Croud’s co-founder and Director of Product and Strategy, Ben talks us through the latest in search attribution following Google’s recent announcement.
Last click has been dead for many years when viewing Cross Channel attribution – the latest analysis from eConsultancy stating that only 20% of advertisers are still using last click analysis. The reasons and rate of change has been in my experience fundamentally due to the huge effort required to change client systems while investing in people and/or tech to make this happen. While barriers to change have been high we are finally getting to a much better place for Cross Channel analysis – this has been helped of course by things like Google’s Multi Channel Funnels in GA
This week Google announced it’s latest Attribution offering – hence my opening statement is last click in search about to die?
I have started with a quote from Qubit
Why do we need attribution?
Imagine if every year when your birthday came around, a whole group of people banded together to buy you a present. Would you thank them all equally, thank the person who gave it to you, or try and work out who actually put the effort in to try and make you happy? Now – friendship groups don’t work on the same principles as businesses do; sometimes the people who contribute the least might mean the most to you. But when it comes to profit margins, it is of paramount importance to figure out who is actually driving the value to your business.
Consider this journey: a consumer Googles your brand and lands on your site through an affiliate link or an industry blog. A couple of days later, they’re on their phone and click through to your website. Later that evening, they’re on their laptop, they see your AdWords ad and click on it. Once they’ve made a purchase from your site, which channel do you reward between these three? Businesses need attribution so they can figure out how to allocate their marketing spend.
Google’s offering this time is within the PPC user journey where last click analysis still remains the champion.
Specific click analysis has been available by all the major search technologies for many years – BidBuddy (now gone the way of the dodo) launched their offering back in 2005 and subsequently this, to me was the natural process of optimisation and offers many benefits to both performance and analysis
Google’s offering is shown – 5 models of attribution now available without 3rd party tech
» The customer journey to conversion is increasingly complex. On this journey, customers can interact with many different ads.
» Most advertisers, however, give full credit for the conversion to the last click. In doing so, they undervalue other pivotal touch-points in the customer journey.
» Attribution helps you to better understand the contribution of each keyword on the conversion path. You can optimize accordingly and generate more conversions as a result.
Within Adwords search funnels we can run all user paths simultaneously and analyse results based on each model – not just the chosen path.
The benefits of rewarding keywords that generated the first click in the user journey rather than just the keyword that converted is more than significant. You for example, will find that for a client like Guardian Jobs that the keyword, “jobs” just generated an extra 1,000 sales last month that you had been attributing to the keyword “guardian jobs” – your frantic activity of reducing visibility and CPCs on the term has been fundamentally wrong. Google clearly know this as rewarding the high cost generics will certainly not hurt their purse but for clients it will give them very accurate insight into performance and what is actually working a win win for both parties!
Examine the “Path length” report; this shows you how many clicks it takes for your customers to convert
The data in the Search Funnels Attribution Modeling Tool (SF AMT) is based on fractional attribution, where a single conversion gets distributed across the conversion path
An example of “Fractional” attribution means that is a user on the liner model that had 5 clicks before purchase and made a sale of £50. In this model each click would be given 0.2 of a sale and a sales value of £10
Examine the “Time Lag” report; this shows you how many days it takes for your customers to convert (and to see the full impact of your marketing spend!)
This time lag report is important for optimisation but you need to be aware of implications for analysing data and results
» Sales will need to be backdated according to your conversion journey window.
» For something like Interflora or Credit Expert from Experian where cost of purchase is relatively low and user intent is high – i.e. I need to get some flowers I will search who I am going to use and will buy then and there. 95% of these users will convert within 3 days
» For travel it’s more likely to be something like 80% after 2 weeks, 90% after a month, 95% after 2 months but this will change on every individual client and by the value of purchase being made.
» Therefore your sales volume at month end will decrease over time – which means you will need to complete some data analysis to maximise your sales volumes
» Although gone are the days of overly manual reporting it does mean that you need to backdate sales for past months every time you report. Think about this process, the tech you are using and ultimately if excel based the macro’s you build to make reporting as efficient as possible
» When I was looking after Hertz on a 30 day conversion window I would still get sales from 267 days ago.This essentially means that a user has come back every month for the last 9 months and refreshed the 30 day cookie window
Working to a CPA target
» In a past life I worked with a dating client who had a CPA target of £25 for paid up subscriptions
» Looking at the “click to sale time” user path we quickly worked out that at the end of the month we needed to be at a CPA of £35 to maximize sales volumes.
» This CPA would then be at £28 at the end of the following month and at £25 two months later – therefore optimizing to get maximum available sales at target became a straightforward process
» The beauty of this model is when you are over target you know it will come down. You also know that if you are still fractionally over target in 6 months time the client will have forgotten!!
Choosing the right model for your business – the Adwords way
Search Funnels are a set of reports in AdWords which provide visibility into your customers’ paths to conversion
To access Search Funnels:
» In AdWords, select “Tools & Analysis > Conversions”
» Click “Search Funnels” in the lower-left corner
To start with Search Funnels:
» Examine the “Path length” report; this shows you how many clicks it takes for your customers to convert
» Examine the “Time Lag” report; this shows you how many days it takes for your customers to convert (and to see the full impact of your marketing spend!)
If you use MCC Conversion Tracking, go to “My Client Center > Conversions” and click “Search Funnels”
The Search Funnels Attribution Modelling Tool (SF AMT) allows you to distribute credit across the conversion path using 5 different attribution models.
The SF AMT shows you:
» How many conversions and how much revenue was generated under each model
» Multiple levels of granularity (campaign, ad group and keyword)
» Comparisons between different models
To start with the SF AMT:
» In Search Funnels, select “Attribution Modeling”
» Select your current model (e.g., “last click”) and up to 2 different models to compare against
Note that the SF AMT only uses clicks, while the Search Funnels reports include both clicks and impressions
The below screenshot form Google hits on some hard truths. There is no such thing as the perfect model!
So click attribution is not new by any means, 1000’s of technologies offer it. With the latest mass market and free release, I think Google have finally killed last click attribution in search for the businesses that were not able to use this, due to cost prohibitive 3rd party solutions. Something long overdue and something that will benefit every client
It’s important to note that, by dead and I’m hoping it will die very quickly – death in reality will be a number of years before the majority of clients use this as standard!