Customer Journey Cookie Attribution – Statistically Flawed?

Customer journey and commission attribution were mentioned in my previous blog post about a4uexpo and after a few days of “affiliate flu” I’ve been mulling over the talks that covered these issues. Whilst it’s interesting to see stats that prove or disprove affiliate concerns and also to see networks and agencies looking into commission attribution I do think there’s some serious holes in the plot.

If I’ve understood the talks right it’s claimed that cashback shoppers now go direct to their cashback site to shop. What this means is that when studying customer journeys it’s noted that they do their shopping in a single cookie setting session. That on the face of it is quite a revelation, and completely contrary to what many affiliates believe actually happens. The conclusions drawn from the stats may well be right, but stats don’t often paint a true picture.

For example – do the stats analysis take into account that cashback shoppers are encouraged to shop in a single session.

Here are some lines from Quidco’s user guide with regard to best practice for tracking sales:

2. Clear cookies & cache, disable pop-up, ad, & image-blockers.
4. Don’t use voucher or other discount codes unless they came from the Quidco site.
6. Complete your purchase online, all in one session.

Friend of cashback Martin Lewis also has some useful advice for cashback shoppers:

Step 3: Clear your cookies.
If you’ve clicked through to several different sites that collect cookies, which will often include the product site itself, through a comparison site, this site, or another cashback site, it may not track unless you clear your computer’s cookies first.

Could this be a reason for seeing cashback shoppers shopping in a single session?

Now, it may well be that those studying the stats have taken into account this data. My initial inclination is that they haven’t! I’m not sure many people would seriously believe that on significant purchases people would not click around for a good deal, do a price comparison or head to a review site.

The data can’t pick up when people are using different computers (work to home, pc to laptop), reading reviews or doing price comparisons but being careful not to click on any links (so as not to burn cookies).

If, as demonstrated above, cashback shoppers are being told or encouraged to watch their cookie activity, then that in itself throws a significantly large sized spanner into the works. The data for looking into customer journey becomes seriously skewed even if a small number of cashback shoppers are deleting cookies or taking steps to make sure they’re only clicking one link.

The data may well be telling us one thing, but without knowing the full facts about why it’s telling us things means that it could actually be impossible to make serious judgments on bringing in cookie attribution systems over the current last click wins system.

Understanding customer journey and cookie attribution should make for interesting discussions over the coming months.

23 Comments Customer Journey Cookie Attribution – Statistically Flawed?

  1. Chris Johnson October 20, 2009 at 12:42 pm

    The only difference being that Cashback models just wouldn’t work without the Last Click Wins model, whether they clear their cookies before or not.

    Imagine seeing “Take out Sky through Quidco today to get £2-£80 Cashback”

  2. Jason October 20, 2009 at 1:01 pm

    not entirely true Chris.. what’s being suggested by the data (as I understand it) is that because a lot of cashback sales are “one referral to sale” then 100% attribution goes to the cashback site (i.e. no other site is involved in the sale process) so there would be little need for them to offer a range of cashback amounts… it all goes to cashback.

    However, my argument is – you can’t look at data and say “yes cashback sales are a one cookie journey” without covering the fact that cashback shoppers are being told to delete cookies or be careful in their shopping experience.

    If we’re going to go down an attribution pathway then imo these things need to be accounted for before deciding what stats are telling us about consumer journeys.

  3. Chris Johnson October 20, 2009 at 1:07 pm

    True Jason, I’m not denying that users are being told to delete cookies to ensure cashback is correctly allocated to them, however the meer oversight into multi attribution of the cookie just won’t work with Cashback Sites, should it come into play!

  4. Jason October 20, 2009 at 1:46 pm

    Think we might be at cross purposes a little 🙂

    Yeah I agree that if attribution is brought in then if a user has mutliple referrals then that’s going to cause other issues for cashback sites (part of me thinks “tough” as it rewards other sites in the journey but it could get very very messy).

    The argument given in the talks to counter that attribution is bad for cashback is “well cashback shoppers don’t have multi referrals” so they’ll be getting the full commissions regardless (one cookie = 100% commission)… which imo loads favour back into the cashback corner and doesn’t move us on from last click (which at least we all know where we stand with).

