The online innovation machine chases after a new buzzword: messenger communication. Instead of sharing, publishers want to send news directly in chat. Spectrm are the chatbot pioneeres. Max Koziolek and his team want to restore web news – and, for a little bit, save online journalism.
The location of the office space where Max Koziolek and his team are based speaks for itself: Potsdamer Platz, Sony Center, the digital heart of Berlin. The tiled counter on the seventh floor is a café and reception in one, and those with some time on their hands can drink a sweet latte macchiato while getting the information they need. Today, Max doesn’t have time for coffee. It’s a cold Friday morning. In an hour he has a meeting with his team, and then he’s off to the central editorial office of the Welt newspaper – by subway, or course.
Digtator: Max, how many Messenger tickers have you already received today?
Max Koziolek: Ten, maybe fifteen. I don’t have a weather app, so I use Poncho, which is a weather bot. The guys behind Poncho are very likable, so Poncho is, too. Apart from that, I’ve got ones from Sportbild, Bild, Curved, and Techcrunch.
Unless they’re sitting in office space above Berlin, most people probably don’t know much about chatbots. What do you actually do?
We make it possible for publishers and brands to send messages to people on Facebook as if they were friends.
You and your co-founders started spectrm in your parents’ house. How did you come up with the idea?
Our co-founder Jendrik and I noticed that we weren’t reading as much news as we used to. We came to the conclusion that it was because publishers are becoming less and less important. More attention should be put on individual journalists. So our original idea was to create a platform for people’s favorite writers.
A curation Web site?
Exactly. But we didn’t have any proof for our theory. We spoke with media representatives at the re:publica trade fair and many of them didn’t believe in this kind of personalization. To check whether our theory had any foundation, we decided to create a prototype – not on our own platform, but on WhatsApp.
Why not on your own platform?
Around that time, I got to know our CTO Manfred at a party. Manfred had spent ten years building up a start-up in Austria that sent weather warnings by text message. So he knew all about distribution. Our prototype was unbelievably successful. However, we noticed that this success wasn’t down to the journalists, but to the method of distribution. Online journalism doesn’t have a content problem. It has a distribution problem.
What do you mean?
A lot of media companies no longer reach their readers because there has been a shift in the channels where people read news. The majority of people in the USA now get their news from Facebook. New publishers are successful thanks to innovative, digital distribution.
You are now using Facebook Messenger. Why is that?
WhatsApp isn’t an open platform. Every WhatsApp service is a hack – an unauthorized solution that can be shut down at any time. We thought to ourselves: “If we can create a hack for WhatsApp, then we can also do it for the Messenger.”
You are Facebook partners, and your office space in the Sony Center in Berlin is directly above Facebook Germany. How did that come about?
Our first Messenger project was the BILD football transfer news ticker. We created a Messenger group to test the ticker. At some point, a Facebook employee signed up. The next day, several more had registered. We googled the users who had joined our group. And as soon as people from Silicon Valley started to join, it was clear they weren’t interested in the news at Werder Bremen football club.
You say yourself that personalization is a matter of opinion. But your users receive personalized messages nevertheless.
There are two kinds of personalization. On the one hand, there’s personalization from the author. There are bloggers who use spectrm to customize content. But what we actually mean is personalization from the reader’s point of view: Perhaps they have to go to work early and want to read the news at 7 instead of 8 am. Perhaps they’re interested in technology and sport but don’t care about politics, in which case most of the content they get should be on those topics.
In other words, they don’t read all the news that’s relevant, but only the stories that they like personally. Isn’t that the same thing that people are criticizing about Facebook?
We also have to provide users with content that they haven’t signed up to. Content that they wouldn’t normally click on but which, using information from other conversations, we know could be relevant. We call that “random exposure.” Here, the goal is to create a similar feeling to when you open up a newspaper and “stumble” across an article.
And for that you use the conversations your bots have with other users?
Exactly. For example, the bot asks: “What did you think about the article?” The user gives a positive or negative answer, and a bot is good at understanding the two contrasting responses. What’s more, we also get very valuable feedback for personalization. Facebook uses metadata. We ask the readers directly.
The problem is that every news site tries to attract readers with the same things. That’s why the Spiegel, Zeit, and FAZ homepages sometimes look so similar that you can easily forget whose site is whose. The bot, however, can use the data it collects to establish what the user knows. It gives publishers the possibility to consider the best way to prepare content. Sometimes, that not only means a 5,000-character long-format article, but also perhaps a video or interactive story.
How will the relationship between Messaging services and journalism evolve?
Messaging will become one of the major distribution channels. It isn’t comparable to email or direct traffic, because the messages are tailored exactly to each user. It will make reputable and political journalism more relevant again.
