Mar 31, 2017

Every company that will use AI faces a trade off

To fully function at its best potential, AI needs to learn "on the job", and not in a simulated environment. So everyone willing to use AI will have to live with a weaker product in the beginning, and will have to try and make it as smart as possible before launching it. Interesting read about this challenge:

Object recognition changes search for photo and video

We are used to tag stuff. Even texts, in order to make them searchable or be able to quickly put together compilations. That might be a thing of the past as AI evolves and object recognition gets better and better. Google unveiled a search recently that is able to determine objects in videos. And in November 2015, the open-sourced tensor flow, their machine learning project, which is now used in Google Photos to recognize objects. Soon, tagging might be over. 

Agility vs Consistency

When everyone goes agile and instead of research just throws products out there and sees how they are doing, who ensures consistency? And why would consistency be a success factor? Short and interesting (older, Jan 2017) piece on Havard Business review around this topic. 

Sports Highlights on Instagram

Nice Overview on Wired about how Sports is one major topic on Instagram. The article has a few interesting numbers in it, for example that House of Highlights, owned by Bleacher report, has a (stunning) interaction rate of 2,43% on Instagram. 

Mar 10, 2017

Overview of machine learning / AI potential in 12 sectors

Apparently taken from a McKinsey report, this article on medium lists 12 industries and evaluates how much which area within those will be impacted by automation and artificial intelligence. AI is always scary and fascinating at the same time, and you cannot begin to imagine how many "human jobs" will be obsolete when all this happens. On the other hand, everything will be more effective, cheaper, accurate and powerful. The article offers no interpretation, just a list of business areas and how they may change. Still a great resource:

Computers with an IQ of 10.000

A genious human has an IQ of about 200. Softbank Robotics CEO Masayoshi Son predicts that by 2047, in 30 years, a computer might have an IQ of 10.000. Which means they will decide and do things that humans are completely incapable of understanding.

[German]: Grundkurs App Economy von Hannes Kleist

Sehr interessanter Podcast, der mindestens zur ersten Hälfte als "Grundkurs App Economy" taugt. Hannes Kleist, Gründer der App-Agentur und -Publisher Stanwood, erzählt die Geschichte ihrer TV-Programm-App "On Air", und gibt dabei sehr lehrreiche Insights preis, wie man erfolgreich Apps in der derzeitigen App-Store-Economy betreibt. Toller, ca. 1-stündiger Podcast hier:

Gartner says Facebook becomes the WeChat of the West

A year ago I started giving talks about Messengers, and halfway through I had a chapter called "the WeChat-ification of Facebook Messenger". That was not based on the monopolization of mobile user attention, which is very much the case in China with WeChat, but on Facebook's move to make Messenger a platform on which 3rd party applications (bots etc.) can be built, distributed, conducted that may even threaten the app economy as such. This TechCrunch article lays out the recent developments on Messenger and how this could be the next big platform:

A conversation with Elon Musk

35 min talk in the United Arab Emirates with Elon Musk. At 8:35 he starts talking about AI and autonomous cars, predicted that in 10 years from now the vast majority of newly built cars would be driving autonomously. At 10:00 he classifies this as "narrow AI", while the real opportunity - but also threat - will be "general artificial intelligence" which he sees developing like super intelligent aliens. If you don't have time and want to be scared a bit - start at minute 20.

Facebook Live "TV-shows"

We are seeing many publishers experimenting with Facebook live, and Facebook recently announced they would introduce video ad breaks in live and on-demand video, fed by the Audience Network. While this type of monetization is important for (automated) scale, I keep thinking that branded content and sponsoring is a great way for publishers to monetize their social media presences. Perform's is running a "TV-like" show that monetizes on the integration of brands, and some insights are shared here:

Stadium WiFi and tech monetization in the NFL

Very good article from TechRepublic about how the NFL uses in-stadium WiFi and apps. Super Bowl 50 saw 10.1 terabytes of data flow through the Levi's stadium's WiFi (63% increase to Super Bowl 49). Team apps feature in-stadium sections with the ability to order food&beverages to your seat, which obviously increases turnover, merchandising discounts and info on waiting lines for restrooms. The Patriots' app is used by 18-22k people each gameday - with 66000 in attendance. More great stats and info:

State of the German Internet (3-17)

Nice presentation from Online Marketing Rockstars and the relation of German internet companies to GAFA (Google, Apple, Facebook, Amazon) and other global players. The link features both the talk (video) and a slideshare.

Feb 3, 2017

Digital in 2017 - collection of global stats

We are social frequently releases an aggregation of research results and statistics, and the 2017 report (which of course is a 2016 report :-) features some nice insights about digital: internet penetration, smartphone penetration, social media usage etc. etc.

Global report here:

Germany starts on slide 78:

Fun fact for Germany: 80.6mn population, 109mn smartphone contracts. 136% penetration.

The schedule and the stream

Long read, but worth every minute: An analysis of the stream as a relatively new way of organizing information (approx. 10 years): one of my main topics in consulting. This piece compares it to the schedule as one of the many "old" ways of organizing information and content, and how both differ in creating public spaces, and what that means for our societies, politics etc. Highly interesting.

1mn concurrent streams for a counterstrike match on Twitch

1mn concurrent viewers is nothing for linear TV, especially not for a globally accessible program. But in streaming esports, that's a real milesstone, and the match was also available on TV, ELeague's website and on YouTube. Streaming is special because there is no "audience flow" like we know it from linear programming, no accidental audience because of zapping or viewers that watched it "becausethere was nothing better on other channels". You also don't typically stream and do the dishes in parallel like with linear TV - meaning: a streaming audience is most likely a dedicated audience, and 1 mn concurrent viewers for a platform like Twitch is a stunningly high number.

