Based on anonymous usage data collected at Stinky Teddy we have written a paper entitled A Measurement of the Social Media Impulse Response Function. This paper is of interest to Internet startups, public relations professionals, and anyone interested in the new web-blog-social media information sharing ecosystem.
Like most alternative search engines, Stinky Teddy doesn't get much traffic. On an average day we get a few hundred searches on our site (Google handles about 1 billion searches per day worldwide). It doesn't help that our advertising, marketing, and public relations budget is $0. This is not strictly true - we once spent $40 on a Facebook advertising campaign, but that experience warrants a separate blog post.
We do, however, get an occassional surge of traffic. Somebody writes an article about us on their blog and we get a bunch of people checking out the site. Our first surge of traffic came in October, when Frederic Lardinois wrote a short piece on ReadWriteWeb entitled Stinky Teddy: A Cool Real-Time Search Engine with a Rather Odd Name. We didn't know the article was coming, and only noticed that it had been posted when our site crashed (we had a memory leak, since fixed). Before this ReadWriteWeb article we got no traffic whatsover as we had not yet released the product.
Being a scientist, I couldn't help but to utilize this ReadWriteWeb post as a chance to do an interesting study. The Internet Entrepreneur's dream scenario is the following:
- Build Great Product in Secrecy
- Using PR, generate massive news coverage on day of launch
- Go viral, with peer-to-peer messaging on social media leading to massive adoption.
This scenario never works for new search engines. Nonetheless, there are a bunch of people out there willing to try a new search engine, and positive news coverage is the way to get on their radar screen.
When it comes to planning for this glorious launch, however, there is one question that the Internet Entrepreneur wants answered that nobody will tell them. How much traffic will I get, and how long will it last?. The study I performed using the ReadWriteWeb post addresses the "how long" question.
The basic premise behind the study was that this single ReadWriteWeb post was singularly responsible for all traffic on Stinky Teddy for the next month. Our traffic before was nil, and we did absolutely no marketing or PR during this period. Therefore, any visitor or user on our site during that period was directly or indirectly related to the ReadWriteWeb post. For the first time, we were able to measure the "Impulse Response Function" of the web-blog-social media ecosystem. The "Impulse Response" is how a system responds to a sharp input signal (for a detailed discussion, read the paper). In this study, we measured the hourly/daily traffic on our site. That is the data we need to determine the impulse response. Here's the traffic in the 100 hours after publication:
This shows an interesting two-peak structure to the traffic. The first peak is obviously direct traffic from the ReadWriteWeb blog. We suspect the second peak is due to social media (e.g. Twitter sharing) and news readers (Google Reader, Netvibes, etc.). The second peak corresponds to 9 AM on the East Coast of the United States, so these are people checking yesterday's news when they arrive at work the next morning. We also looked at traffic on Stinky Teddy for the next 25 days:
Here we see something very interesting. In the web-blog-social media ecosystem stories "ring" for a long time. Half the traffic attributable to the ReadWriteWeb article came more that 4 days after the article. Only 10% came during that initial 5 hour burst from the ReadWriteWeb page.
This is a one-time only experiment. We've had several other momentary spikes in traffic, but only for this period in October through Novembmer could we definitively attribute all of the traffic back to a single source. It would be interesting of others repeated this study to see if what we observe is universal. Our main findings are:
- Only 10% of traffic eventually generated by the blog post came via early direct clickthroughs from the ReadWriteWeb home page.
- There is a two-peak structure in the traffic during the first 24 hours, with the second peak likely associated with "first thing in the morning" readers of yesterday's news through social media sharing or readers.
- Half of the traffic (from both direct and indirect sources) came four or more days after the article was posted.
Please read the paper
and leave your thoughts below.
Posted by David Hardtke
@ 02:37 PM PST