Referral stage experiments are focused on creating viral content or working on processes, which encourage your users to tell their friends and colleagues about you.
The importance of working on virality is to spend less on acquiring new clients eventually. Good virality can reduce the need for paid traffic. But if you'd be able to create a viral loop that would encourage your existing users to invite to your business more than 1 new customer on average, this need might be completely thrown away.
There are three types of viral processes: word of mouth, inherent, and artificial. The last two processes are quite similar.
Word of mouth is the hardest one to track, but it’s super effective - because people are more likely to listen to the views of their trusted ones. Inherent virality, on the other hand, is easy to track and implement, since it’s built-in in the product - and it’s used when users share something from the product with others (i.e. a link to Google Docs). And artificial virality offers users some rewards for spreading the word - for example, extra storage in Dropbox for inviting others.
Word of mouth is primarily about the social sharing and customers from social networks. For example, the experiment saying to interview the client about his journey and write a blog post with pictures (that would be interesting enough for other people to share) isn’t straightforward. So for it to get started, its expected effect and confidence in it should be at a high level. Okay, but first, how can we measure this experiment?
The simplest solution is to count how many people have read the article. To do this, we only need to specify a page and the metric called Users.
But this metric is really “too vanity”. It’s unlikely we could gain any insights from it. But if you have a social plugin for Google Analytics connected, you can draw a line in the sand, for example - the number of social actions - tweets.
But it’s far more interesting how much of new social traffic this article leads.
And even more interesting, how many sign-ups or sales the article brings.
You can also calculate the revenue from users that came from the article and paid.
Remember, it’s important to accurately determine the measurable goal of experiments already at the stage of hypotheses scoring.
Here’s an example of built-in virality in Tom’s product. Each tour contains a map that is easy to share in social networks. People read travelers’ posts, click the link to view the map, and sign up for the service as well. Turns out to be a Viral Loop:
1. Paying customer shares his travel map in social networks.
2. His friends like and share the post.
3. His friends and friends of his friends click on the link.
4. They sign up for the product.
5. They buy the product.
Tom's team knows that their users usually share the map at the beginning or the end of the journey. The team decided that they can increase the number of map shares if they let users indicate their current position on the map and the percentage of the road traveled. Then customers will share the map several times during the trip, telling more friends about it.
Though the task isn’t easy to implement, the team expects it to have a great impact on the metric - the number of map shares increased at least two times. If this action is tracked by, let’s say, TripMapSharing event, measuring the success of the experiment is simple.
If the link from the post is tagged with UTM, you can measure how it will increase the flow of new users for this campaign.
Or even how the number of sign-ups will increase.
If you are experimenting with a viral loop, it’s important to break it down into steps and improve each step separately.