Finally: some Snapchat numbers

Snapchat filed for an IPO and therefore had to release some numbers on users and usage: Topline is 161 million daily active users (DAU), 53mn of them in Europe, 400mn revenue in 2016 (coming from 59 mn), most of it in the US, and staggering >500mn loss. If they manage to create similar revenue growth in Europe and Asia, it can become a very interesting company though. I also found this statement from their filing very interesting:

"Snap Inc. is a camera company.
We believe that reinventing the camera is our greatest opportunity to improve the way people live and communicate.
Our products empower people to express themselves, live in the moment, learn about the world, and have fun together."

The full filing is here:

Two nice TechCrunch articles with key charts & stats here:

Pokemon Go makes 1bn of overall 35 billion global app revenue in 2016

We all know the app business is very short tail, due to app store structures and bad discovery mechanisms, but also app access on our phones (which is essentially a nicer bookmark list with icons on our smartphone screens). Still waiting for a smartphone notification stream that can be fed by hundreds or thousands of apps, but i am sure the day will come. For now, the short tail is very short - Pokemon go alone took in 1 billion of an estimated overall developer revenue of 35 billion. More interesting numbers from App Annie and Sensor Tower here: 

Jan 26, 2017

Smartphones and health

Maybe the next big thing for smartphones: to greatly contribute to our health. is always a great resource, and the article linked below gives us a good overview or where smartphones can be beneficial to our health (although, for elite athletes smartphones may be the opposite).

Harvard Business Review on collaboration work models

Being kind of an introvert and having the feeling that I am most productive with a switched off phone alone at my desk, I am sometimes positively surprised and sometimes really annoyed when groups of people get together not to dicuss and decide (what meetings should be for), but to actually work and produce together. I think it is a very complicated thing, beginning with hierarchies and team dynamics and ending with practical distribution of workloads. Here are some interesting thoughts from HBR on "collaborative overload":

Uber releases public data platform (a drop in the ocean)

I have always argued that we have one major construction bug in the web: Data that I cause should be mine, and not belonging to the service provider. I can lend this data to the company in order for it to be able to provide better services, but I also should be able to delete it, or take it elsewhere, or publish it if I want. Everyone posted their Spotify data analysis at the end of 2016, but I didn't like it because it showed data that I felt was mine, and that I should be able to take with me to Pandora or elsewhere if I wanted to. Same with my movements: Car2Go and DriveNow have data on my automotive history, Lufthansa and AirBerlin on my flying, DeutscheBahn on my train rides etc. - and Google maps, if you have location tracking on, all of them - but these should be mine. If we all put them together or give them in trusted hands, the public could greatly benefit from it. Uber has tons and tons of data on traffic, and - kudos - has decided to make it public. I guarantuee that no single traffic planning government entity in any city has comparable data to invest our, the tax payer's money, in how to design cities and make traffic more effective and safe. You have to be grateful to Uber for publishing this, but it is also a reminder that there's so much super-useful stuff out there being held away from our societies.

Strategic view on Amazon Echo as an operating system

This article is three weeks old, but I just recently received my Alexa/Amazon Echo product (love it) and had a few thoughts that I may have put together as a bigger blog post. Doing some research, I found this one and it summarizes everything better than I ever could: The "design" of Amazon Echo as a platform and why it has this huge potential:

Micro-Moments: Realtime, hyperlocal matters

I recently discovered the weekly newsletter on and sometimes I can find interesting links to articles on sites I didn't even know. "Street Fight - Inside the Business of Hyperlocal" is one of them. Although I don't really like to declare 2017 "the year of xyz" withing January, and the author might have an interest in doing so, being the CEO of uberall, a location marketing company, the article gives an interesting perspective on "micro-moments" and why they matter. Having had two years of "micro-services" in any IT-discussion, maybe we will see this term more often on the marketing side (side note: at the time of writing this post here, the article had 4 shares on G+, 44 on Facebook and 100 on LinkedIn - interesting spread).

Time to read up on algorithms

If I would have my own "Google Trends"-like analysis of words used in professional conversations and presentations, I guess that "algorithm" (maybe besides "data") would have the steepest growth of all. Amidst the Trump-domination of my newsfeed, algorithm-driven, I stumbled upon this interesting article (pure politics, Trump in perspective) and found it interesting enough to see what else was published on "Los Angeles Review of Books". And here comes one of the most interesting reads of this year so far: An exhaustive examination of algorithms - how we use the term, what they are, and what impact they have. Lengthy article, but worth every minute.

Internet, radio and politics

Fake news and alternative facts and their impact on society and politics are widely discussed. I found an interesting thought about this on meshedsociety, comparing radio and its role in the rise of Nazis in Germany with the internet today. Not very long, but good food for thought:

Interesting statements from Facebook's Head of Advertising

Andrew Bosworth, two weeks ago on AdWeek, was quoted with this:

"Hyperlocalizing the expansive digital network is the next step, he added. In Bosworth's mind, Facebook should help people directly order food, buy tickets to events, find out what to do on a Saturday night and shop until their heart's content straight from the app."

I bleieve that one day Facebook will make more money from transaction provisions than from displaying advertising. But that's a long road. Other than this, the article is a documentation of how FB's platform is too advanced for the industry it serves:

Cars as feature phones

There's a ton of articles out there dealing with the upcoming disruption of the automotive industry, from various angles and perspectives. Besides big data, intelligent co-pilots like this interesting-looking start up from Germany ( and self-driving, cars are stunningly un-connected nowadays. Given that the average car age on German roads is 9.2 years, this should be a home-run for products like this Bosch-powered car/smartphone dongle: But somehow none of these things really take off. An interesting perspective on the situation of cars is to compare them with feature phones, pre-iphone: interesting article by Benedict Evans from Andreessen Horowitz: