If someone asked me why I loved online marketing so much I would probably have to say the analytical nature of it and the fact I can see so much detail around what I am doing. I love the fact that at the end of the day I can tell someone my work has generated £x or 5000 incremental visitors, whilst the traditional marketers in a dying industry come up with some ropey figures to support a print advert or billboard.
But how accurate is the data you see from your analytics in terms of PPC revenue or other tracked online marketing activities? I would argue if you use a standard out of the box 30 day cookie…..not very!
Why Do I Think This?
Most web analytics suites such as Omniture and Google Analytics come with a standard cookie, which tracks recurring purchases for 30 days. The vast majority of advertisers either never change this or unbelievably justify increasing it to 60 or 90 days.
Here is an illustration to try and highlight the problems of using a recurring 30 day cookie with the green representing a customer’s journey through search, life and the universe.
In my example this is a customer who regularly purchases from an online sports retailer called just-t.com, they are looking for some trainers so they broadly search on that term. They spot a PPC advert for just-t.com and as a result of knowing the brand they decide to click on the advert. They do some searching on the site, but they can’t find the product they want, so they abandon the website carrying with them a treasured just-t.com 30 day cookie.
7 days later they wander into a high street shop and find a perfect pair of trainers, which they purchase. 24 days after their original visit to just-t.com they navigate directly to the website and decide to purchase a t-shirt highlighted on the homepage. The 30 day cookie is still present on their machine, so the conversion is attributed to PPC, and the keyword trainers.
Naturally the PPC team is happy at their work, another conversion against trainers HURRAH! Management see the latest PPC reports 5 days later and are pleased with the excellent levels of ROI visible. They sanction an increase in investment. Little do they know that the figures they are seeing are fundamentally flawed.
This is just one of many possible scenarios, but it helps to highlight even at a basic level how ropey PPC reporting can be. This example doesn’t even consider that this same customer could go on to buy 10 more items from just-t.com after the t-shirt purchase and providing they are within the 30 day active cookie period everyone of those sales will be attributed to the keyword “trainers”. In reality the PPC advert failed to convert the customer on the product they wanted, but as far as management are concerned it drove additional orders.
What is the Solution?
Unfortunately I can’t give you the perfect solution I can only recommend a more robust solution that will give you a more realistic summary of how successful your PPC activities are.
Whilst I criticise those who use 30 day+ recurring cookies, I do appreciate that in many cases customers will convert after 7 days for example because of the PPC advert, but in the main I think the negatives outweigh the positives.
So here are my suggestions…..
Reduce your active cookie period
Pretty simple really, if your organisation looks at PPC success based on ROAS then reduce your cookie period to 1 or 2 days and let’s see just how successful it is at driving short term incremental orders. Forget the folk who claim industry averages, they are just protecting their jobs!
By all means have a broader level campaign that just focuses on generics for branding, but you should have campaigns focused at long tail conversion terms right? These should be using your short term cookie.
Expire on conversion cookie
Recurring conversion cookies are baloney, dump them and set your cookie to expire on conversion. Track the customer’s first purchase to the PPC keyword, not their subsequent orders. Yes, again it will reduce your numbers drastically, but better to provide a more robust picture of return so your management can make informed budget decisions rather than providing numbers that have been falsely enhanced.
Double drop cookies
This is my favourite solution, but not everyone will be able to do it depending on their analytics provider. I know Omniture although it through multiple eVars, but I can’t comment on all of the other providers.
The basic idea is that whenever a user visits your site through a PPC advert you drop two tracking cookies.
Cookie 1 will track for 1 day, whilst cookie 2 will track for 30 days. This way you can use the 1 day cookie to really see the value of PPC at driving incremental sales, whilst the 30 day cookie will allow you to keep an eye on the longer term conversion tail. I personally wouldn’t use the 30 day cookie reports for senior management, but I would look to bash it against customer data to get a feel for new and existing customers driven.
Any way, these are just more thoughts and frustrations on the area of PPC reporting in particular, but it does ring true for plenty of other online marketing campaigns which use the same flawed methodology. Ultimately if your PPC reports say you increased sales by 100%, but your total web sales are flat…..what does that tell you?
Interested to hear any other ideas about improving the robustness of online reporting or simply criticism of my ideas! Thanks for reading, please share.
{ 6 comments… read them below or add one }
CoreMetrics (I think) allows you to setup complex rules to ‘decay’ or proportion attribution – it is quite amazing how often various sources will claim a single sale, this can include newsletters, organic, branding and of course PPC, plus display and affiliate … It is possible and even frequent that two or three will have an impact on a sale. It is almost too complicated for any marketing department to attribute fairly! Many ‘display’ companies also claim that seeing adverts on the web is like seeing adverts on TV, so they will use 3rd party cookies (Atlas etc..) to claim attribution…
Glad I don’t manage advertising budgets!
Hi Gerry, thanks for your comments. You’re absolutely right, most marketing vehicles are challenging each other for the same conversion. I am always left frustrated mainly around internal campaign reporting, for example customer clicks on a merchandising banner, and buys something. Business reports benefit against merchandising, but have no idea what the customer actually bought. Internal campaigns are always targeting specific products or series, so unless a customer buys that product then the marketing team shouldn’t be claiming a successful conversion.
Most organisations just need to recruit more analytical folk who are less concerned with justifying their jobs, and more concerned with giving an accurate picture of what is going on.
I agree with alot of the sentiments your mention however there are a few instances where i believe paid search should be rewarded. Adopting the current Google analytics structure. Basically if somebody accesses the website through a paid search keyword for example: ‘garden water features’. They look around the site, love what they see but decide to shop around before making a purchase.
Upon shopping around they dont find a product as good as the one accessed through the website and decide in their minds they should return to the original website. They remember the name of the original website however they didnt bookmark this page. They make a search on google under the Brand name and access the website through the organic listing which will appear top under the brand name. They then buy the product. This search type would be attributed to ‘organic’ revenue source or too ‘brand’ revenue source.
Although the conversion was made through accessing the route domain, it wouldnt have occured if it wasnt for the paid search exposure previously. Its becomming increasingly common (and annoying at the same time) that the general public and those less web savvy individuals use Google for everything including typing the full URL’s into the toolbar. If they would of come directly to the website then the credit would be with the paid search however through entering the website through organic listings it cancels out the cookie already placed within the paid search.
This for me is the flipside to your excellent arguement. There are times when paid search keywords should be attributed to the results for aquiring new customers and revenue however they are placed upon the organic results.
Just a different side to the argument ive come across recently with freelancing for a client.
Thanks for your comments, though I don’t believe an organic click would ever steal the paid conversion because usually an organic referral wouldn’t over write a paid search cookie. Usually when a customer arrives from organic a cookie is dropped, but it will be a site cookie, not a typical campaign based cookie. You would probably just double count the conversion….once to organic, once to paid.
Unless of course you included campaign tracking codes on the end of all your site URL’s, but I can’t image anyone doing something so dumb.
Some very valid points.
Although AdWords does now allow you to see ‘Conversions (one per click)’ – so you don’t have to attribute multiple conversions to one cookie (as I understand it)..
Hey Jordan,
Thanks for the comment, I will be sure to look into that and modify the post accordingly if that is the case. I’m looking forward to seeing the latest updates to GA rolled out shortly. The new goals section is a massive improvement!
Cheers,
Ben