    It’s also wrong as as in the blog post… cashback shoppers aim to go for one click/one referral… making the argument for attribution and any data showing affiliates don’t lose to cashback slightly ropey (imo).

    I’m far from convinced attribution is the way forward and think a better reward model could be introduced in terms of tiers and/or offering a mix of CPM/increased CPA deals to good affiliates – at least to give them a fighting chance against cashback.

  5. Kevin Edwards October 20, 2009 at 1:55 pm

    Hi Jason,

    I was really pleased that multi or value attribution of commissions was a running theme at this year’s Expo and this hopefully shows how we’re a maturing industry; looking beyond the basics and starting to grapple with the more sophisticated but hugely complex area of rewarding for value (or as Julia Stent’s excellent presentation called it, ‘influence’).

    Julia’s presentation was probably the most insightful I heard during the Expo because it demonstrated how clued up merchants are appreciating that CPA marketing is no longer unique to the affiliate industry and that we cannot collectively bury our heads in the sand if we value the longer term sustainability of our industry.

    Any future move towards value, influence or multi-attribution models will probably benefit affiliates disproportionately as well as acknowledging that whilst last click is still a credible solution it will become increasingly obsolete as merchants look to understand and reward traffic appropriately. This is an inevitability of progress, better reporting, analytics and more knowledgeable employees.

    In a sense the fact that cashback sites need a fixed CPA is a minor one that can probably be overcome with little issue. This shouldn’t detract from the debate that is gathering pace amongst more enlightened merchants.

    I’ll hopefully get around to blogging on this in more depth in the near future.

  6. Chris Johnson October 20, 2009 at 2:28 pm

    FYI – i’m not in an arguentative mood, just curious as to how this would work (i know it’s a different angle on what Jason is posting).

    @Kevin – “In a sense the fact that cashback sites need a fixed CPA is a minor one that can probably be overcome with little issue.” Very interested to note how this is a minor one – is it simply by having cashback sites on a different commission structure to others? If so – makes sense. If not – would love to know 🙂

  7. Kevin Edwards October 20, 2009 at 2:39 pm


    What I mean is compared to the logistical and technical (as well as theoretical) issues of changing commercial models the fact that one type of affiliate requires a transparent CPA shouldn’t detract from the much wider issue merchants are looking into – that of how every aspect of their online traffic (not just affiliates) adds value.

    In my Expo presentation I concluded by presenting a ‘value attribution’ model that could include a number of different elements including where an affiliate is likely to occur in a sale path in order to arrive at a fixed commission.

    Value attribution will happen so we need to start discussing workable solutions sooner rather than later.

  8. Chris Johnson October 20, 2009 at 2:43 pm

    @Kevin – Sounds good mate. I’ll keep schtum now until i’ve watched your presentation back from the expo – should be live in a week or so!


  9. Matt Bailey October 20, 2009 at 4:38 pm

    Whilst I think that it is extremely important and credible to begin looking at value attribution on a basis that takes into account more than solely last click, I fear that we are viewing it too narrowly by concentrating solely on the affiliate channel.

    If we accept that in certain consumer journeys there will sometimes be more than one affiliate involved, then there must also be multiple channels involved (eg PPC, SEO, display, social media etc) which must be taken into account as they will also act as influencers on sales.

    Whilst only affiliates would be affected by a move to paying on a multi attribution model, all of these other channels still need to be included in any analysis so as to see the full picture.

  10. Clarke October 20, 2009 at 5:09 pm

    @Jason interesting article, certainly going to make me dig around and have a look at a few things. One thing might want to keep in mind as many of people presenting data are themselves not in possession of the full facts. Standing up on a stage and saying something doesn’t meant your data is the full picture, indeed no ones is but some people have much better data to work with than others. Merchants will be in one of the strongest positions with regards to data but you can bet that the majority have not a clue on what they are really tracking and how it is really working and what it’s really doing. Most openly admit that Affiliate is not a big deal in there marketing mix so the priority won’t be to make sure it’s fair for affiliates or indeed for all, what they are looking at is how they can get as many people promoting them and pay either the same or less.

