Or, How Human Errors Cause Computers Such Grief
Yeah, it’s you. You broke the bidding tool. My personal favorite tools for this are the free ones provided by Google AdWords, like the Conversion Optimizer, Display Campaign Optimizer, and Return-On-Ad-Spend bidding strategy. But any of these things are terrible in the wrong hands, possibly your hands if you don’t understand how they work. As they say, a tool is only as effective as the person using them.
The Basic Causes
If you’re trying to diagnose a massive problem, start with the simple stuff. If you’ve covered it already, you can skip down to the more advanced section.
- Something has changed – You added negatives that were converting great. You changed your ads. You changed your landing pages. Your competition has finally gotten a clue. Your Quality Scores have dropped. Google changed something. Your conversion tracking broke.
- Your data was flawed – The AdWords Conversion Optimizer says you can’t even use the tool until you have at least 15 conversions in the same campaign. But, everyone recommends having more. Why? Because more data is what the tool requires to operate. This is the case of any algorithmic based software. A sudden spike in conversions or the wrong page designated as the point of conversion with cause bad results. And if you think you can get around it by stuffing a series of unrelated ad groups into the same campaign to get your numbers up, well, your results are going to reflect that.
- You tried to force a low CPA – Sorry, but the recommendations given by the Conversion Optimizer are there for a reason. I’ve gotten the AdWords tools to operate well below their minimums, but it’s not a guarantee. And other tools are going to have similar problems if you try to force a $40 CPA when you’ve been at $120. Remember, CPA bidding tools work toward a cost-per-lead, not volume.
The Complex Cause
This is the one that is not so obvious, and it requires a little math. Don’t worry, I’ll do the math. You just need to get the concept. Let’s start with a hypothetical situation. We have a campaign that receives 10 leads per day at $10 per lead. By allowing the bidding threshold to increase, we also get higher volume. But decreasing the threshold nets a reduced volume. Let’s say we have a perfect tool that by increasing to $11 we receive 11 conversions, and when we reduce down to $9 we receive 9 leads.
By that logic, reducing to $8 should get us 8 leads. But in this scenario, we don’t want less than 9. In this scenario, we’re looking to get our cost per lead down to $9. So, we go in and switch the simple setting. Perfect bidding tool, right? We check the reports every day and start seeing the numbers go crazy. Our volume has plummeted, and our cost has barely dropped.
I’ll show you why this happened in five day increments. If the tool is looking at the last 30 days of data, it’s seeing 300 leads at $10/lead. To create a lower average, it needs to balance out the average by going much lower, like to $8. But then its volume is too low to counteract the volume from early on.
Starting
Day 0 – 300 leads, $10/lead
Day 5 – 250 leads, $10/lead; 40 leads, $8/lead = 290 leads, $9.72/lead
Day 10 – 200 leads, $10/lead; 80 leads, $8/lead = 280 leads, $9.43/lead
Day 15 – 150 leads, $10/lead; 120 leads, $8/lead = 270 leads, $9.11/lead
Day 20 – 100 leads, $10/lead; 160 leads, $8/lead = 260 leads, $8.77/lead
As you’ll see in the chart, that means that we’ll need to go sometime after 15 days at fewer leads than we need to get to the desired average CPA. But let’s take a more extreme example. In the following scenario, we’ve decided we want a $7 CPA. Our bidding tool is going to have to take extreme actions, and they won’t be pretty.
Starting
Day 0 – 300 leads, $10/lead
Day 5 – 250 leads, $10/lead; 20 leads, $4/lead, = 270 leads, $9.56/lead
Day 10 – 200 leads, $10/lead; 40 leads, $4/lead = 240 leads, $9.00/lead
Day 15 – 150 leads, $10/lead; 60 leads, $4/lead = 210 leads, $8.29/lead
Day 20 – 100 leads, $10/lead; 80 leads, $4/lead = 180 leads, $7.33/lead
Day 25 – 50 leads, $10/lead; 100, $4/lead = 150 leads, $6.00/lead
As you’ll notice in the second chart, it takes roughly three weeks to get down to our CPA goal of $7/lead. Meanwhile, our lead-flow is a disaster. Those are the sort of months that can get an agency or employee replaced. While the math I used was extraordinarily simple, it should make you realize why CPA bidding tools go terribly wrong. It’s much better to ease them down a bit at a time.
Now You Know What To Do
Now that you’ve got the concept down, you should have no problem succeeding in the future.
Now, go make some money!