Apr 142008

Affiliates: 6 Ways to Catch a Thief

On the Today Show this morning Matt Lauer helpfully offered advice to consumers that the best way to shop online is to type the name of the store followed by "coupons" into their favorite search engine. It doesn't take much to set me off on a rant about affiliates, but that certainly did the trick!

Apologists for the industry will argue that coupons help tip consumers who might take their business elsewhere to close the transaction on your site. Others will say that affiliates simply siphon traffic that was heading for your cash register through their system, giving discounts to users who would have purchased without them and then charging commissions for "driving" the sale.

The truth, as usual, lies somewhere in between. Below are six tests to help get a better sense of whether affiliates are doing more harm than good to your business.

Most retailers prohibit affiliates from advertising on their brand name and trademarks. That's smart. Not only does such advertising drive no incremental traffic, it raises the cost retailers incur for running their own brand ads. In some instances, retailers allow this as a means of pushing competitors off of their brand terms, which may make sense in some markets, but not in most.

However, forbidding advertising on trademarks doesn't actually prevent it. Here are two tricks to catch the cheaters:

  1. Time of Day: some affiliates have learned that running ads on your brand at night is a good way to avoid detection and make a lot of easy money. Take a look at affiliate orders as a fraction of total orders by hour. You may find that affiliates are driving a larger chunk of the total at night and on the weekends when the cat is away.
  2. Geographic Targeting: Similarly, some vipers have figured out that by running ads on trademarks everywhere but where the corporate headquarters is they can break the rules with little chance of being caught. We recommend asking your industry friends, and/or Aunt Lois in Missoula to search on your trademarks periodically and send you screen shots along with the "properties" text of the links. You can also use Google's Ad Preview tool to see what things look like elsewhere.

I've spoken with some retailers who issue warnings to offenders. I don't understand that. How many warnings would you give to an employee who was caught embezzling?

On the broader question of whether affiliate sales are incremental or cannibalizing natural search, consider the source of traffic. How do your top affiliates generate traffic? If the answer is that search on your trademark yields natural search results for "Acme coupons" right after your link, then you may have your answer. Four useful tests to determine the source of an affiliate's traffic:

  1. Conversion Rates: Take a look at the conversion rates of traffic from each affiliate. If they all share the same source code target them by referring domain. Compare the conversion rates of the affiliate traffic to conversion rates of PPC traffic on your brand terms and on your non-brand terms. We all know that traffic on brand PPC terms converts at 5 to 10 times the rate of non-brand traffic -- these are "lay-down" orders. If a given affiliate's traffic converts at rates approaching your brand terms the reason why should be obvious.
  2. Click to Order Time: Do a similar comparative analysis on the mean and median intervals between the affiliate click and the eventual order. If the affiliate intervals are markedly shorter than non-brand PPC, this indicates folks have already made a purchase decision before they sought out the affiliate.
  3. New-to-file Percentages: Defenders of affiliates will often point out that while the "new customer" fraction is lower than other channels, it's not zero, therefore affiliates are driving incremental business. Hold the phone! New-to-file does not mean "incremental." When your loyal customers tell their buddies to shop at your store, those folks shop by your brand name. They're new, but neither your PPC program, your organic search program, nor your affiliate program should get credit for driving that order. Compare the New-to-file percentage of your affiliates to your brand PPC ads, if they're about the same, that says a lot.
  4. Coupon Percentages: What fraction of affiliate sourced orders come in with discounts/coupons? For these coupon buyers, what's the lifetime value? Given all that we've seen above the greatest apologist for affiliates would have to conclude that there is some cannibalism here. What fraction of the coupon orders would have to be incremental to offset the cost of the discounts given to the cannibalized orders? Does that fraction seem plausible given the rest of the analysis?

Ultimately, determining what is incremental and what isn't is tough. One interesting data dive might be to track three numbers over time: Online sales attributed to Brand PPC, online sales attributed to "natural search", and online sales attributed to affiliates, each as a percentage of total site sales. Since the start of the affiliate program, has the "growth" in affiliate sales come at the expense of the other two? Does one of these three go up when the others go down and vice versa? These patterns can also be indicative of problems.

