Comments around the affiliate channel looking for a new solution to last click is completely fallacious as are many of the articles and panels I have seen around attribution modelling.

I have seen many articles written about attribution models over the last year along with the flux of roundtables, conference agendas etc. that now include this mythical model. 

This is much to my frustration as I have yet to see anyone come out and give me an example of an attribution model that is providing great insight and benefit to an advertiser. 

An additional annoyance is that there seems to be much discussion about attribution and the affiliate channel, with claims that the channel is looking for a “new solution to last click”, which is completely fallacious. 

There are so many incorrect assumptions about attribution that I felt the need for a myth-buster, and to be honest a bit of a rant at the so called experts that are frankly filling the trade press with utter drivel.

Myth 1: the affiliate market is looking for a solution to the last click model

No it isn’t. This is a channel that pays a commission per sale and the most viable, fair way to do this is on a last click basis. The industry is not looking at a solution to replace this, it works. 

However, the channel, like the rest of the online industry, is looking to understand the role of its affiliates within the customer buying journey. We want to see to what extent affiliates are contributing to a sale, whether that is through multiple touch points or one touch point that did not result in last click. 

This does not mean that as an industry we are looking to change the payment model. It just wouldn’t work and to split a commission would be time consuming, resource heavy and most likely turn out with a similar result to the last click model anyway. 

All evidence I have seen points to a split commission model ultimately delivering the same results as last click. Think about it, it’s likely that an affiliate will win and lose, therefore end up more or less in the same position. 

There are obviously some areas of promotion that need to be considered more carefully (voucher and cashback being the most obvious), but split commission is not the answer just yet in my opinion.

Myth 2: attribution is about splitting commissions

I’m not really sure where this came from; attribution is about understanding what drove a sale, not splitting commissions. 

Splitting commissions potentially could be an outcome of understanding and guiding marketing spend to channels driving the best ROI, but as per above I can’t see how this would work. 

For me, attribution modelling is simply to understand what drove a sale and how customers interact with a brand and how it leads to a purchase.

Myth 3: attribution modelling is all about affiliate marketing

I’m not sure when and where affiliates became the centre of attention for attribution modelling as I see it as a vital tool across all online (and offline) channels. 

Typically, big brands spend a lot of marketing cash on getting people to the top of the funnel, common sense and evidence shows this tends to be the display model. 

I personally see an attribution model initially being fundamental to understand if and where this spend is working, what part of your display plan is driving people into the funnel and what isn’t. 

If I, as an online marketing manager, could understand this then I would invest my marketing budget where it is most effective. A similar story for social media and long tail search terms. 

The affiliate channel has developed and can be seen as a driver for the top of the funnel as well as the bottom of the funnel, so attribution modelling does apply here as well but it is no more important or less important than understanding attribution across all channels. 

Of course the affiliate channel is complex in that it consists of hundreds of affiliate sites, but so do display media plans and long tail search strategies.

The need for data

I am by no means an expert in this field but I think advertisers / agencies / networks need to start thinking about what attribution really is and I would love to see some real data and conclusions start to come out of this rather than conference agendas essentially about Clients de-duping, which is not an attribution model.

Looking at attribution within the affiliate channel specifically is complex and I see it developing to potentially offering a wider variety of payment models to affiliates. 

For example, a large affiliate publisher is not that different to a display publisher, so why not look at post impression tracking potentially or paying a fixed cost to secure traffic levels.

My thoughts on attribution modelling are:

  • It is a tool to understand touch points within the customer journey which includes but is not limited to display, affiliate, mobile, social media, natural search, paid search and direct traffic.   
  • This should provide insight into how best to spend marketing budget not split commissions.
  • It should inform the advertiser about the customer journey and present opportunities to improve and invest in key areas. 

I recently attended an attribution session and at the end of one of the panels a question was put forward to the panel which consisted of advertisers and a technology provider that had implemented an attribution model.

The question went something like this “So, can you tell me what changes you have made to your marketing strategy since implementing an attribution model. For example did you find opportunities to move marketing spend to more effective areas of a channel or from one channel to another?”  

None of the panel could answer this and openly admitted that they didn’t know...


Published 19 July, 2011 by Helen Southgate

Helen Southgate is Senior Online Marketing Manager at BSkyB and a contributor to Econsultancy. 

