The marketing potential of Advanced Attribution is huge, and many companies struggle to devise a strategy that’s suitable for them.

Everyone is talking about it, yet marketers find themselves confused, stuck, overwhelmed by the many options, model types, and data sets.

Although adoption is gaining momentum, with brands allocating more resources and budget to attribution modeling, only 26% of companies use advanced attribution that goes beyond last-click according to an Econsultancy Quarterly Digital Intelligence Briefing.

This is truly alarming, considering that companies spend sums of up to and more than £5m annually to drive customer acquisition in online marketing alone. This is huge investment that is not necessarily ROI proven.

So, how can you use the platform to turn data into insight and action?

Here are three ways marketers can approach it, without becoming stuck.

What is advanced attribution?

Advanced attribution will change the way we approach digital marketing forever. Is not a trend, it’s not a fad, it’s here to stay whether you chose to change and implement it or not. The longer you take the faster your competitors would  take advantage.

Advanced attribution is simply shifting the way you measure your marketing valuation. This means going from last click where 100% of the valuation for credit, ROI and revenue is attributed to the last marketing channel before a purchase, to valuating all of the marketing channels in a customer journey.

What this means is that you are giving proper credit to your complete marketing mix, instead of only one channel. Makes sense right? Why wouldn’t you.

Why advanced attribution?

Let’s start off with the name. The industry calls it 'advanced' attribution because the current last click model you have been using for years is also an attributed model, it’s just the most simplistic version.

When you hear advanced attribution you automatically know that the value is being attributed across the whole marketing mix. The reason why this is so important is to basically prove 100% of your marketing spends in terms of ROI across all your channels, not just one.

Not knowing which 50% of your marketing works and which doesn’t is no longer acceptable. Some of the benefits of advanced attribution are:

  • It helps you understand the value of marketing activities that build demand at the top of the funnel. For example, can provide guidance on which keywords to bid on during a marketing campaign, those keywords you had no idea your customer started their search with, because you don’t even bid on them.
  • It's useful in discovering the true value of affiliate referrals and determining payments, no more duplicate payments. Who still pays double now a days?
  • It helps marketers find the optimal frequency of ad serving such as for display ads, meaning you don’t spam and scare your prospects by flooding them with your ads. 

The challenges: don’t become stuck here

Recently I was sitting in an Econsultancy attribution roundtable and learned from different marketers and agencies how confused the topic gets them.

These are the three things that will take the confusion away and make it simpler:

1. The  model: go for algorithmic

There are several ways of measuring advanced attribution. Essentially these divide themselves in two. First, Rules based models. Second, Algorithmic statistical models.

Please don’t become stuck here, make your decisions and move forward. If it’s algorithmic what you choose, you would be 110% better off than using your current last click model.

Rules based models are exactly what it says in the tin. You put rules based on the position of where the marketing channels are in your customer journey. For example, if you have a customer journey that looks like this:

Paid Sarch>Display Ad>Email> Conversion

Under a rules based model, you say, whatever comes first will get x% that’s assigned by a rule you make. This means that paid search will get the same percentage every time is first, as so very other marketing channel that is first. You don’t want rules based models.

Why? Because they become a self-fulfilling theory. Your marketing will work as per the rules you assign, not very smart.

Algorithmic Statistical models are what you want to get. There is a plethora of different models and how they work. Don’t get stuck here. Just make sure you go for this option.

What’s important to know here is that it's science and it's data based. It's science because it’s a mathematical algorithmic statistical model. Not to worry, it just means there’s maths instead of rules doing your calculations for valuating every marketing channel.

It's data based, because it’s based in your own customers data. This means it looks at your brand's past customer journeys and how each channel interacts differently in getting you a conversion.

For example, if you have a customer journey that looks like this:

Paid Search>Display Ad>Email> Conversion

And another customer journey that looks like this:

Paid Search>Social>Display Ad>Email> Conversion

The first channel, which in this case is paid search, will vary significantly in its valuation. Even though each example starts with Paid Search, the customer journey is different, therefore the credit for the first channel is also different.

