In the first session I attended on Day 2 of Web 2.0 Expo 2010, Dropbox and Xobni shared their stories and lessons from launch. It was a really good presentation (Adam Smith from Xobni posted the slides on his blog post, linked above).
A lot of the advice is stuff we’ve all heard before: make sure you have a great product, test a lot, get user feedback, create scarcity to build buzz, be responsive and bold, focus on doing the critical few really, really well. Even if that advice has been beaten to death, I still find startup success stories inspiring.
One thing that caught my attention was an almost throw-away statement that Adam (I think it was him) made at the end. He said that users they acquired through PPC didn’t convert to paying at a very high rate, but they tended to refer a lot of customers who did. He also mentioned that Zynga measures referrals through Facebook as well. The interesting thing is not that the referrals happen, but that they measure it and account for it in the ROI analysis of their marketing campaigns.
I might be reading into his really brief comment, but if it’s true, that’s definitely an area where I could learn a lot more.
At Homestead and mostly at Intuit, we only do the first level accounting in measuring the ROI of campaigns. That is, we count the number of users who are directly attributable to the campaign. It’s definitely easier measuring users by their propensity to sign up/purchase than measuring users by their propensity to refer others. All referred sign ups are really just considered gravy. This first level of ROI accounting has been really successful for us as it still continues to scale. But the question remains: are we leaving money on the table?
There are a lot of questions in the details of how you’d go about doing this “second level accounting”:
- What are the best ways to track a second level user back to the referring user and then back to the originating lead gen source?
- Assuming we would track using links of some sort, should we try to account for “offline” referrals (using untracked links)?
- What is a reasonable time frame to count the referrals from a first level user?
- Does a first level user get any credit for third level signups?
- Do you account for the speed of the referral? For instance, you could credit the first level user with all $ collected from the second level user within 12 months of the first level user’s signup, thus rewarding those who refer faster.
I’m sure there’s more.
If we can figure this out with any reasonable amount of accuracy, this could significantly change the economics of how we do acquisition marketing right now.
Ramblings of a Short Man 
Hehe, I’ve been using Xobni on and off for the past couple years. I forget if I found it via paid link but I didn’t pay for the premium features, yet did recommend it to others. I believe I did this more to use these others as guinea pigs, not because I was espousing any benefits of Xobni. In fact, ultimately, it tended to slow my outlook down and the insights if had weren’t very satisfying…so I no longer use it.
If I recall correctly, Xobni had a built in recommendation feature where you enter your friends’ emails and it sends them some kind of invitation. I would hazard to guess that this is where they are getting their numbers. I know Facebook loves to get people to recommend things to their friends as well.
The fact that the people who were recommended something by friends were more likely to convert than people who found some service or product through a paid advertisement makes sense to me…certainly there are issues of trust and personal relationship at play. The person I “recommended” Xobni to actually apologized to me when they uninstalled it (for the same reasons I stopped using it), as if I would be offended on a personal level.
Of course, this does not answer any of the “how to measure secondary referrals” question…the lesson here is don’t use Xobni.