1 more post from this author

You might be interested in

Comments (13)

Save or Cancel


Spot on. This is about understanding your company's increasingly distributed conversion funnel, and is a black art at best. It's important that companies understand where the decision to buy/convert happened so that they can weight their spend/effort to the right areas (which could be literally anything). Also, last click modelling makes a lot more sense in acquisition programmes, but less so for conversion, where it is merely another indicator.

about 7 years ago

Alec Cochrane

Alec Cochrane, Head of Optimisation at Blue Latitude

This should provide insight into how best to spend marketing budget not split commissions.

Spot on Helen.

You should use any data you can get about the variety of channels that users responded to to get information to optimise your marketing. Trying to use the information to split commissions won't work. We tried it 5 years ago and it didn't work then, it won't work now. My Affiliate website doesn't care what other channels led to a sale on your website, because I sent you traffic that converted so give me my god damn money.

I think part of the problem is that senior people want things to add up to the total number of sales. They rarely do because of the way that the users move in and out of the various channels over time meaning more than one channel can be attributed to the sale. More than that, the data coming out of the analytics package is probably not very accurate anyway, so it is pointless trying to reconcile it to be the total of the number of sales.


about 7 years ago


Tom Young, Acquisition Manager at Citi / Egg

Great piece Helen.

Attribution modelling stikes me as falling somewhere between a hammer and the change in house prices.

Fundamentally understanding the effects on conversion and sales that different parts of a media mix have is a useful tool - much like a hammer - only when you have cause to use it. Whether this involves split commissions, increasing investment in certain areas or ownership of inventory to prevent competitor threats, etc.

I'm not sure that many folks have got the above right yet. I'm not sure whether this is because of the precision of the data, the agility of the organisation the financial model or just a general desire not to change. Genuinely I have no idea.

Finally - Attribution Modelling is like the change in house prices. It's a conversation piece that never seems to die. Round tables, conferences, and late night conversation - Attribution modelling has a fair chance of being discussed.

about 7 years ago

Neil Sampson

Neil Sampson, Head of New Business at DC Storm

As a technology and service provider that presents at and attends many talks and roundtables I would like to add my comments.

I agree with Helen that the most important part of the process is in understanding the contribution that your channels make at all stages in the path to conversion. This allows you to re-invest budget in ways you would not have considered based on last click measurement alone. You also come out of the process having a priceless knowledge of your customers’ visit behaviour and the impact that external influences have on that behaviour such as the strength of your brand or the influence of offline advertising. This has always been our approach to attribution at DC Storm. Any resulting model is just a way of measuring success more accurately, having done that crucial analysis.

My feedback from speaking to advertisers and affiliates alike is that last click wins is not necessarily the right or wrong approach and to assume that one approach is best for any given business is short sighted. As the search engines force affiliates to create richer content sites that add value to the user journey it’s natural that affiliates will start contributing more and more at the earlier stages of the path to conversion. Surely affiliates want to be rewarded for their efforts in this respect? Also, I think it’s a contradiction to say that last click wins works best for a given channel but then say that you place value on earlier contributions. If you place value on an early contribution then why not eventually reward it and encourage similar activity from others?

People often single out the affiliate channel when talking about attribution. This is because the affiliate channel covers 2 crucial points. You need to consider 1) how you measure performance and 2) the reward model you apply to your performance partners. I think that the reluctance to move away from last click has come about by a lack of understanding and an assumption that attribution changes both the measurement and the reward. Perhaps in some platforms it does. I strongly suspect that you’d upset most affiliates by suddenly overhauling their reward model. However, affiliates are adaptable. It’s possible to measure their performance based on overall contribution and still reward them on last click. Why not start by introducing a bonus for earlier contribution while the payment model is still based on last click? Through clear communication from advertiser to affiliate, attribution models need not be so overly complex that your partners don’t understand or accept them. A model can be expressed as a series of logical statements about how your business values contribution rather than a set of charts or graphs. By giving your affiliates visibility of their last click performance alongside their attributed performance they can understand how any change in reward model will affect their income and put in place changes if necessary.

I completely agree with you on your point that de-duping is not attribution. Sadly it’s a method that some solution providers use to justify the investment from advertisers in their attribution tools. Granted, it’s always going to be hard to quantify the benefit in money terms when you won’t understand the impact of attribution until you’ve carried out the analysis. Carrying out the analysis and having the right experts to understand the data is in itself what makes the investment worthwhile.

