Forecasting Website Traffic, how much search volume do you need?

Being asked “How much traffic do we need to generate?”, I believed would be a common question for someone in a digital marketing or website management role, but unfortunately it rarely came up in conversation before joining The Royal Mint as Senior Website Manager. I often felt like I was walking through treacle when guiding individuals through the process of modelling and forecasting website sessions, which surprised me as the concept of understanding how many website visitors I would need to meet a revenue, account creation or data capture target intrigued me.

Over the years I’ve read in-depth about time series modelling to help go some way to forecasting website traffic, but I’m far from a statistician, whilst I can use analytics tools to interpret performance and user behaviour, time series modelling needs a level of focus that evades me with a full-time job and young children, I found it hard to commit too. Perhaps I’ll revisit it one day, there are some great articles and an in-depth paper linked below, should you have the time and want to read more.

It was in one of those articles that suggested a lazy version for forecasting over time and it was that word “lazy” that inspired me to write about the process which I’d been using to identify targets for website sessions. Before you start we need to understand the objective of the website, whats the ultimate goal? Is it e-commerce sales, lead generation, subscribers or perhaps just the number of readers? The model I adopted and developed is reliant on being able to measure that ultimate goal and for this article, I’ll reflect on my latest experience in e-commerce where I used and encouraged marketing teams to adopt their thinking and planning around it.

I begin with the end goal firmly in mind, knowing what our sales target or amount of email captured or customer accounts we want to create is essential. In this example that I’ll focus on our sales target, this will be for a hypothetical busy e-commerce website, with a nice clean and simple £100,000 period sales target.

Next, you’ll need to understand your average conversion rate and order value, hopefully, you’ll have Adobe or Google Analytics fully integrated and capturing these data points already. If you don’t have analytics tools across your website, I’d recommend getting these in place to have complete visibility across your website.

Again as this is hypothetical I’ll pluck a 2.5% Conversion Rate and a £90 average order value out of the air, and that’s all we need our target or what we want to achieve our conversion rate and average order value.

Using this equation we’ll see an estimate of the sessions we need to achieve to meet our £100,000 sales target.

(Target Revenue / Average Order Value) / Conversion Rate = Target Sessions

\[ (\text{Target Revenue} \divide \text{Average Order Value}) \divide \text{Conversion Rate}) = \text{Target Sessions} \]

This is just an estimate and a basic one, other variables will affect how the sales target is achieved, but it will provide an idea to help and support marketing teams to understand what traffic they need to drive traffic.

I’ve used this method to apply context to revenue and sales targets for many years and it can be surprisingly accurate when combined with solid averages gathered over a rolling period. I’ll keep that in mind for future posts and elaborate on how I made this model even more accurate through data gathered Google Analytics.

More in-depth reading on forecasting website traffic