“Social” distribution platforms prefer entertainment because it is harmless and easy to read. Whenever you share a critical article on Facebook, you always have to be prepared for someone to disagree with it. People simply feel less comfortable about that and it involves more effort than sharing cat content. Messaging services can help publishers find loyal and regular users for good content. It is the digital subscription of the future.
Facebook is already trying to move in this direction with its news offerings.
But a social network that focuses on news rather than contacts will lose its appeal in the long term. Facebook knows that. That’s why it wants to sell the Messenger as an alternative to publishers: as a digital subscription. We are already seeing this development in China with WeChat. Some 80 percent of all business accounts operate in the information sector. Users can add credit-card details to their account and subscribe to the publishers through Messenger.
Talking about payment: Which industries profit from communication via Messenger?
I see three areas where bots are a benefit. First, customer service and customer support. Most companies in this sector work with templates, because they always get the same queries: “What happens if I haven’t downloaded my ticket?” or “Where is the next store?” That’s not only a silly job, but also one that companies spend an incredible amount of money on. It therefore makes sense to transfer these templates to a bot and automate this support.
The second area is e-commerce. More than 60 percent of traffic in online shops comes from mobile devices, but only 20 percent of users actually use these devices to place an order. That means a large number of potential customers stop at some point during the order process. It’s no fun having to enter your name, address, and credit card number on your cell phone. Bots provide a much more pleasant experience: I simply say what I need and the bot recommends various products. If there’s one I like, I can buy it with a single click. Facebook is already going in this direction: Facebook’s Vice President of Messaging Products, David Marcus, used to be the president of PayPal. So they have someone who knows how payments work.
The third sector where bots can be used is content and marketing. Messaging services provide completely new and exciting possibilities when it comes to storytelling. The interaction between the user and the bot can be very smooth. We designed a user journey for Universal Music that was around 10 minutes long. Almost all users went through this journey right to the end. There is no platform that has such high engagement rates.
Is that because the communication comes across as more human?
Yes, there’s that. And there’s also the fact that the major instant messaging services have more users in total than the major social network platforms. So we go where the users are and address them in the same way they communicate with other people: with language, one to one.
You are a German start-up, and are also off to a flying start in the USA. How come you are so successful overseas?
On the one hand, the majority of people in the US use Messenger instead of WhatsApp. But far more important is this: the publishers there have a better understanding of the digital world. In Germany, the only publishers that can afford to pay for digital strategies are the major ones. The local village newspaper has to go without. In the USA, however, it’s the other way round: the market is unbelievably competitive. Things move at a completely different pace. Whenever people see a new platform, it takes two days before they say: “okay, we have to try that out.” Then, the decision processes are fast, and off they go.
Who benefits the most from Messenger from a journalistic perspective?I really hope it’s the local papers, as it will finally enable them to reach their readers again. A story about a vandalized local bus stop isn’t going to reach any of these readers on Facebook. There simply isn’t enough interaction with that kind of article for it to be picked up by Facebook’s news-feed algorithm. Still, this is news that could be of interest to anyone living locally. So Messenger would be a good solution for local journalists, as they could build up a number of subscribers without being dependent on Facebook algorithms.
It is becoming increasingly more difficult to tell bots and real users apart. We saw that to extremes on Twitter during the US election campaign, and even the AfD party isn’t ruling out using social bots during the election campaign in Germany.
We have to differentiate between two types of bot: social bots and chatbots. The social bots we saw during the US election run on platforms. They are intentionally made to lack transparency, as they don’t want to be exposed as bots. We, on the other hand, are developing chatbots that function via a conversational interface like Facebook Messenger. These bots are transparent. My Messenger contact is the company. And while we can create conversations that are very realistic and work well within a certain framework, nobody looking for a good conversation will turn to a bot in the near future.
Are we losing interpretive authority on the Internet?
There are figures that show how the majority of Internet users today aren’t actually people. However, I believe that hate speech and social bots are phenomena that, in two years’ time, will no longer play a role on the Internet. Google’s AI can already see with a probability of 60 to 70 percent if a comment contains hate speech and whether it should be deleted. The same will happen with social bots. There will no longer be political correctness but rather AI correctness.
So you aren’t worried that at some point you won’t be able to tell who is a bot and who isn’t?I find it exciting, as it makes us think differently about human communication. What makes our communication unique? Imagine you could transfer the entire conversations of someone who has died prematurely to a chatbot, and their friends could simply carry on chatting with them. This throws up a lot of questions, especially ethical ones. Perhaps we’ll have to think about chatbots that, now and again, will say, “Hey, I’m just a robot. If you have a problem, ask your doctor or pharmacist.”
Interview: Simon Schaffhöfer
Translation: Toby Skingsley
Photos: Markus Braumann