While the incidence of fraud in the partner and affiliate marketing portal – https://affpub.com/ channel is lower than for many other sectors of digital, owing to our primary focus on “hard” metrics like actual purchases versus “soft” measures like impressions and viewability, it makes sense to be concerned and learn more about how fraud affects the category.
As the fraud threat grows in some regions, marketers need new and powerful ways to spot and eliminate it from their programmes. Fortunately, careful examination of your partner and programme data can reveal signs of possible fraud.
It’s important to note that unusual data are not always conclusive proof of fraud but rather provide signals to investigate more closely. There will always be some variation in the metrics generated by individual programmes and publishers – it’s in the examination of the most pronounced variances that marketers can spot the warning signs of fraud.
In my work helping leading brands make the most of the partner marketing opportunity, I am often asked how to spot possible fraud in partner and programme data. I recommend that you start by focusing on a few key metrics that can reveal suspicious results and would-be fraudsters. Here are four telltale metrics to examine:
1. Meantime to conversion data
Time passes between when a user clicks on a link and when they transact. The metric “mean time to conversion” calculates the average time your buyers took to complete their user journeys.
Naturally, there is a significant amount of variation in time to conversion for individual users. But take a look at your average time to conversion by partner and partner category. When you do this, you will likely find that many of your partners have similar or slightly varied mean time to conversion. Of course, some variation is perfectly normal. Further, larger variances are more likely for partners and placements that drive relatively lower volume.
But take a look around the edges of your data – to see if any of your publishers show both significant volume and very short mean times to conversion relative to other partners. Such figures may indicate cookie stuffing – that a bad actor is simulating clicks to a set of cookies periodically. Long mean time to conversion statistics can indicate that a fraudster is dropping cookies on all of its site visitors. Both are real problems and warrant examination.
2. Publisher conversion rates
Many people who click on a link don’t end up converting. A conversion rate divides the number of conversions by the number of clicks in order to determine an average percentage of people who end up making a purchase. Publisher conversion rates calculate this metric for the clicks and purchases driven by each individual partner. Again, there will be natural variation in those figures. But if you see one or a few partners with significant volume but extremely low conversion rates, it may be a sign of possible click fraud. Click fraud artificially inflates reported clicks, and is most likely to occur in CPC programs or where brands pay for specific “real estate” on a partner site – like a top feature position. Those are circumstances in which it behoves a bad actor to drive up click counts.
3. Conversion rates by placement/IP address
Here you take a look at the rates for different placements – like a specific web page. Abnormally high conversion rates can be indicative of creative fraud — in which fraudsters alter your messages.
While sample sizes for individual placements can be quite small, the ability to click a link and instantly see the creative at a specific location can offer immediate evidence for or against fraud. Remember also that low traffic placements are less likely to be fraudulent because a bad actor is unlikely to go to all the bother for a small payout. Where fraudsters reveal themselves is in a combination of peculiar data and significant volume or scale.
4. Traffic by URL rank reporting
Digging into where your traffic comes from can help you spot cookie reuse and unauthorised placements. What you’re looking for here are large numbers of clicks associated with particular URLs – scale makes this fraud economic for the fraudster.
Many of your highest traffic URLs are going to be for your largest partners. That’s totally normal. What you are looking for here are URLs that aren’t associated with a legitimate partner, or where the traffic generated is out of sync with the expected scale offered by a partner.
Just a starting point
The four metrics discussed above are both simple to understand and provide strong indications that specific elements of your programmes warrant investigation. Over time, you can identify if there are specific elements of your programmes that seem to be under threat.
Remember also that data anomalies are a reason for an investigation, not accusation. Shared data and analysis are a great way to collaborate with great partner
- Street: Noordmolendwarsstraat 184
- City: Rotterdam
- State: Illinois
- Country: United States
- Zip/Postal Code: 3035 Wh
- Phone number: Noordmolendwarsstraat 184
- Listed: June 14, 2018 9:35 am
- Expires: 80 days, 12 hours