Google AdWords Conversion Metrics – We Need to Go Deeper

Google AdWords

Last few days I saw some weird new metrics on my Google AdWords report center. At first I didn’t pay too much attention, but then came Google’s message and it got me thinking about the whole concept of conversion attribution models and how easily we can get the wrong conclusions when not analyzing the data properly. I will now present how easily we can persuade ourselves to take wrong campaign actions based on data which may seem very promising.

Problem #1: Working solely by Cost per Lead

The simplest example I can think of can be in the finance sector. Let’s assume we are running two campaigns generating leads for mortgages, and focusing on optimizing cost per lead:

Campaign #1 generated 1000 leads, costing $5 lead, $5,000 total cost – and we seem very happy with it.

Campaign #2, generated 500 leads, costing $10 per lead, $5,000 total cost – and we are considering pausing it / changing or whatever else.

Now let’s add to the equation the following parameter:

Parameter k: ration of qualified leads vs. junk leads

Junk leads are those leads where you either find that the data is not validated, unresponsive to phone calls and emails and basically those leads which you find out are worthless i.e. probably not even interested in your product.

Quality leads are those leads which you can put into your sales pipeline and have a probably chance to convert to paying customers.

The ratio of quality leads vs. junk leads (or the proportion of quality leads from the total leads count) is super important, and sometimes will change your overall decision making, such as the following example:

Campaign #1
Campaign #1 had 40% quality leads = 400. Its “effective” cost per quality lead is now $5000/400 = $12.5 per quality lead.

Campaign #2
Campaign #2 had 80% quality leads > 400. Its “effective” cost per quality lead is now $5000/400 = $12.5 per quality lead.

Are you confused? You should be. Numbers can be very misleading. We now see that our preliminary analysis was misleading us to think one campaign is outperforming the other, while they are actually quite similar.

Ratio of Quality Leads vs. Junk must be put into the Cost per Lead Formula

But let’s make this even more interesting, and move on to Problem #2:

Problem #2: Ignoring the differentiated Value per Lead

Now let’s add to our equation Parameter m: average value per average mortgage requested.

Campaign #1

Campaign #1 generated 400 leads for mortgages valued at $40,000,000.

Campaign #2
m = $150,000

Campaign #1 generated 400 leads for mortgages valued at $60,000,000.

Bottom line here is that campaign #2 has the potential for much more business than campaign #1, even though its preliminary cost per lead seemed high.

Finally, my point is we really must calculate our entire conversion funnel, from each ad impression cost to actual income, in order to get proper decisions for our campaigns. Decisions which are not based on complete analysis, can sometimes lead us to the complete opposite directions and conclusions.

Leave a Reply