From this article you will learn:
- Why you should automate bid management
- What are the main principles of bid management
- If Google Automated Rules can work
As every PPC specialist knows painfully well, bid management is a tedious and time-consuming job. It is overwhelming at best and impossible at worst. That is why bid management automation is a must, it takes all the grunt work off your back, makes all the necessary calculations and simulations instantaneously.
Google caters to that need by providing automated bid management option in campaign settings. However, not every advertiser likes the idea of completely giving away control over bid management to Google. More advanced users prefer to set up their own bid management automation and for those users, Google Automated Rules were introduced. At first glance it may seem like a really good idea to have a set of your own automated rules adjust the bids the way you like. But can it really work?
Before answering the question let’s have a closer look into bid management itself. Any bid managements system, to be of any use, needs to comply with few principles. Namely:
Bid management is a problem of the most optimal allocation of limited resources. That being said, keywords cannot be effectively optimized out of context, without taking into account other keywords within the same campaign or dynamic market.
So let’s say you set up a rule to reduce bids on most expensive keywords and decide to reduce the bid every time keywords CPC exceeds $2. But if you are in a competitive market soon enough $2 for a bid won’t be enough to keep you on the first page. If you wanted to keep clicks on the same level you would have to put up with a higher CPC. However, fixed value of $2 in your rule doesn’t allow this kind of flexibility.
On other hand, if you are doing really well and are able to increase your Quality Score, $2 might become too indulgent.
It is also not a very good idea to use the same rules on a campaign in which you are spending your entire daily budget with one in which you’re not exhausting your set budget.
The whole idea behind bid management is to best predict consequences of bid changes based on historical data. But the data taken into consideration must be relevant. In other words, circumstances in which data were collected must be as close to current circumstances as possible.
For instance, let’s say you set a rule (Rule A) that reduces bids on keywords with a high conversion cost every 14 days. Looking back on the 14 days, the cost per conversion of any given keyword was very high. However, the max CPC of this keyword was significantly reduced 7 days ago by another rule (Rule B), and for the last 7 days the cost per conversion was low.Circumstances have changed and the data older than 7 days is no longer relevant, as it does not correspond with the current bids. Unfortunately, Rule A is oblivious to this fact and unnecessarily reduces the bids further more.
Even if the bids stay the same there are an abundance of other circumstances that might warp the relevance of the data. The rule of the thumb is that the older the data the less relevant. The oldness of data is also known as “recency.”
It is important for the data we use to support bid management decisions to be reliable. As in game of dice you would never bet all your money on 12 just because you happened to throw it on the last roll, you should never attempt to base bid management decision on thin data.
Let’s say that this time we set up a rule that bids up on all keywords with conversion rates higher than 2% in the last 7 days. Keyword “Headphones” gets 10 clicks and 1 conversion so its conversion rate is 10%. However, can you trust this keyword will maintain such a high conversion rate based on just one conversion? What if this one conversion was by chance and in the next 100 clicks the keyword will not get any other conversion? Again, the rule we created ignores this problem altogether and bids up on the keyword.
The principles listed above are just basic rules every bid management system should abide by. It doesn’t guarantee success but it’s a good cornerstone to build up on. As you might have noticed relevance and statistical significance can be often mutually exclusive. Especially in the cases of small or long-tail campaigns, it can prove impossible to collect significant amount of data within reasonably short window of time, when data is still relevant. In case of long-tail campaigns this problem can be often minimized by a portfolio approach where keywords are aggregated into groups of high inner correlation. However, if campaign has very few clicks altogether, effective bid management automation is not possible.
Now, back to our question. Can Google Automated Rules work? If you read this article carefully, you already know the answer.