Top down forecasting for SEO - let's discuss...

First of all, what is a top down forecast? 

A top down forecast is where you would take the overall revenue figure for a client / business you work for then work backwards to eventually get to a sessions figure that would be required to deliver the revenue figure.

How does this type of forecasting differ from what a “regular” forecast may look like?

Generally in my experience you would more commonly see a bottom up forecast – this is where you would take the sessions data as your starting point then work this through whatever forecasting model you would use to reach a revenue figure as your final output.

It’s important to note that there isn’t a right or wrong way to go about this, but this post isn’t here to debate that so let’s get back to looking at why a top down forecast is useful and the process on working through creating one. 

What data do we need?

This is split in to two parts, firstly you need;

• Business revenue target for organic specifically
• Conversion rate for organic
• Average transaction / lead value for organic

 You require business data such as overall revenue target, conversion rate for organic and average transaction / lead value (there are other metrics but let’s keep it simple for now).

 Secondly you need to gather;

• Google search console clicks, you can get these by category for specificity but again let’s keep it top level for now (once you have this data look at the brand and non-brand split %)
• Organic sessions data, ideally for the last 24 months but 12 will suffice if you can’t go back further (data availability will depend on when you set up GA4)

We want as much sessions data over time as possible so that we can get a decent baseline showing:

• What YoY growth has looked like over a period of a couple of years, previous growth is NOT indicative of future performance but this is still useful to see
• If there are any anomalies within the data that don’t match up with regular seasonality for example, if the brand has executed any ATL activity this may have inflated sessions so we need to try and exclude this from our numbers

Ok so we have the data what’s next?

Now we move on to the fun stuff, let’s discuss how we can get to the sessions figure that will deliver against (or ideally exceed) revenue targets.

At this point you may be reading this post thinking “If we start with a revenue figure, how do we get to a sessions figure?”.

Well, we also work this from the top down. I’ll explain and give an example below but first let’s quickly look at what a bottom up forecast would look like.

 I’ve listed the metrics in the order you would build the forecast.

Sessions > conversion rate  > transactions >  average transaction / lead value > revenue

In this instance you take the known which is based on sessions and transactions data that you actually have available to you. The other metrics that follow would also be readily available in GA4 or from the business data you have available.

A bottom up forecast would look like this:

 Revenue > average transaction / lead value > transactions > conversion rate > sessions

The part where the top down forecast differs is that you wouldn’t have a sessions or transactions figures to start with, these would ultimately be your outputs and would be calculated using the above business metrics in the order given.

Calculating the output

Once again I will quickly reference a bottom up calculation to help give context and segue in to the top down.

Let’s assume the following figures for this section just to make life easier. 

The figures assume that we have taken and applied the non-brand split % taken for clicks from Google Search console.

Sessions – 10,000

Conversion rate – 10%

Transactions – 1,000

Average transaction value = £1,000

Revenue = £1,000,000

For a bottom up calculation we would do the following, sessions X conversion rate = conversions.

Then we would calculate transactions X average transaction value = revenue

When looking at this from a top down view, we just take the above but do this in reverse order to get the sessions and transactions output.

Obviously in this instance the sessions and transactions figures would be unknown at this stage because we are working back from a revenue figure.

Revenue / Average transaction value = Transactions (1,000,000 / 1,000 = 1,000) 

Average transaction value / Conversion rate = sessions (£1,000 / 0.10 = 10,000) 

Once you have the above at a top level you can then go nuts and phase this over a 12 month period based on the peaks and troughs of the business. 

(When working from a top down perspective we still need to ensure we are giving as much of a non-brand view as possible, the best way to do this for now would be to work back from the overall revenue figure to get your sessions output, then apply your non-brand % against your sessions figure to give your non-brand sessions. You can then apply your business metrics to the non-brand sessions figure to get your revenue output. for the purposes of this article I've assumed the figures already have the non-brand split applied against them.)

Final considerations

There are 3 key levers that make up any forecast that we need to track and if needed, adjust based on performance:

-         Sessions

-         Conversion rate

-         Average transaction / lead value

Any of the above 3 levers can hugely affect how we track against our forecast and will need to be monitored and accounted for accordingly. 

If for example our conversion rate increases then in theory we would need less sessions to hit revenue target.

Conversely if our conversion rate decreases we will need more sessions to make up for this. 

The above examples would also apply to average transaction value.

This post is a bit of a beast as it is so I will keep it fairly top level and won’t delve in to product specific forecasting, but this allows us to forecast based on better converting or higher value products which can be advantageous, especially when working with a larger more complex business.

However that is for another post, another time.

If forecasting for SEO is a headache for you please feel free to reach out ben.barker@connective3.com or drop me a DM on here.