When it comes to trying to make sense of website traffic, the thought of extrapolating numbers drives many site owners insane. Traffic is naturally bumpy, and may double one month only to drop by 40% the next.
In this post I’ll discuss how to evaluate traffic growth for your website using two key principles: long time horizon and linear extrapolation.
Take a Long Time Horizon
When trying to evaluate traffic growth, it’s extremely important to take a long-term perspective. If your site is less than a year old, this becomes a nearly impossible task, and you can’t rely on data accrued during the first 6 months or so anyway.
Try to use a minimum of a 6 month trailing growth percentage, but it’s better if you have a year. “Stepping back” like this will allow you to put your growth in a broader context, and you’re less subject to making false assumptions based on short-term spikes.
I think it’s important to take a conservative approach to extrapolating traffic. If I’m evaluating a site I want to make sure that I don’t set expectations too high and then be disappointed (or risk losing my investment) when my projects don’t pan out.
When it comes to traffic, I achieve this by always taking a linear approach to my estimates. While many websites do have greater than linear growth (such as NlogN, quadratic, or even exponential growth) I don’t want to bet on the fact that it’s going to have that level of growth.
In general, it’s better to under-promise and over-deliver. Expect a base hit, but hope for a grand slam.
That means that even if the traffic numbers look like they’re starting to take an exponential turn (up 10% 1 month, then up 20% the next, then up 40% the next) I want to be more conservative.
I usually take an average of a long-term growth (a year) with a medium term growth (2-3 months) to come up with a decent percentage to work with. Then I assume that the site is going to average that growth over time.
As you try to evaluate your traffic, remember that website traffic is notoriously difficult to predict, especially when the site is young. Use any numbers you develop with a grain of salt, and don’t over-rely on your predictions.
If you want to be extra-conservative, it’s probably a good idea to round down any growth numbers, sometimes by as much as 20-30%. This provides a buffer to ensure you don’t have unrealistic expectations for your site.