June 1, 2012
As PPC managers our job is to continually be optimizing. But what happens when we over-optimize? Over-optimization is the result of many changes within a small window of time. They are typically all individual great ideas but as a whole they can actually hurt performance.
Worse than just an uptick in CPL or CPA, when over-optimizing an account it becomes increasingly harder to revert changes. This usually leads an account into a downward spiral where only small incremental changes will right the ship. Which in itself is something that will blow your desk phone up with angry client calls where they are wondering why more isn’t being done.
The following is an example of how making a few layered optimizations can lead a PPC account manager far off track before they even know it.
Over one week the following changes are made:
- During peak 6 hours of day increase bids by 10%
- Increase bids by 10% on keywords that convert 10% or more below goal
- Conduct ad review and leave highest CTR ad running (as long as CPL is decent)
- Duplicate best campaign and launch mobile campaign
All seem like decent ideas. Nothing seems to conflict and none of them are major changes that would have a drastic negative impact or be complicated to reverse if it hurts performance. Let’s take a look at what could happen.
First let’s go over some assumptions. Let’s say that while CPC is going to almost always be lower than bid price, for the sake of this example let’s say they are the same thing. Of course if you are already in position 1 and increase bids by 10% you probably won’t pay nearly 10% more for clicks, but in this example let’s assume that you are not in position one so a 10% bid increase will results in a 10% CPC increase. Let’s also say that conversion rate and impressions stay the same.
Here’s what could happen…
Original bid on top keywords is $5. After 10% increase it goes to $5.50, during the 6 peak hours that you are bidding up by 10% your bid is actually $6.05 (a 21% increase in bids.) CTR was at 1% and now it is at 1.2%. Because clicks were at 1,000 they are now at 1,200. This means that while originally you were paying $5,000 for 1,000 clicks and 30 leads at a $166 CPL you are now paying $7,260 for 1,200 clicks and 36 leads at a $201 CPL for $7,260 during the six peak hours of your day.
Here’s a chart that might make this easier to follow:
The worst part is, now you have layered changes and you don’t necessarily know which one to revert. You can revert them all, but then you didn’t actually make any progress or learn anything. AND you’ve done this to your new mobile campaign, in which the only conclusion will be that mobile doesn’t work for your client and you’ll cut bait.
To course correct, you could start with one change; let’s say discontinuing bidding up on peak hours but what if that was actually helping? If bidding up during those hours was actually increasing conversion rates and lowering CPL, stopping it will make your account worse. At that point, you may think that you need to undo another change, such as raising bids on top converters, so you do that, but don’t redo the bidding up on the top performing hours of the day. So now you are just running you highest click through rate ads, which may have been the original problem and now the account is worse.
Meanwhile, the client/boss is calling and emailing wondering what is going on, so you are forced to come up with a strategy for new optimizations and implement those right away. At this point, you become lost in a cycle of over-optimization where you begin to lose site of what is working, what isn’t, what changes are making what impacts, and how to get everything pointed back in the right direction.
The thing to take from this is that clients often request big changes fast. There are some instances where the optimizations that are needed are obvious and it makes sense to put the pedal to the medal. There are other times, like the example above, where it is important to manage client/boss expectations and focus on incremental changes where you can isolate issues when the occur and can easily revert. Do one change a week, or two-weeks if you have a low spend account, until you can determine if the change helps or hurts performance. While waiting for results you can conduct keyword research, do a competitive analysis, make landing page optimization ideas, or write new ad copy to test.