Posted in Articles By Sam Applegate
However, how do we use this data to generate decent predictions on how much traffic we might receive at different ranking positions?
There's no exact way to do this, as every single search term is different. But, this is how I go about it.
Plot a Ranking Distribution
Let's follow on from the example I used in my last post on search volume accuracy, and look at the [web design bath] keyword.
According to the Adwords Keyword Tool, the approximate 12-month average number of search terms matching this keyword is 720. If we want to look at specific months, we can download the trending data, but this keyword doesn't have great variance over the year, so the 12-month average will do us for now.
I want to break down the search volume into each ranking position. We can make a start at this by using some data from an optify organic CTR study performed last year, and applying an error margin of 25%. It's a crude way to start our prediction, but by simply multiplying our search volume by these numbers, we can get an idea of the actual traffic we might see, should our website reach these positions.
|Rank||Average CTR||Lower CTR||Upper CTR||Monthly Traffic|
|1||36.4%||27.3%||45.5%||197 to 328|
|2||12.5%||9.4%||15.6%||68 to 113|
|3||9.5%||7.1%||11.9%||51 to 86|
|4||7.9%||5.9%||9.9%||43 to 71|
|5||6.1%||4.6%||7.6%||33 to 55|
|6||4.1%||3.1%||5.1%||22 to 37|
|7||3.8%||2.9%||4.8%||21 to 34|
|8||3.5%||2.6%||4.4%||19 to 32|
|9||3.0%||2.3%||3.8%||16 to 27|
|10||2.2%||1.7%||2.8%||12 to 20|
How many times have you worked hard to reach the top spot, only to find a miserable amount of traffic compared to what you were expecting? This is because you probably overlooked certain factors in your estimation.
Analyse the Texture of the Search Results
The guess work starts here, this can be finger in the air stuff at times, but you need to take a long hard look at the texture of the search results. What do I mean by this? Well, basically look for things which might cause variance in our simplified traffic vs rank distribution.
PPC Adverts - Is the search page full of them? Or is there just one lone advert on the right hand side? Are the adverts well optimised and compelling? Ads are going to skim traffic away from the organic results.
Strong Brands - Are there any major brands in the search results? These will probably attract a higher percentage of the traffic.
Videos/Images/Maps - Keyword dependant, but do some videos, images or map locations appear in the results? This may result in an irregular distribution of the traffic.
Rich Snippets - I talked before about how rich snippets can be used to increase CTR. Look for these in the search results as they can attract a higher CTR.
Snippet Text & Title tags - If the competitors have relevant meta descriptions, these may appear in the snippet. Together with the title tags, they can attract a higher CTR if well written.
Site Links - Any websites appearing in the SERPs which contain sitelinks may attract a higher proportion of the traffic.
Obviously, the less of this stuff there is, the better. Let's look at the texture of the search results for [web design bath].
At first glance, the texture looks pretty standard. I would expect to see this page fit closely to the general traffic distribution plotted above. There are no major brands (that I am aware of), no rich snippets and just one site link.
The only problem for us, is that there are plenty of well written adverts. It's hard to say exactly, but expect to see our traffic numbers reduce by about 5%-10%.
Looking at a Measurable Example
Now I'll show you an example where I can reveal the actual traffic numbers. I have a website which ranks at #1 for the [lawn mower review] keyword. Before I reveal the answer, let's see if we can predict how much traffic there is at this position.
As you would expect with this keyword, the search volume is very seasonal (this makes sense, as more people mow their lawns in the summer months). Therefore, using the 12-month average search volume data would result in inaccurate monthly predictions. We have to dig deeper and fetch the search trending information.
Using the Keywords Adword Tool I have fetched the following data. If you don't know how to do this, take a look here.
Now I'll predict the traffic for a website ranked #1 for the last 3 months, using the method I outlined previously. For a #1 ranking position, we assume a CTR range of 27.3% to 45.5% (using a 25% margin of error):
|Month||Search Volume||Monthly Traffic (at #1)|
|January||28||8 to 13|
|February||58||16 to 26|
|March||140||38 to 64|
I next take a look at the texture of the search results, and look out for anything which might affect the distribution of traffic.
The first thing to point out is that a major brand is appearing in the #2 and #3 positions - which.co.uk. I predict these results will have a higher than average CTR for their position. This, along with the high number of PPC adverts, means I believe my actual traffic will be towards the lower end of my estimated range.
So, Was I Right?
The moment of truth. I load up Google Analytics and check how much traffic I actually received on my site, from the [lawn mower review] keyword during these three months. First looking at January:
|Month||Estimated Traffic||Actual Traffic|
|January||8 to 13||11|
|February||16 to 26||21|
|March||38 to 64||66|
Not bad. The presence of the PPC adverts and branded result didn't seem to affect my traffic too much. In March my traffic was even slightly higher than I would have predicted. I can put this down to my exact match domain (EMD), which no doubt helped my CTR for this particular search query.
So remember to always consider the texture of the search results page before coming up with any traffic predictions. Also remember that the distribution can vary wildly from the standard 36%-12%-9% pattern above. You should usually include a margin of error with your numbers and always prepare for the worst. It's better to underestimate these values (especially if you're working for a client) and be pleasantly surprised, than to get your hopes up with massive amounts of traffic, only to appear to under deliver.
Want to hear more from me? Subscribe!