    My only hope is that the new cookie stuff is brought in one Merchant at a time and that the poor folk up first trying to work out what to do and cocking up their program in the process get another chance by all the people involved. But I can bet you this, Google won’t change what they charge a Merchants, Agencies won’t change what they charge a Merchant, Big Display Ad companies won’t change what they charge, neither will TV, Radio and Magazines so that will just leave the easiest people to bully lead by some of the most clueless individuals. Can you just imagine it, Merchant with shocking PPC ad campaign that costs a fortune and interferes with the cookie or mass banner dropping site or toolbar/adware all interfering with a shopper getting a slice of the action. It needs a lot of work before it will be workable, the whole Voucher Codes thing was solved by Graeme and Myself at Paid On Results in 2007 and still same complaints are raised by big merchants and agencies, to them I say either shut up or start working with some real experts that will look after you and your brands.

    I do agree that last cookie is not ideal in 100% of cases, and certainly the best we have just now but before we go lock stock an barrel give up on it fully, we might want to try fixing some of the things that are wrong instead of ignoring them or worrying that X Big player will bully us, our spread untrue rumours or stop working with us. The Merchants after all pick what percentage Affiliates get paid and that’s meant to be based on the advice as to what they can afford and what they are paying elsewhere, potential leakage and cross over etc, unless that all been a big lie and the experts advising merchants are actually a bunch of lazy charlatans themselves.

  11. Clarke October 20, 2009 at 5:27 pm

    @Matt to be honest that’s all Affiliates are interested in, how it will effect our bottom line and you as leader of the IAB should be fighting our corner to make sure we don’t get screwed out of the deal. But you have vested interests as Agency head honcho and not really about looking after Affiliates as your main concern. Like I said to you and Will Cooper (NMA) at the Expo, real solutions exist but unless it’s a massive brand or a big agency or flavour of the month Network saying they know what to do, no one actually gives a monkeys to get things sorted unless they are the ones making out of it.

    Thing is if this was all just about Merchants understanding their spend and applying the correct spend in the right place, than that would be cool (couple of ad networks and “agencies” out of a job right away), as after all that’s what they should be doing, and the Agencies advising them should be helping with. But when Merchants start saying this is how Affiliates should get paid and don’t have a clue about the Affiliate channel, this is where you end up with problem, examples at the A4uExpo where terribly one sided, narrow views to get crowd pleasing cheers. Most of the crowds didn’t have a clue what was really going on and I dare say some people presenting data didn’t understand what they where saying, with the exception of a few fly guys playing smoke and mirror tricks with data.

    Seriously, more effort needs to be put in to what’s going on just now with the current system but almost every Network is too scarred financially to stand up to some of the things that are not right so are passing the buck to the “everyone gets a slice of the action” (but not really) models.

  12. HelenMarie October 20, 2009 at 5:30 pm

    Interesting blog Jason which has raised a few discussions. I think firstly on your point about incentive sites encouraging users to delete cookies. I have done a bit of work now on path to sale back when I was at dgm and now at client side. I had the same theory that a lot of sales attributed to incentive sites would have been attributed to one click only. But in fact this was a very small percentage and I suspect that most people do not read / bother / know how to delete cookies (I don’t for one – bother that is, I do know how to!).

    On the subject of multi-attribution I think that perhaps everyone is trying to jump the gun a bit and agree with a few comments made here. From my point of view as a client I firstly need to know the “real CPA” of my online activity. So, for example display costs as we all are aware are higher than that of other media, however this is only looking at last click. If you took into account that they influenced a sale and allocated a worth to this then it suddenly changes the way you think about this channel / allocate budget etc within the whole mix. This does not necessarily mean we would beak down the CPA but it is important to realise the role of each media channel within a sale.

    When it comes to affiliate I think the same thinking is needed but the difference obviously is that payment is made on sale. I do think that in order to protect the affiliate channel we need as an industry to look past last click, or, turn it on its head and affiliates need to develop their strategy to ensure that they are last click. This is extremely complex and I do not think the answer has been found yet nor do I think it will be blanket across the industry as will vary per client / sector.

    Really interesting topic and good to see it was on everyone’s mind at the Expo …

  13. Jason October 20, 2009 at 5:54 pm

    @Helen “once click….this was a very small percentage”… isn’t that contrary to what Julia was suggesting though? Now even I’m confused by it all lol!!

