Analysing Conversions: Time Data – When Enough is Enough
Posted on: March 24, 2013
Running an Adwords campaign without conversion tracking is kind of like going around a race track without being timed. Yeh, you can run it, even be a bit of fun, but if you are truly serious about it you measure your performance by your lap times.. or your conversions.
I’m not going to go into detail about how to set up a conversion or anything like that, I’m going to talk about analysing the data and to keep things simple, I’ll be using a sale as a conversion and a target cost per acquisition of $10. Nice round numbers and an easy target.
The most obvious thing to do is for a campaign manager to look at the conversion costs and then make a decision based on cost per conversion per keyword or adgroup. That’s fine and of course, best practice. So as an example, we find a keyword converting at $30, which is considerably above our CPA target. Naturally we then go and reduce the bid of that keyword by 10%… run it a bit.. then edit accordingly. Simple stuff and provides great value to the client.
But is that it? A lot of the time, especially with smaller campaigns, the answer could be yes. However lets take a different scenario. Let us now assume that your target CPA is still $10 but your a running quite a large campaign with 1000+ keywords. If 500 words are converting you are still left with a considerable number of keywords that are yet to convert or convert well over that $10 CPA target.
Now I just want to look at those keywords with 0 conversions. There may be a good chance that over 2-3 months there are a fair few words that are yet to convert. Some of these may even have only minimal spend appended to them. What do you do when say, 100 keywords have spent $7 each over 3 months? That is $700 without a conversion that can distort your figures. What is the plan of action? You cannot simply dismiss the keywords as they have not broken the $10 CPA thresh hold to positively conclude that they are bad performers. But at the same time, the culmination of these keywords brings down the CPA performance of the campaign as a whole.
What is the right way forward? Do you give these keywords more time and continue to push through until you have deemed each and every one of them as statistically relevant enough to decide whether they stay or go? A few years ago I would have said yes of course, and even today, when we are dealing with longer term campaigns and bigger budgets then yes, let’s continue to give them a chance to prove themselves.
However over the years, working on hundreds of Adwords campaigns in different phases of development, I have drawn my own conclusions. This is what I call, Time Data. Time Data does away with a purely Spend VS Conversion analysis and brings common sense and judgement back into play. In simple terms the theory behind it is this:
If it hasn’t converted yet… then it’s probably not going to convert
This sounds very low brow but it is really that simple. Why would you continue to do the same thing over and over again and expect a different result? It’s Insane (did you like what I did there??).
Where do you draw the line and decide on using Time Data? I use a really simple equation:
x = Conversion CPA x30
y = 3 month moving total of non converting words.
If y > x then remove KEYWORDS to the value of x starting from highest to lowest.
(note that this is assuming the non-performing words have had a full 3 months of being run)
As an example, I have a campaign with a monthly spend of 1k working towards a CPA of $10 using 100 keywords.
3 months is a guide which you should look at having a think about, there are no hard and fast rules, its common sense. If you are tight on budget and ROI requirements are pressing, then try 1 month. If you have some lee way in time and budget then 3 months, or possibly even more, if fine.
The key is to not simply remove keywords and that’s it for your campaign. You should be testing new variations regularly, using search terms as your basis of building out high quality, target keywords. Once you have exhausted this avenue, move onto thinking of broad terms that you may not have exploited in the early stages of the campaign.
Remember to find a way to separate the new keywords from the old ones. Set up new campaigns or adgroups clearly labelling or distinguishing the age of keyword to avoid a mix up which can be time consuming trying to decipher.
If you have your own methods of using information other than purely CPA and conversion data then I would really like to hear it. Also let me know what you think of the equation above and how you would look at analysing it as that’s just my take, you may know of a different, possibly more effective method.
William Bakhos is an SEO and Adwords professional and has been implementing best practice online marketing initiatives since 2004. Along with consulting to National and International companies and digital teams, William is the founder of the Web Marketers Crew, a Sydney based optimisation and online strategy company.