Attribution is all about understanding the contribution of each of your channels and how they interact in the path to conversion. An attribution model allows you to measure the performance of your channels taking into account this deeper understanding.

about 7 years ago

Hero Grigoraki

Hero Grigoraki, Head of Media Product at

"Myth 2: attribution is about splitting commissions
I’m not really sure where this came from; attribution is about understanding what drove a sale, not splitting commissions."

I wholeheartedly agree with your points and angle on the matter - I believe that in order to make business decisions, you first need to evaluate the data, and we're still trying to figure out ways to collect and analyse the data, much less make decisions on who to pay and how much.

HOWEVER - on the Myth 2 point. The rationale is pretty simple. There will always be the need for advertisers to cut down costs. No one will ever say "let's continue paying everyone until we can analyse solid data"; instead, they will start funneling all channels through the grossly faulty last click wins model regardless of type of activity, payment model, cookie type etc. To clarify - last click wins is a perfectly good model when comparing like for like. Faulty when you use it to dedupe CPM, CPC, tenancies and CPA alike.

It's inevitable that channel value attribution will be forever interllinked to payment attribution. Especially when you have companies who believe they've cracked it and try to impose their methods on everyone around them. In my opinion, the complexity of the attribution question comes from the signle fact that there is no single solution. No "one size fits all" approach. We can't have one model applied to every single client, in every single vertical. Why? Because each site is unique and therefore their marketing mix is different, the consumers need influencing from different channels and in different sequence.

Tell you what though - Once Google Analytics' Multi-Channel Funnels gets released, a lot of people will start grasping what this is all about. And, best of all, it doesn't come with a forbidding price tag to balance the complexity of the project.

about 7 years ago


Helen Southgate, Online Marketing Controller - Strategy & Planning at BSkyB

Thanks for the comments all, suspected this would fuel a good debate.

In terms of costs, I have not seen any strong case to say that rewarding on attribution reduces costs. Logic would say that it would either:
A) Keep costs the same - on an payment by attribution model my guess is that the winners vs the losers balances out and overall you end up paying the same anyway. All channels, not just affiliate, work to a CPA model and whatever way you mix it up you are still going to be working to that same CPA, therefore costs do not change.
B) Increase costs as you end up paying out several times across one transaction and due to not being able to control transactions your CPA is not fixed - this is not a good way to work and I can't see anyone employing this strategy. An example would be where ppc search is the last click, this carries the same CPC whatever I decide to pay at other stages of the customer journey, so if I do decide to pay a commission to an affiliate and /or a display partner for their contribution then this is going to increase my costs. Unless anyone wants to convince Google to work on a CPA?

My biggest issue however with an attribution model is how do you decide on what click is more valuable, it is a purely arbitrary metric and I can't see how this can ever be accurate. In one case a first click may hold much more weight whereas in another transaction it could be the middle click. Without getting inside an individuals consumers head this surely is impossible.

about 7 years ago


Alternative viewpoint

I agree with the argument against attribution modelling to determine CPA/payments. However, to look at this another way...

Let's say you measure the notional value of a customer on the basis of the profit they generate in a 12 month period (but could be any period appropriate to an individual business).

Last-click attribution modelling is in my opinion of lesser importance than measuring the lifetime value (or post-purchase 12-month value) of customers that generate their last-click through a given marketing channel.

If channel "A" delivers new customers that generate an average £200 profit over the 12-month post-purchase period, then that will be more valuable than channel "B" that consistently delivers new customers that only make a company £20 profit over 12 months.

One click is not the same as another; one acquisition is not the same as another; one customer is not the same as another.

I adjust my channel budgets based on the lifetime (i.e. period of 12 months in our business) value of a customer and not solely based on which channel refers most customers on a last-click basis.

about 7 years ago


Helen Southgate, Online Marketing Controller - Strategy & Planning at BSkyB

Alternative Viewpoint - this is essentially a revenue share / value attribution model which I am all for and I think many Clients are moving to more of this type of model. There is also opportunity to look at the retention opportunities that affiliates can present as well as upsell and crosssell. I'm all for this way of thinking about all online channels, as well as afiliate.

about 7 years ago


Online Insider

Hi Helen,

You are one of the few advertisers who agrees that re-attributing based on fantasy logic is a silly idea.
In a industry overflowing with trendy charts, it takes a level head to see through it. If weighted click made financial sense, then right now we would all be using the sliders of Atlas' "Engagement Mapping" Tool.

