Merchant Advice: Checking your Affiliate stats to Identify abuse or fraud
3 Mar
This is mainly a post aimed towards merchants and agencies, but if you are an affiliate or network then hopefully you’ll find my first blog post in what seems like years an interesting read all the same.
So let’s get straight into it and be honest here – whilst the perception of affiliate marketing continues to increase (and rightly so) this does also mean, even with a strong approval process on your affiliate programme, that the chances of abuse or fraud get bigger. Detecting it is challenging but it’s one of the most important tasks to ensure that you don’t pay out when you shouldn’t.
Lead programmes are often the most common problem but it really can affect everyone in the industry. For example, a few months ago I was innocently browsing a bittorrent site and an advertising message came up saying “This site is brought to you by Sky Digital” – the whole page was taken over with the Sky site and further investigation showed it was an Affiliate on their buy.at programme opening the site in an iFrame (screenshot below). I passed on the details to Helen and the affiliate was removed from the programme (and hopefully the network) the next day.
Imagine how negative this could have been for the Sky brand… You can almost see the headline “Sky Digital sponsoring illegal downloads”!
From a merchant and agency perspective it’s actually quite difficult to stop this kind of abuse, after all, not many networks give you the referral URLs of where the sales or leads have been generated to protect the affiliate from having their ideas of information stolen, but there are tools that you can use to help you.
Tip 1 – Lag time analytics
Most networks will let you pull of a report that shows the time between a click and a sale – this may be presented a number of ways, either with the actual time that the last click & the sale were generated or actually be displaying the “lag” time in seconds.
A suggestion of what you can do here is start to categorise your affiliates and look out for non-standard lag times. For example the lag time coming from a voucher code will generally be very low (as it’s often the last place that a user comes before completing their transaction) whereas a non-brand paid search affiliate of a content affiliate will be longer.
If you start to see very short lag times for affiliates that don’t fit into the brand/voucher/incentive types then it may be inductive of an affiliate using adware/spyware or at the very least someone to check with your network to find out more about how they are driving the traffic. Very long lag times may be where affiliates have been cookie dropping (like the example above) and be especially important for some of the larger brands – again, categorise and check up on it, often it’s not that the affiliate is doing anything wrong of course.
Tip 2 – Checking Transaction Times
Is going against your programme terms abuse? We’d say that it is and you can use transaction times to check out where this may be happening.
For example, if you see a high rate of out of office hours transactions coming from a particular affiliate then it’s likely that they are using a day-parted PPC campaign – not always abuse obviously but if you have tight PPC guidelines (that more and more merchants do have these days) and block brand then they may be doing this to make sure they are bidding on brand when you are less likely to be checking.
You can also look out for temporary increases in traffic and sales too – this could just be where an offer has been featured on one of the deal sites but it could also be that an affiliate has sent out an unauthorised email campaign and if you’re very brand conscious then this may be against your terms and conditions.
Tip 3 – Checking Conversion rates
Again, just like with Lag times different types of affiliates and different promotional methods will give higher or lower conversion rates – someone that comes from an incentive site is likely, for example, to have a higher conversion rate and grouping / categorising your affiliates comes in handy again here.
High conversion rates will often indicate brand bidding and very low conversion rates may indicate adware/spyware so its definitely another one to check.
Tip 4 – Using Analytics
You can use Google Analytics (or other analytics packages!) to get to grips with your inbound referrers – not everything will be picked up because the traffic comes from an affiliate network but as technology evolves more information is starting to be made available here.
Bits you can look out for include:
Referrals from Webmail clients (yahoo, hotmail, etc) means that one or more of your affiliates has either featured you in their newsletter or send a solus email out. Not indication of abuse but if it’s getting lots of clicks then you should check to see if it is something that you have provided them with or been authorised to send.
When I worked client side for a gambling company we used a few incentive sites and the volume and quality was often low but manageable – one day an incentive site decided to feature us as their main offer without telling us and it cost us a lot of money.. It would have been nice for a warning so that we could have at least budgeted for it!
Referrals from Search Engines may well show that you have affiliates using paid search and sending the traffic direct to site. One of our clients at AffiliateFuture recently noticed this in their google logs and as the affiliate was going against the Paid search policy they were removed from the programme and their commission was revoked but they were a “house hold” name in the affiliate industry. Perhaps an honest mistake, perhaps a deliberate one.. who knows?
If there is traffic coming through with no referrer then it may well be affiliates masking their links and removing the referrer from the users session. Some affiliates just don’t trust networks or merchants to see where their traffic is coming from but personally I always suspect that affiliates masking their links on purpose may well be doing something against the terms.
Tip 5 – IP Addresses
This tip really speaks for itself. Check the IP addresses for transactions, if you are seeing multiple transactions coming from IP then you may have problems. This could be that your network is using some advance technology (so for example with Veracitag on AffiliateFuture all these transactions show on the same IP address) but often it is more likely a someone doing something dodgy. We kick affiliates off our network daily for filling out multiple lead forms time and time again. It must take them ages and I can’t believe they are stupid enough to think that they won’t get caught out but there are ebooks being sold on ebay which tell people to do this to earn additional income – scary.
Tip 5 – Check the sales or leads even after you have paid out
I’ve left this tip till last as its one of the most important things that you can do and really does show where there could be fraud on your programme, not just abuse. This may be from an affiliate or it may be from an incentive site user but the outcome is the same.
As more and more networks and merchants look to improve the validation time and payout to affiliates this can lead to potential problems.
For example, depending on your sector you should be looking for an abnormal amount of chargebacks (gaming) and returns (retail and telecoms). Validating quickly can lead to a stronger affiliate programme but as mentioned, it’s always worth checking for abuse here.
If you’ve got a lead based programme then you should really validate the email addresses and make sure you are not paying for bounce backs too – also check how long it is before a user unsubscribes, if it’s after the first email (or before) then they have probably signed up for an incentive and you should check if you are allowing this.
I’m sure there is plenty more of this but the overall advise is there for all to see. You will get abuse, you might even get fraud. Plan for it, be wary of it and make sure that if you’ve got an affiliate programme you’ve also got the resource and the systems to manage it correctly and check things like this.
Recent Comments