CHATBOTS Case study: Goal.com

Andrey Brych
Chatbots bible and More.

--

Yep, I heard it, the last year’s messenger bot hype has vanished, but guys at U.K. football portal Goal.com, are holding their ground and have spent the last 9 month building and polishing their bot, which goal is to build brand awareness.

The Goal.com’s chatbot has 120 thousand subscribers and that number keeps on growing, adding up to 600 subscribers per day. It is interesting that the guys don’t rely on this rather impressive number since it doesn’t mean much (just like App downloads). What they take into account is a number of daily active users, and they impress me even more with this KPI — 17'000 daily engagements. That number didn’t come out of nowhere, the bot team took a constant experimenting approach and increased the number of engagements for the past 6 months.

Author: Last week I wrote an article on chatbot KPI measurement, I suggest reading it after you finish this one.

Their current goal is to expand their reach to non-English speaking countries.

So, what have they learned:

Polish the initial bot intention rather than scale the untested idea

Goal’s first intention was to let users search and automatically receive news on their favorite players, clubs or leagues. Their website has all the information on most players around the world, but their bots initial focus was on 5 major European leagues:

1. La Siga
2. Premier League
3. La Liga
4. Bundesliga
5. Ligue 1

This stacks up to hundreds of teams and thousands of players that need to be manually added. This gives users access to the latest stories about their favorite Lionel Messi news, photos, and stories. Apart from top players, the team added hundreds of less notable players just to make sure the bot had enough information to respond to customers.

Very soon it became obvious that it was a waste of time and energy. They saw that users didn’t use a bot to search players. According to Paul Rayment, users felt more comfortable using Google for that. So they went ahead and integrated the bot with the search API to decrease the amount of time spent scaling the bot database. This ended up in an increase of free time to focus on bot usability.

It didn’t matter how many keywords had the team build into the bot, users refused to use the search function and the team focused on looking for ways to prolong the engagement time.

They made a strategic decision to add shortcut buttons to the most popular subjects at the end of the stories. For example, the “Transfer news” button was added to every single story about transferred players during this season.
After they saw a rise in user engagement, they started digging the same hole even more. Naymar’s transfer was the most expensive one in 2017 with a fee of 222 million Euro. The goal added Naymar quick button to all transfer related stories and saw another 2-time increase in engagement time. They then adopted the same strategy for other big players.

Another stumbling stone was the incorrect name spelling, due to name complexity and rarity. Every time, users incorrectly spelled the name of the desired player the bot couldn’t provide the right news. To fix this problem, the team taught the bot to understand nicknames, like in case with Oxlade-Chamberlain (hope I spelled it right) if the users type Ox, the bot will understand it.

Timing, timing, timing!

Lots of football publications use social platforms to distribute exclusive stories to stimulate readers engagement. Goal conducted an experiment with this as well but used their bot to deliver the exclusive interviews first thing in the morning, according to user time zones. It failed because the information couldn’t be delivered at the time the users expected it, and it didn’t matter if it was exclusive or not.

Conclusion

The Goal bot drives between 4 and 6'000 users to their website every day and if you take the amount of total daily users (claimed to be ~2 million visitors/day) it is a very small percentage. According to Rayment, bot counterweights email newsletters with more personalized information, (i.e if a user cares about Messi only, he will get news on Messi only) and raises the brand awareness in a very unobtrusive way, because users have to subscribe to receive bot messages.

--

--