mFilterIt Blogs

How Conversion Rate links with Ad Fraud

What is Conversion Rate?

When advertisers run ad campaigns, their ads are displayed thousands of times on the internet. Conversion rate (called CVR sometimes) refers to the number of conversions compared to the clicks received (being a more performance-based metric, generally advertisers use clicks as the base instead of impressions).

CVR can be either a heavily watched metric, or a completely ignored metric depending on what the advertiser is paying for:

  1. If Advertiser is paying on CPC model, then CVR will be the most important metric most likely. Advertiser will spend ages trying to increase the CVR and optimizing their strategies and targeting to get better CVRs.
  2. If advertiser is paying on Conversion (CPS / CPL / CPI etc models), then the CVR is generally the most ignored metric. The logic being that if you are not paying for clicks, then why bother with how many clicks came, or what was the CVR for a source.

This dichotomy doesn’t make sense, since the basic premise of HOW the advertising is being run remains the same. Only the payout model has changed. Irrespective of whether the payout is on click or on conversion, the sequence which is required is

  • Impression
  • Click
  • Conversion

Which means that any publisher to run ad campaign, the final comparison metric is CPM only. If the campaign makes sense with normal CPM metrics, then the publisher would normally continue a campaign. Whether the payout is on conversion or click or impression for that matter is immaterial.

Consider the below metrics:

  1. An advertiser pays $0.50 for every install of their app.
  2. Consider an CVR rate of 0.1%. This implies that for getting 1 install, a publisher has to trigger 1000 clicks. And the effective CPC earning for the publisher: $0.0005 / click.
  3. Let’s go one step back. To get 1000 clicks, how many impressions will it take? Let’s say the CTR is 1%. This means that to get 1000 clicks, publisher needs 100,000 impressions to be served. The effective CPM rate here : $0.000005 per 1000 impressions

Now, the question to be considered here is, does the above make commercial sense for a publisher? Is the CPM rate this low to be justified for a publisher to run this campaign when there are multiple other campaigns available at better CPM rates? The main reason for this crazily low level of CPM rate or even CPC rate is the extremely low CVR of 0.1%. The only way this business model makes sense for publisher to run an AdCampaign at this CPM rate is AdFraud.

When advertisers ignore the CVR rate, and assumes that if they are paying for conversion and hence clicks don’t matter, they are turning their back to a key metric which can identify fraud on their campaigns.

Why low CVR indicates AdFraud

One key element which is generally missed when running a conversion linked campaign is AdFraud types:

  1. Click Spamming (app-based)
  2. Cookie Stuffing (web-based)

The point of both above strategies is to steal Organic traffic and ensure that end conversions which were already occurring organically are rehashed as inorganic conversions. The Advertiser basically pays for his own traffic.

The way these frauds work, is to take advantage of the last-click attribution model. When a conversion happens, the last click is searched for. If the last click was found to be a non-organic source, then it is attributed as the source of the conversion and gets paid for it.

This means, that as a publisher, I can keep firing clicks repeatedly for different users and deviceIDs (in the background), and if any of those users go organically to trigger a conversion, I will get paid for it.

Obviously for this to succeed, I will have to fire millions of clicks and then hope that some of these end up triggering organic conversions, which I will then steal. But that means that my CVR will be extremely low.

Here is an example of an advertiser who ran a subscription-based campaign in the middle east.

Android Campaign :

Date Clicks Installs Conversion Rate
01-Jan 1,672,490 180 0.01%
02-Jan 1,525,557 227 0.01%
03-Jan 1,658,371 206 0.01%
04-Jan 1,618,298 216 0.01%
05-Jan 1,510,869 211 0.01%
06-Jan 1,438,888 209 0.01%

IOS Campaign :

Date Clicks Installs Conversion Rate
01-Jan 1093,751 141 0.01%
02-Jan 1026,334 140 0.01%
03-Jan 954,728 152 0.02%
04-Jan 913,661 124 0.01%
05-Jan 839,953 163 0.02%
06-Jan 828,295 137 0.02%

So, in 6 days, this source basically triggered 15m clicks across Android and IOS. This is amazing. The total population of UAE (target market) is ~10m! So, if this source is to be believed, this source covered the entire country in 6 days and then started over again!!

From any logic, there is no way this traffic makes sense. It is impossible for a genuine publisher to waste so many clicks and impressions and earn so little. It is impossible for so many users to click so many times before installing the app. A 0.01% ratio means that users clicked on the ad 10,000 times before deciding to install the app. Would you ever have the time or energy to click 10,000 times before deciding to install an app?

So, understanding the CVR metrics for your sources and tracking it regularly is important irrespective of whether your payouts are linked to impressions/clicks or not. They hold important trends for you to understand fraud on your campaigns.

Also remember that an extremely good performing source (in terms of ROI / ROAS etc) doesn’t mean it doesn’t have fraud. The extremely good performing source, who has a crazily low CVR is most likely stealing organic traffic from you.

So, as a thumbrule:

  • Extremely bad performing sources are obviously bad
  • Extremely good performing sources are most likely also bad!

And that is the key message from this article which any performance marketer should take back.