2. It’s not about data, it’s about what to do with it

Basically, the biggest internal organizational challenges a company has once advanced attribution has been implemented is to take decisions on the data it gives you. This basically means driving the insight. Tell me what to do and how to do it?

Boom! We have attribution implemented, all channels are integrated. We have spend, cost, revenue data, and ROI figures for every digital channel. We even integrated some offline channels like call centre data and point of purchase.

Amazing, we finally have one single source of truth. After this implementation you got  access to 354 reports. I believe data must be turned into insight, and so does Econsultancy.  

What does this mean? It simply means, tell me what to do and where to take action. It’s awesome that we now have 354 reports to look at data and customer journeys and all the amazing things advanced attribution gives me insight into.

So what? How do I really get the action I need to be a source of customer, competitive and marketing advantage?

Look for platforms that have gone a step beyond, and include a scenario planning capability that allows you to predict future behaviour which is fundamental to creating strong customer lifetime value models and optimising marketing effectiveness.

And as Econsultancy stated in the Marketing Manifesto, digital channels provide new and valuable sources of data and customer insight that can be acted upon in real time.

This is what you really need, besides the 354 reports, you really need to be told what to do with the data. Where to spend your money? how to better forecast to get £1M in revenue, or how to drive 10k conversions, or where to spend the additional £150k for the Christmas campaign.

If you do not see this feature in your platform, then you don’t have a modern platform in front of you.

3. Change management: this must come from the top down

Advanced attribution will impact the entire marketing organisation.  

The current challenge, according to Jim Sterne, founder of eMetrics Summit & Digital Analytics Association, is that too many large organizations are using incentive structures that foster (dis)integrated digital marketing by issuing bonuses to individual teams based on the performance of the marketing channels they are responsible for.  

This we all know. Now think about it, if you are a channel manager driving attribution, you will have a channel conflict with every other channel manager.

Because each team (search, social, email, display, affiliates etc.) must prove their own channel’s performance for compensation, they’re shoved into internal competition and left to seek metrics and attribution models favourable to their particular channel.

This creates both operational and data silos. This means that for advanced attribution to be successful and for channel managers to optimize, forecast and use actionable data to drive better ROI, the direction needs to come from the CMO, Marketing Director, or whoever owns the budget across all marketing channels.

Adoption of attribution modeling is accelerating because digital media keeps growing new channels, making it harder for marketers to understand what drives ad performance, meaning whatever your product or service, there’s a huge potential to better understand the other 50% of how your marketing works.

Digital marketers must step up and start things moving with advanced attribution. The longer you take, the faster your competitors will take advantage.

Sebastian Gutierrez

Published 30 October, 2013 by Sebastian Gutierrez

Sebastian Gutierrez is eBay Enterprise Senior Product & Partnerships Manager  and is a mentor for online start-ups at Wayra. Find him on TwitterGoogle+ and LinkedIn.  

8 more posts from this author

You might be interested in

Comments (6)

Save or Cancel

william christi

A good job and well written points, all the words were represented pictorially perfect and it helped me out a lot.

thanks for spending your valuable time.

Good one


almost 5 years ago

Sebastian Gutierrez

Sebastian Gutierrez, Senior Product Manager at eBay Enterprise

Thanks William, great feedback

almost 5 years ago



Thanks for the great write-up.

I agree with all of the points you made - re the Figure 4 that shows that the lack of knowledge as the main culprit of not doing attribution, what do you think are good resources for gaining that knowledge? That's resources in addition to signing up with econsultancy of course!

almost 5 years ago


Matt Lovell, Head of Customer Data, Insight & Analytics at Eurostar International Ltd.

An interesting article but I still think it misses some of the key problems people have with attribution in terms of a simple issue that an algorithm no matter how advanced can't truly understand a customer's behaviour without

a) all of the data touch points
b) someone to actually diagnose what the data is telling it

I've worked with a number of so called 'advanced' attribution technologies and there are still a number of problems with a lot of them in that all they are doing is identifying whether a particular touch point was involved in the journey to a conversion. Now you may be thinking to yourself - that's exactly what I want to get but there are problems in that with many channels, this tells you something you already know and doesn't really help you change your marketing or adjust you spend to make a significant change.

