Here’s a script to help manage bids by keyword match type.

Some account managers like to use the same keyword in multiple match types in their accounts to get better bid control or to show more relevant ads for queries that are closer to the keyword.

To execute the strategy properly, bids for each match type need to be kept at a certain level relative to the other match types. For example, if the exact match keyword’s max CPC is $1.00, the advertiser may want to bid 80 percent (or $0.80) for phrase match and 60 percent (or $0.60) for broad or modified broad match.

This is called “tiered bidding” or “stacked bidding.” There’s some debate about the merits of the strategy, but ultimately, I believe every advertiser should be able to test it if they want. Since my job at Optmyzr is to give people technology to streamline account management, I wrote an AdWords Script that will check if your bids are following the expected bidding tiers.

Why you should automate tiered bids with scripts

In a debate on #ppcchat, Kirk Williams reports that the reason most people don’t like the stacked bidding strategy is that it complicates account management. For example, if you found that the keyword from our example above wasn’t performing as expected, and you lowered the exact match bid to $0.50, you’d have to remember to scale back the other match types’ bids, too — or risk triggering the wrong keyword in a future ad auction.

It’s a real pain point, but passing up the possibility to achieve better results because of management concerns is not a great argument, especially when it’s relatively simple to deploy technology that can help. An AdWords script is an easy-to-implement tool that can make your life easier when testing a stacked bidding strategy.

Sometimes agencies have to do things they don’t believe in

Whether you believe in this strategy or not, if you are a PPC consultant or agency, at some point you will come across someone who pays your bills who insists you try it. That’s exactly why I wrote this script myself in the first place, many years ago, when a client wanted me to test it for them.

Just a few weeks ago, an Optmyzr client asked if we could help with stacked bids, so I dusted off the old script and made it work with the latest version of the reporting API. Now, I’ll share it with everyone reading this.

About the script

The AdWords script can be put on a schedule of your choosing, and every time it runs, it does two things:

1. Identifies when bids are no longer tiered correctly.
2. Outputs a spreadsheet with new bids to upload back to AdWords.

Because I’m giving you the full code, you can extend it as you want.

What is stacked bidding?

In stacked bidding in AdWords, the same keyword is added in multiple match types to better control costs. The closer a query is to a keyword, the higher the chance that it could lead to a conversion, hence the more an advertiser is willing to spend.

For example, say the keyword is “Titanfall xbox game” and the following three searches trigger an ad for that keyword:

• Titanfall xbox game — This is exactly what the advertiser chose as keyword, so it should convert quite well and it can have a high bid.
• Used titanfall xbox game — This is still very close to the keyword, but the conversion rate may suffer if the advertiser doesn’t sell used games.
• Used xbox game — This is a much more generic search. It’s no longer for the specific game, and it’s for a used game, so it has an even lower chance of leading to a sale. These broad matches will probably convert at the lowest rate, so they should have the lowest bid.

At an intuitive level, stacked bidding makes sense because the closer the query is to the keyword, the more likely it is to be valuable, and the higher the bid should be.

However, the reason I am not entirely sold on the strategy is that using bids to control which keyword Google selects may be difficult in situations where it would make a big difference.

How Google prioritizes keywords

It helps to understand what happens if you have the same keyword in multiple match types and Google needs to decide which one to use. Google’s own help materials explain how similar keywords match to search terms pretty well. It can be summarized as follows.

The sequence in which Google determines which keyword to use:

1. The keyword whose text matches exactly to the text of the query.
2. If there are multiple keywords with the same text, the exact match version.
3. If none of the keywords’ text matches the query, then the one with the best ad rank will be used.

Remember that in stacked bidding, the strategy is to always have each keyword in multiple match types, so rule 1 won’t apply, because it assumes there is only one keyword with text that matches the query. We would always have to evaluate rules 2 and 3.

Rule 2 says that the exact match variation will win, so having a different bid for the different match types doesn’t matter to Google, because it simply takes the exact match variation.

Rule 3 says that the ad rank determines the winner if none of the keywords had the same text as the query. In this case, we can already eliminate the exact match variant from contention because it is too restrictive to show an ad for a query with different text. Phrase match may enter the auction if the word order of the keyword and query text match. So at most, we have to figure out whether phrase, broad or modified broad wins the auction. What happens now becomes dependent not only on the stacked bids, but also in which ad groups the keywords are, because this impacts quality score.

How Quality Score works when the same keyword exists in multiple match types

When adding different match types of the same keyword to the same ad group, the real-time quality score (QS), the one used to calculate each auction’s ad rank, will be the same for all keywords, regardless of match type. This is true because they would show the same ad, with the same landing page, and have the same predicted click-through rate (CTR). All three QS factors are in effect the same.

If the different match types of the same keyword are in different ad groups, things change, because QS may differ when the ad groups have different ads, different landing pages, or are part of campaigns with different targeting options. In this case, you cannot control which keyword wins the auction, because it’s hard to predict the QS of every auction.

So here’s what we know:

1. If we put the different match types in the same ad group, tiered bidding might not actually change what Google would have done anyway, which is to prefer the more restrictive match type.
2. If we have different ad groups for each match type, tiered bidding can help by boosting ad rank for more specific match types, but there is a significant degree of uncertainty due to QS variations.

There are two typical ways advertisers structure ad groups when they do stacked bidding.

The exception to the rule

It gets more complicated. Google’s help article describing the keyword priority sequence says that “on rare occasions, the system will prefer to use a [less specific] keyword that is cheaper — meaning it has a lower cost-per-click (CPC) bid — and has a higher Ad Rank.”

So basically, a high stacked bid doesn’t guarantee anything in terms of prioritizing what Google serves in the auction. An unsophisticated advertiser may find the lack of control palatable because Google won’t pick a less specific keyword that has a higher CPC; in other words, they’re saying, “Don’t worry, your CPC was lower, and we thought you’d like the extra click.”

But for most advertisers, CPC is just a factor that ultimately determines ROAS (return on ad spend) or CPA (cost per acquisition) — so a low CPC for a click that won’t convert because it went to the wrong landing page is far worse than a more expensive click that went to the right place where it had a decent chance at converting.

Alternatives to stacked bidding

When I was at Google, we never recommended following a stacked bidding strategy, or even having the same keyword in multiple match types, for that matter. But Google states the problem very well themselves: “In rare cases, the keyword with the highest Ad Rank might seem to be less relevant to a particular search term than other eligible keywords. Because higher relevance is generally correlated with a higher Ad Rank, this should happen infrequently.”

So what do we as PPC managers (and control freaks) do? It basically comes down to managing queries in two specific ways:

1. Put yourself in the position to have a keyword that matches the query text as often as possible. So use query data to add good new keywords to your account on a frequent basis.
2. Find those rare cases when Google doesn’t respect their normal prioritization rules and add negative keywords to force them to serve the keywords you want.

The first strategy is easy enough and can easily be done in AdWords, but the second one’s a bit more complicated because you need to run both a keyword and query report, and you need to prevent yourself from adding negative keywords that cut off traffic when the more specific keyword is disabled due to low search volume. (While I don’t have a script to share for these two strategies, we do have automations for them in Optmyzr if you need some help.)

Conclusion

While I do believe in the idea of granular bid management, I am skeptical about how much we can control what Google does by simply maintaining stacked bids for different match types. But this is a strategy advertisers want to test, and some see success with it. So for those of you who’d like to make managing stacked bidding a little easier, I hope this script will help.

Get the script