it is running a set of applications for different purposes.

Servers are the backbone of the internet economy. Internet of things is especially a huge driver for the growing demand. While it may seem to many that this migration to cloud has been on going for a while now – we’re still far from saturation.

I believe both Raspberry Pi and cloud server providers will continue to see increased usage in the coming years as devices and services alike penetrate our lives even deeper bringing up the amount of services we communicate with daily, and also those services continue evolve – requiring more and more computing power.

Very exciting times to be in the cloud computing industry!

Polar Loop and OnePlus One

I’m writing an SEO optimised title for this post on purpose. I was unable to find any other information on this topic so hopefully others will be able to solve this faster with the help of this post.

I’ve been playing around with my OnePlus One tonight and one of the most important things I wanted to get working after syncing my calendars and contacts was Polar Loop (well done Polar, you’ve got me hooked).

Polar does not state on their website that they support the OnePlus One on their website. However since OnePlus One has Bluetooth 4.0 support which is the version that also has support for Bluetooth LE this means that Polar Loop works on the OnePlus One.

To get it to work, you’ll need to unpair it from the previous phone. You can do this by putting the Loop into airplane mode and then bringing it back before syncing it with the Polar Flow app on the phone. To put the Loop into airplane mode, follow the instructions for Other Display Views on Polar’s website.

After this has been done, go to Polar Flow and click on “Settings” in the menu and then “How to Sync”.

Works like a charm.

The role of distribution

I’ve been thinking about distribution quite bit with regards to different businesses. I believe the role of the distribution network is usually overlooked or not given enough importance with regards to the potential growth speed of companies. One of the reasons Supercell was able to ramp up sales of almost a billion euros in just three years was due to the existing and extremely well functioning Apple’s app ecosystem. Hundreds of millions of devices that have a very tightly knit integration with the credit cards of their owners to the accounts on AppStore. There is very little friction between discovery, purchase and actual use.

This is not the case in all industries. For example with UpCloud, we rely fully on the internet as a distribution channel. This of course means that we too have a powerful distribution platform on our hands, but it is not as tightly integrated as that of Apple’s. Our customers need to enter their credit card details to their account on UpCloud to be able to use it. Needless to say we have tried to make the sign-up and purchase of our service as easy as possible.

However, I believe a great way to understand the role of the distribution channel can be summed up through the amount of friction between discovery, purchase and being able to actually use the service. Distribution plays a key role, but so do the types of equipment required to actually consume that service. This is also one of the reasons why HTML5 and other cross platform languages are extremely crucial in designing and building online services these days.

One of the reasons I believe you need to assess the path to purchase from the discovery is because that is ultimately the point when the customer receives the value of the service in return for the money she paid. In the quest to lower friction between the different steps, tutorials and other methods are extremely important in creating a successful on-boarding experience.

I got the idea for this post after I decided to pull the trigger for the Android phone I blogged about previously. While the shopping experience is one that tries to increase the attribute of exclusivity – it is not very scalable nor very enjoyable. I had to wait 90 minutes before I was able to enter the store and then after a few failed attempts, I was able to finally buy the phone.

It might create appeal towards the phone, but it is not very scalable. One has to remember that growing companies have very concrete boundaries of growth that are sometimes very hard to overcome, especially in the case of physical products like that of the OnePlus One. Despite all this, my order number was north of 145 000 from which I gather that OnePlus sold more than 145 000 phones yesterday in just three to four hours.

Despite the tough to scale sales process (one through invitations) – selling 145 000 phones a day is impressive.