    “affiliates need to develop their strategy to ensure that they are last click” – this is actually the easiest (and possibly the most sensible) solution.

    Affiliates do have to be smarter (adding voucher codes, keeping feeds up to date, offering value) and things could be improved merchant side. The only area where I can’t compete is 100% cashback. My guess is that’s the same boat for a lot of affiliates.

    One further concern with attribution is will complex rules put off new blood or stifle creative ideas from ethical affiliates. It could become almost impossible to figure out “user value”.

  14. Jason October 20, 2009 at 6:11 pm

    I know I am probably going to sound dumb, but what about going back to the first click wins? This seems to me, from a basic standpoint, to reward the place which started the buying journey, and probably done enough to persuade the visitor to purchase said item and thus go on to seek out the bargain/voucher/cash back.

    I’m sure the theory is flawed, and hopefully someone will put me straight, but a few that I spoke to at Expo thought it is credible enough to think about.

  15. Kevin Edwards October 20, 2009 at 7:05 pm

    @Clarke, I have to say you have an incredibly dim and dismissive (as well as particularly assumptive) view of a number of your peers.

  16. Clarke October 20, 2009 at 7:22 pm

    @Jason (BigIdeas) yes it’s not a very good idea 😉 but seriously so many scamming bad apples around like adware, toolbar and iframe who would spoil that barrel the day it went live. Also the good guys would be all dashing to get the 1st click, when a new merchant went live, “click here for xxx”, oh I lied a little it’s not such a great deal, oh well, cookie has been set, so don’t care, it’s the merchants problem now. So not so good, any system that rewards laziness will end in tears.

    You don’t give the “hi welcome to pc store” person the commission for selling the PC extended warranty if the sales guy sold it, do you? Equally you don’t give the commission to the person ringing it through the till, so it’s all about rewarding the right person and last click is far from perfect but everyone getting a slice of the pie is even worse in most cases.

  17. HelenMarie October 20, 2009 at 7:22 pm

    Hi Jason – I think the confusion is that one merchant’s data is only relevant to them and not necessarily reflective of the industry as a whole. The path to sale is defined by the customer’s decision making process and this as we know will vary vastly depending on what the purchase is. I don’t think there is a blanket answer to the question of attribution; I really think it will vary depending on the client.

    With regard to paying out on first click, I was asked this question recently. Within the media mix affiliates would lose out significantly to display but just within the affiliate channel I can see an argument for this. However, my personal opinion is that more influence on the sale can be attributed nearer to the sale, so essentially last click. Take eBay again, their QCP model attributes more value to the click closer to the sale and I think many clients would take this view. Certainly open to debate though.

  18. Jason October 20, 2009 at 7:31 pm

    @Helen, Kevin, Clarke etc… what data can be “seen” by a merchant/network. Are you able for example to tell if Johnny Consumer clicked around, set a few cookies, deleted them and then went for his cashback?

    How much data is available – can it be drilled down to such levels.. and here’s another thought.. do consumers know that you/merchant/network know or potentially know this level of activity? Is that a hurdle to jump in the future or will shoppers become less keen to click on links/affiliate links for fear of tracking?

  19. Clarke October 20, 2009 at 7:40 pm

    @Kevin, you’re entitled to your view point and so am I. Sorry if it’s not sugar coated lumps of everything is wonderful but when you see (and record) the amount of things I have you start to question what people say (people question me all the time and rightly so). But being the biggest trumpet blower or the guy with the best parties doesn’t cut much ice with me. You’re entitled to your view and how you express it and I am of mine and if you don’t agree fair play.

    I do wonder why so many ex-DGM members of staff rally around to try their best to take shots at me any time they can, if it was over DGM Pro then I am sorry but I lost a lot of money when that system went live but even I moved on and forgave what happened back then, maybe you and a few others should consider that also. As I am prepared to work with people in an honest and open fashion but if they don’t do the same back then they can’t expect me to be nice all the time.

