Two key metrics publishers take into account when picking sales affiliate campaigns are commission percentage, which is the percentage of total sales value on referrer sales which they receive and cookie length, which is the period of time after the initial referral the affiliate is entitled to commission for.
These metrics can be hard to compare though, am I better off taking the higher commission rate with a short cookie time? Or should I take a lower cut and bank on the fact that customers will return during the cookie period to make additional purchases?
Another factor to consider is also the tracking methods used by a network. Standard cookie tracking deteriorates over time as people will either clear their cookies or have browser settings or automated programs that do it for them. According to statistics from the Affiliates4u 2011 performance marketing guide, after 30 days 43% of sales are coming from alternative tracking methods. This means for a network using standard cookies your commission is getting cut by 43% after 30 days, a longer cookie length, whether this stat just gets worse, may not be very beneficial.
To provide a rough guide I have analyzed some of our personal account merchant referrals over a long enough period of time that they would be pretty stable, especially for the first merchant. The first merchant is on a 30 day cookie and is recorded from a pool for 20,814 sales over 17 months. The second merchant is on a 45 day cookie and is recorded from a pool of 2526 sales over 6 months.
This first chart for merchant A shows that 59% of sales occurred within 24 hours of the original click, this would mostly be people who clicked through and purchased in the one internet usage session, but also include those who may have made a mental or physical note to come back later in the day and make a purchase. As you can see this tapers off quickly with less than 5% of sales coming in the second 24 hours period and settles around the 2% mark for each day of the first week. Taking into account the perviously mentioned issue of regular cookies being removed over time it seems that after the first week there is a fairly uniform chance that someone will return on any given day to make a purchase.
It may be better to think in terms of the next graph, the amount of total sales completed on or before each day of the cookie period.
Here we can see that by day 10 you have 80% of the sales value. To get an idea of how cookie length and value would be interrelated here we will insert some numbers to put a figure on different levels of revenue. Say that you have 1000 sales at an average of value of $100 with a commission percentage of 10% over the 30 day period. This means the total revenue over the period is $10 * 1000 sales = $10,000, $8000 of which comes in the first 10 days. Meaning if your commission was changed to 15% on a 10 day cookie you would get $15 * 800 sales = $12,000 which would be better value but unless you have a rough idea of the distributions of sales you are getting from merchants it may not be clear.
In merchants we have ran 2 commonly offered cookie lengths are 30 and 45 days, to get an idea of difference between these 2 we examined the second merchant.
Here we see an even more pronounced effect towards the first 24 hours, with about 85% of sales occurring in the first 24 hours. It also features a similar tapering off after the first couple of days. What we want to check here though is how much extra revenue does having an extra 15 days of cookie length generate and if presented with competing offers, one of which offers a longer cookie for less commission percentage on each sale, what is the best course of action?
The graph confirms what you would have already been suspecting from previous graphs. The extra 15 days only generated an extra 2% of sales. 45 days at 10% or 30 at 12.5% are offers you may be likely to see. Using 1000 sales at an average of $100 per sale again as an example, with the first offer you would get $10 * 1000 = $10000 and with the second you would get $12.50 * 980 = $12250. For both methods to break even you would need 20% of sales to come in the additional 15 days of cookie time. It is possible that depending on the products being sold this could happen but from the data we have to go by here it’s definitely not in our niche.
Another aspect to consider is that of sub sales, those are purchases by customers referred that have been completed in transactions after their initial transaction by a merchant. Sometimes these sales will attract a different commission rate than first sales as some merchants feel they have already paid for and acquired the customer and shouldn’t be rewarded as handsomely on additional transactions that still occur in the cookie period. This data is taken again from merchant A, but from a smaller poll of 6952 sales which could have their sub sales more accurately tracked. From this, 8.3% of sales were sub sales and their distribution across days is show below.
Apart from the first day when people tend either buy things across multiple transactions but in the same session of browsing, there doesn’t appear to be a great pattern in when people come back. It could be said though that people who have purchased are more likely to come back across the cookie period than those which haven’t before. When considering a different rate for sub sales it is useful to figure out what percentage of your sales are sub sales, also what is considered a sub sale, is it just people making a second transaction on the current cookie or is it also those that had your cookie 6 months back and have just required it and made their first purchase on this cookie?
AffClicks is well poised to pose and answer these questions on others in future about affiliate marketing with anonymous user data (opt in of course). Will be interesting to see how it develops and any extra data we can provide to better inform affiliate marketers and help them make better decisions in their businesses is a great thing.