Is it time to say farewell to the old model of assigning all the value to the last click? Many marketers want the ability to measure customers' paths to purchase and understand the influences of each channel on a sale.

It would be a dream to accurately measure the impact of your TV ad on generating uplift in an outdoor campaign, driving interest in your press ad, which subsequently delivers a sale.

But can this now be achieved online?

I promised in my previous article, It's not all about the last click in search marketing, that I would share some insight into how I integrated my paid search, display, affiliate and email marketing channels to give clear visibility of customer journeys.

I have developed such an approach for a number of companies, including Warner Breaks.

I was fed up with treating each channel in isolation, measuring its individual impressions versus clicks, new versus repeat, spend versus return.

I knew that channels were positively impacting each other. I knew that my display banner advertising was delivering uplift on my paid search results. I knew my paid search was having a positive result on my affiliate networks. I knew my email marketing was driving visitors to brand keyword searches.

But I had no way of accurately measuring this effect.

I spoke to numerous companies to find a solution. Analytics companies can track the clicks, identify where users have come from (the referrer), but I wanted to also track customers off the site, including impressions (those who don’t even click). So I turned to an ad-serving company for the solution I needed.

The first step was to track all channels (impressions and clicks) through a single universal tracking tag. This meant regardless of the marketing channel, all user interactions were tracked to a single cookie.

By tracking the date and time of each impression or click, a report was produced to map out first interaction through to the final conversion on the website.

In its simplest form, a single customer path may look like this:

A customer journey

In addition to recording the date and time, it also captured the campaign name, the creative name, creative format, the website placement and even the keyword (for search marketing).

So what was the benefit of this report?

  • Measure channel cross-over and collaboration
  • Understand the online marketing mix
  • Measure the impact of online branding
  • Allocate budget more appropriately
  • Reduce media wastage
  • Improve online conversion.

I began by understanding the impact of the marketing mix. Was there evidence of display delivering uplift to search marketing?

I devised a simple pivot to look at how display banner advertising impacted the other channels. Top line results showed that of all customers that first interacted via a display banner, 25% ended up booking via paid search marketing and 12% via an affiliate network.

Display's impact on sales

I also found that of all customers that first interacted with an affiliate network, 23% ended booking via paid search marketing - proving affiliates had value in raising awareness.

I also looked at the creative impact. Did display banners advertising the Spa drive customers to search on Google for Spa Breaks or Spa Hotels? Did our Rich Media creative deliver a stronger uplift than standard banners?

Next I tried to understand how to impact future customers. We looked for opportunities to replicate successful journeys, strengthen messaging or even adding special offers to improve conversion.

It also allowed us to optimise media plans, by identifying which website placements were delivering positive uplift versus those not delivering any influence.

The final setup was to calculate our attribution modelling, sharing the CPA (cost-per-acquisition) back with all the channels that influenced the sale – a far more intelligent model than last click.

I have spoken to numerous companies over the past few months who are now taking this approach to their online marketing. Be warned, the volume of data maybe a bit overwhelming at first, but crack it and you have an incredibly powerful report that will change your whole outlook on your online marketing.

So is it RIP for last click? Personally, I think the time has come to be a bit more realistic when measuring online marketing, and this report could be key to that change.

 Matthew Finch - view blog

Matthew Finch

Published 14 January, 2008 by Matthew Finch

25 more posts from this author

You might be interested in

Comments (10)

Save or Cancel
Richard Hartigan

Richard Hartigan, Industry Manager at Google

Great topic Matthew!

This is a very interesting approach and a logical step for most organisations. Optimising the mix is a priority for all marketing professionals. This is unachievable if companies continue to operate with silos of investment.

Over the last six months, we have certainly seen the dividends of applying this method of attributing value to marketing channels. We have been able to determine the true value of marketing channels helping us to develop plans and allocate budget accordingly.

Personally, I am surprised that more analytics companies have not exploited this opportunity and used advanced attribution methodology as a unique selling point. Attributing across impressions is also a valid point, although in our experience this has not resulted in any valid uplift. I would be interested to see what ratio you apply to impressions vs. clicks with regard to display advertising.

The biggest challenge is to convince agencies, CPA partners and affiliates to accept their new found role in the mix. As conversion metrics are increasingly adopted as a means of payment, expect this subject to be one of the key marketing topics of 2008.

over 10 years ago


James Sandoval

Excellent post and nice work Matthew (just saw it now).


over 10 years ago

Ashley Friedlein

Ashley Friedlein, Founder, Econsultancy & President, Centaur Marketing at EconsultancyStaff

Hi Matthew

This is great stuff - very interesting. Out of interest, when you talk about customers who "interacted via a display banner" or "interacted" via an affiliate, what do you actually mean by "interacted"? There is everything from just being exposed to an ad (but not seeing it), to being exposed and seeing it (hard to measure I guess), to interacting with it within the ad (e.g. rich media interaction), to actually clicking through to a destination page.

Did you do analysis on the so-callled 'view through' value (or lack of)? i.e. where a customer was exposed to an ad but didn't click on it and then went on to (search and) buy? And do you have any evidence for these 'view throughs' that the customer did actually *see and notice* the ad as opposed to just being exposed to it (I guess you'd have to ask people to find this out)?

Ashley Friedlein

over 10 years ago


Mark Garner


this is probably the most intelligent thing I've seen about online marketing in a long time.

Like you, I had noticed a long time ago that paid online advertising was having a dramatic effect on my organic search engine rankings.