There are certain classes of affiliates that unquestionably drive incremental traffic, and I'll discuss those next week. Unfortunately, for most retailers, those account for the minority of sales.

Happy Hunting!


7 Responses to "Affiliates: 6 Ways to Catch a Thief"
Great comments, George. As well consider this research and this research and Google's reversal (money grab) on trademark hawking which add more fuel to your fire. A few reactions for you: 1) The "pro" retailers (that ALLOW trademark bidding among their affiliates) believe (have been told by their affiliates and affiliate networks) that DIS-allowing affiliate use of trademarks relinquishes control to competitors and affiliates of competitors. Nice of them to give such "free consulting" away like that hu? It's not like they have a horse in the race, right? Yet retailers turn the blind eye every time. After all this is marketing not accounting. It's a rather mindless "gotta fill the real estate" argument that doesn't get into any measurement whatsoever (again, who wants to audit and measure!?). 2) How top affiliates are generating traffic isn't easily accessed/understood -- purposefully. Again, it suits many affiliate program managers just fine (to not know). Even worse, a vast majority of retailers don't hold affiliate networks accountable for gaping order fraud security holes (see Real Life Tests Confirm the Worst Fear at page bottom). 3) Ultimately, determining what is incremental and what isn't is tough but it IS work... work that affiliate managers will not get paid to do -- in fact they'll get a pay CUT (given how most have their personal bonuses tied to overall affiliate channel revenue). 4) Finally, it's interesting how affiliate marketing is both a darling and a dirty scoundrel -- at the same time. Will legislation in NY serve to support growth -- or diminish it? I'm not sure given the love/hate relationship with affiliates. In the end, as your partner Alan has told me, it's all about incrementality and after doing simple math (audits) this can be determined.
Great comments, Jeff, and many thanks for sharing! You're right on the money that no one seems to have an interest in looking under the covers. The affiliate manager doesn't want to put herself/himself out of a job, and to the folks in the corner office, affiliates are this huge, profitable program to which they'd like to divert more resources. Which leaves us running around shouting that the emperor has no clothes. Maybe someone will listen eventually. Any CFO's out there in the blogosphere?
linkdan says:
Regarding point 3 here, I think the conversion rates of affiliate traffic SHOULD be higher than generic traffic (unless it's a thin affiliate site). If the affiliate is doing a good job of qualifying traffic with (non-trademark) PPC keyword selection, targeted ad copy and a good sales-optimised landing page before sending the traffic on, then the conversion rate of this traffic should be quite high... but of course, most affiliates don't put that much effort into their campaigns.
Interesting point, and no question that in different business sectors, like music downloads, I'm sure affiliate behavior and value is different. For retailers, traffic quality will vary from channel to channel and within channels. We bid differently on different keywords and even different match types of the same keywords because the quality of the traffic is different. Display ad and Content ad traffic converts at much lower rates than non-brand search because, by definition, the users weren't actively looking for a place to buy X. The dichotomy between performance on competitive search (even sku terms) and brand search is so sharp that there is no question that it's a different animal. Affiliates in the retail sector aren't "qualifying" traffic, they're advertising to people who are proverbially in line at the cash register waiting to check out. George
Ryan Douglas says:
George, I find some striking similarities to my recent blog post from the 9th... http://www.plumbersurplus.com/Blog/post/2008/04/Affiliate-Marketing-How-to-Become-an-Affiliate-Hunter-in-6-Easy-Steps.aspx Perhaps its coincidental, but it seems rather odd that you would have the same number of actionable items as I (some of which are the exact same points), and end the post with the exact same "Happy Hunting". If my post inspired you to write this, I would appreciate some credit where due.
Hi Ryan, I don't blame you for asking, but in fact I hadn't read your post until just now. Chalk it up to great minds thinking alike. Had your work inspired me I assure you I'd have cited it. In point of fact, we bring up many different points, and I think the world is richer -- albeit slightly -- for having both articles in it. George
Ryan Douglas says:
George, Thank you for the reply. Great minds do think alike, and I appreciate the citation in your recent post.

Leave A Comment