Let's look at the pros and cons:

* You can finally see any channels, sites, placements, keywords etc. that don't actually influence conversions in any way which means you can remove them. Great - but how much of your marketing is actually that cut and dry

* You can see any channels, sites, placements, keywords etc. that appeared to be driving no or next to no conversions but are actually influencing more than expected and can continue spending on these / invest more in them. Once again it's a nice addition but ultimately given it's rare that this learning then makes these sources significantly more cost effective, the material effect is limited

* For areas where advertisers are really keen to get to the bottom of effectiveness, these tools provide limited benefit for example

* Display activity where it is difficult to justify that cookieing a user for "seeing" (very much in inverted commas) your ad has actually influenced them to drive them to have an interest in your brand

* Re-targeting activity where you are stuck with the difficult question of whether the user would actually have still come back to the site if they hadn't been re-targeted

* Reward driven affiliate sites where without understanding how the customer uses these sites / what the competition is doing on them / whether they could have gone elsewhere, it is difficult to judge the nature of their influence.

And that's just for starters. I appreciate that systems and tools are evolving but suggesting that the wider advertiser base should jump in and start using attribution (because they would be stupid not to) is almost more damaging as it suggests to them that what they are / were doing was wrong while promising to radically change their activity when the truth is that without the resource to run tests and evaluate observations from any data, it really won't...

almost 5 years ago

Sebastian Gutierrez

Sebastian Gutierrez, Senior Product Manager at eBay Enterprise

Hi Matt- these are great points you have raised. I agree with you in that they are important. At eBay enterprise we tackle these issues and solve for them. Let me know if you are interested in reviewing these in more detail and we can set up some time to talk.

almost 5 years ago


Marcos Richardson, CEO at DIG

The subject of attribution, especially third party, is currently an industry issue and everyone is looking for the de facto solution. We have been working on a few tricky accounts such as a country destination example, whereby they are trying to attribute footfall into Canada. They rely on International third party publishers such as National Geographic and partners such as holiday booking companies (e.g. Thompson) and on the receiving end, destination hotels etc.

We have tested the recent BETA launch called Data Hub , We have also tested,,,

To date they all fall down on the issue of tagging, in that if the source is not tagged it cannot be placed into the measurement construct for attribution analysis.

The other barrier to entry is privacy, all of the main tracking solutions such as Adobe, GA, Web trends have the built in capability in log files to match IP and RTGML (Real Time Geo Mobile Location) and personal information but they are limited by legal compliance on storing this information for retargeting. We have persistent and session cookies but they cannot be linked to personal information then linked to upstream/downstream activity outside of the owned company network.

From our interrogation we have found the following solutions…

Pinpoint and advanced programme that feeds from programmatic and can decipher psychographics, which enables hyper targeting and return enables profile attribution

Trillion an advanced tagging solution which also matches user ID’s to a built up database allowing retargeting. This system can tag and track upstream /downstream traffic to patch third party personal attribution

Reverse Engineered Targeting, whereby we look closely at profiles and associated user journeys to hyper target campaigns. This allows us to attribute online/offline traffic based on highly defined parameters whereby rendering all other activity noise or non-associated.

Real Time associated, we look at time sensitivity of matching Advertisements and Promotions (online/offline) with respondent queries and actions

Custom attribution modelling, Last Interaction, Last Non-Direct Click, Last AdWords Click, First Interaction, Linear, Time Decay, Position Based. Example FIRST: 10%, MIDDLE: 50%, LAST: 40%

The conclusion is that much like programmatic, attribution should be a learning machine based on tagging, inference and a linked dynamic growing database.

11 months ago

Save or Cancel

Enjoying this article?

Get more just like this, delivered to your inbox.

Keep up to date with the latest analysis, inspiration and learning from the Econsultancy blog with our free Digital Pulse newsletter. You will receive a hand-picked digest of the latest and greatest articles, as well as snippets of new market data, best practice guides and trends research.