Google swore it should have no effect, but the difference was striking.

There were huge spikes in my organic traffic when I ran the ad campaign.

The results/clicks from the actual ad campaign were almost non existent.

Which obviously opens up the whole question of what exactly is the purpose of online advertising.

over 10 years ago


Sam Lempriere

Good post Matthew,

I have been working to get this system implemented for a while now. It certainly looks like the most sensible way forward.

One thing I am not clear on yet (and I guess once data starts coming in you can make a more informed decision), is how much weight you give to certain interactions.

For example, if someone sees a display banner that is used mainly for branding, then responds to an email relating to a new campaign and finally, clicks on a campaign display banner before converting, how do you split the sale? How valuable was the brand banner?

I'd love to see an article on how different people are splitting the value across the channels. This is where it starts to get complicated in my view.

almost 10 years ago


Jeff Greenfield

@Sam Lempreiere

You have asked the '$64,000 Question' -- how much weight does each interaction deserve?

It really depends on who you talk to.

@C3 Metrics - we have created an Attribution Gateway which is applied to every transaction/conversion.  In the case you mention:

someone sees a display banner that is used mainly for branding, then responds to an email relating to a new campaign and finally, clicks on a campaign display banner before converting, how do you split the sale? How valuable was the brand banner?

Research has shown that if the interaction the email came within a certain time-frame (12 hours - 3 days) of viewing the banner, then the banner should be attributed a portion of the sale.  The C3 Metrics Attribution Gateway would call the banner the 'Originator' - since they were responsible for introducing the customer to the product.

The C3 Metrics Attribution Gateway is discussed in detail in the White Paper called:  The Hushed Hiddend Gaps of Online Media Tracking and can downloaded @

almost 9 years ago


Bertie Stevenson

Hi Matthew,

Really interesting post for me as I'm joining up with Tagman next week.

You might find this blog entry particularly useful at how advertisers can move towards a 'best click' model without ruffling too many feathers with the publishers.


over 8 years ago

Ashley Friedlein

Ashley Friedlein, Founder, Econsultancy & President, Centaur Marketing at EconsultancyStaff

Hi Bertie

Of course it makes sense to understand the true customer journey, the true drivers of value and to do attribution modelling as a result.

However, the question I keep coming back to is *how do you actually split the payments*?

Not only is the administrative burden that much more but also I can't see how you can go back to Google, for example, and say "You know that click we paid you for - well we think 50% should go to XYZ affiliate instead" so could we have 50% back please? 

Or, it is well known that cashback/voucher sites are doing well under the 'last click' regime, for obvious reasons, even at the expense of paid search. But these schemes have promised the user a certain cash bounty or prize so how can you go back to them and say "You know what, guys, we're only going to pay you X now because we realise it was display advertising which actually drove the sale in this case".

So whilst it all makes sense, and marketers need to understand the *true* value chain to optimise their channels and spend, I'm not sure how the practical details of paying out varying amounts to all the parties involved can be addressed?

over 8 years ago


Craig Macpherson

Hi Ashley,

I work with Bertie at TagMan and would like to address your response re dividing bounty across multiple channels.  We've developed a system that deals with the practicalities. 

Firstly, for buying on a CPA via affiliates, affiliate conversion tags often use a basket value to calculate their bounty.  By deploying these tags through a container tag solution, the container can change the basket value parameter before passing it to the affiliate conversion tag.  If the container tag solution has visibility on all click & view events leading to the conversion, the amount passed can then be a function of an attribution model selected by the advertiser.  For example, if there are 2 clicks via an affiliate in the conversion path and a flat attribution/awarding model is used, 50% of the basket value can be passed to each affiliate conversion tag for the affiliates to calculate their bounty.

Secondly, for buying on a CPA via ad networks/publishers/ad exchanges, where the basket value or bounty is not passed to the conversion tag (but negotiated independent of the technology); a container tag solution can deploy these tags a proportion of the time, according to an attribution model.  For example, if a last biased linear attribution/awarding model is selected, for all scenarios where 2 ad network ads are clicked upon before conversion, the tag for the last ad network ad clicked upon could be loaded 70% of the time, and the conversion tag for the 2 to last ad network ad clicked upon could be loaded 30% of the time.  This would also result in the effective division of advertiser bounty across channels, and could run in combination with the ‘dynamic basket value’ method described above. 

Thirdly, you’re right in that when an advertiser pays on a CPM or CPC, you can’t de-duplicate that media spend.  However, these campaigns often provide conversion tags to show the conversions generated (and therefore value of) each campaign e.g. AdWords.   In which case, you could load these CPM & CPC conversion tags a proportion of the time per an attribution model also, channelling ‘rectified’ conversion data into their respective systems; removing the distortions from ‘transactional’ sales channels.  The advertiser would still retain a full click & view path record with their container tag solution, but can use the ‘rectified’ conversion data for their CPM & CPC channels (generated per their selected attribution/awarding model) to make better informed spending decisions.

By combining the 3 methods above, the advertiser can escape reliance on only 1 significant event in the conversion path, be it first, last, or most significant. 

over 8 years ago

Ashley Friedlein

Ashley Friedlein, Founder, Econsultancy & President, Centaur Marketing at EconsultancyStaff

Hi Craig

Thanks for that. Certainly interesting to see how you are going about making the attribution model approach more 'real' and actionable. Do you have any live clients actually doing this yet?


over 8 years 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.