    As for my Dim and Dismissive view, it’s generally only when peoples words are not the sum of their actions. However I can’t change the fact that certain people will not like what I have to say, just like when Spyware was a major issue and how very wrong I was to be rallying affiliates behind stopping this practice that so many of my peers (Networks making money from it) supported and told me was actually a good thing, so I am no stranger to people having strong views against me due to my passionate love of this industry. They normally bring out the “you’re a shareholder in a Network” card but I am and have been an Affiliate much longer than most people have been involved.

  20. Clarke October 20, 2009 at 7:55 pm

    @ Jason – Confused at what Jason is who? Get guys, it brands your comments so you know who is speaking quicker 😉

    There is a lot of data from a Networks point of view but it’s not the full picture, also you need to use technology other than plain old cookies, Google has got great stuff but even that doesn’t show the Merchant the full picture.

    Yes most of us can see who clicked around, some of us can see who deleted cookies and then went on to do a sale, some of us can even see who came from what places with no-cookies and then activate them for buying from a site then turn them off.

    The question is, are we ready to really know how many sales change last minute for different reasons? Will certain big factions really around to try and stop people getting business if they publish facts (blackmail in this industry is saying you won’t work with someone and then publishing it), loads of politics in working Network and Agency side, less so Merchant and even less as an Affiliate. I have no boss so I can say what I like, most are not that lucky.

    It’s no small task pulling data together and checking it and making sure it’s based on facts, of course once you publish it some people won’t believe it anyway but you do need to remember everyone’s data is going to be different, think the DC Storm guys proved they had different data from others but didn’t make the others any less true or good to know.

    A clever guy told me at the Expo, you can make data say anything you want; you just need to know what you want it to say and work from there. Bah, wish I was that clever 🙂

  21. Kevin Edwards October 20, 2009 at 10:33 pm

    @Clarke – I worked at dgm over three years ago, quite what this comment about ex-dgm employees has on this discussion is baffling especially when I worked so hard at the time to raise the very obvious problems internally.

    I’m not dimissing views here – in fact I positively encourage debate as we need to address the subject Jason originally raised collectively as an industry

    Presumably you are concerned about affiliates across the board losing out as more merchants look to de-duplicate all their online activity whilst maintaining last click still reflects the value of their various channels.

  22. Jamie H October 21, 2009 at 11:23 am

    Hi All,

    Great post Jason,

    I think cookie attribution is and will be the way forward. However i have to agree with Matt, the only way it will happen properly is if the entire online journey is included seo, ppc, cpm, and cpa/affiliate which then would include a subset of channels voucher, cashback, content, comparison etc. Merchants are now more informed than they have ever been and want to understand the consumer journey and the TRUE value of each channel. This will inevitably shake a lot of cages as it is well known that, at present allocation is moved around to prove something is working well we all know it is clearly not (pi banners).

    I think it can be used well and if done properly prove the value of good quality affiliates. If its done badly then it runs the risk of destroying the industry. We are not the cheapest route to market anymore and have to ensure merchants understand the true value affiliates/cpa can provide.

    “The first rule of cashback club is to delete your cookies and tracking information off your computer (prior to click) so you know you will get the sale. This makes it really hard for networks and merchants to track this activity in line with other channels.” From James on Geoffblog

    I guess the question is do we move from cookie and look at other ways of tracking, you could do it by IP and User-Agent matching to make a best guess scenario but it is unlikely any networks or merchants have done this. Merchants would be able to see the real value as you wouldn’t need to delete cookies for cashback sites.

  23. Clarke October 21, 2009 at 12:16 pm

    @Jamie H, IP and User-Agent matching is possible but as the hours pass it becomes less reliable and you have to set a reasonable limit on what is considered acceptable. Best case just now is around 6 hours after a click as long as checks are made into instances of more then 2 sales in that time, from a specific IP and User-Agent match.

    You can see from the debate over at that even that has its issues if correct checks are not in place. With regards to how many cookies are deleted by Cash-back users etc, I would need to do some checking into data we hold and can’t say at this point how much information I would be able to find out, or indeed how accurate that could be without spending some time researching it.

    But it is all very interesting pieces of the puzzle that need to be trashed out and debated, but I doubt one model fits all is possible but you never know, I am keeping an open mind on it. Amazon being the market leader, anyone know what they are saying on the subject?

Comments are closed.