Of course it doesn't make sense, and you can't say "this click is worth more... because it's a video!" so what is the best way to use path to conversion data for financial gain?

In my experience you put as much data as you can in a database, and then study similar sets. For example "brand search with prior display" vs. "brand search with no prior display". Looking at the aggregated behavior patterns you can begin to allocate "attribution" back to more influential channels. The most prominent case I've seen was looking at how other channels are nearly always present in "voucher code site" conversions. This was not a pattern seen on any other site or channel. Using A/B testing with a charity creative it was possible to work out the true uplift effect of voucher codes, which was actually positive (surprisingly).

The number one thing you need is a quality source of data. Number two is a decent database analyst. Excel just isn't up to the job, and any off the shelf solution will never accommodate the quirks and peculiarities of a real life campaign. Once you have this, I'm sure you will be reassured by findings that backup up some assumptions, and surprised by revelations that some assumptions were simply not true (search term crossover, or lack of, for example).

I wish you well on your quest!

about 7 years ago

Paul Cook

Paul Cook, Director at NCC Web Performance

Hi Helen, and all.

Really interesting to hear a counter viewpoint.

The real problem for me here is the perception that full user journey and attribution modelling aren't really being used to a. make different budget spending decisions or b. allocate commission payments differently. They absolutely are. I know this because our clients are doing exactly that.

a. Using attribution models for spending decisions: Most all our clients are doing this. There's been little public proof since going from adoption to implementation to results to sign-off on public release of those results is a long process. Luckily, we have at last got to that point with one of our clients, Tui. The case study is at and it shows how Tui has upped SEO by six times and generic paid search by three times purely on the basis of full user journey and attribution reporting. Econsultancy is covering this very soon.

2. Using attribution to allocate CPA payments better: Completely agree that splitting commission payments is difficult, chiefly because of the way rewards and offers sites work. So, the middle ground our clients are either already taking or working to is 'best click', which allows advertisers to award credit for a sale (and the commission) according to a ranking system that gives priority to preferred channels when they appear in the path to conversion. The Tui case study shows how it saves 30% in CPA payments doing exactly this.

I'm always in favour of applying some cynicism to the many hypes of digital. But I'm afraid, in this case, it is misplaced.

about 7 years ago

Paul Cook

Paul Cook, Director at NCC Web Performance

Btw, apologies for making a classic "a. .... 2." error.

about 7 years ago

Robin Davies

Robin Davies, Country Manager UK, DE & APAC at ValueClick

Great discussion Helen, well done for kicking it off and for framing it so clearly.

In response to TagMan post, I see from the TUI case study "A 30% saving in commission payments to CPA partners through accurate deduplication". After downloading the full study I see that it wasn’t actually a de-duplication exercise but a flat attribution of sale to all contributing clicks. Flat attribution is entirely arbitrary and yet not entirely without worth. To be fair to TagMan I also thought there might be a silver bullet in this and even wrote a short article to this effect on iMedia [], back in 2008. We set about trying to make it work, only to find that it just isn’t satisfactory, academically. While it does have the affect of highlighting possible under-investment in upper funnel media and SEO it is nothing more than the re-jigging of a report, just like engagement mapping enabled us to do. How has our industry talked itself into the idea that there is a mathematical expression to tell you where to put your money? I think attribution is unsettling because our industry growth is based on accountability yet now we’ve talked ourselves out of last click and realized that arbitrary attribution isn’t quite enough either. Mediaplex offers a custom solution to its clients but we are not offering a mass market product.

Helen, I think I was at the industry event you refer to and it felt much like 2008.

about 7 years ago

Paul Cook

Paul Cook, Director at NCC Web Performance

Hi Helen and all,

So you can see a full interview now on Econsultancy with the Tui Specialist Travel client Tess Bedard on how they've used full path-to-conversion and attribution awarding to make big decisions on where to put their money.

Hoping this cuts through the idea that this isn't happening. It's true it isn't happening much but that's because you need a certain tag management system to do it :)

almost 7 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.