4 Simple Lessons To Make CPA Bidding Work For You

I’m a huge fan of Google’s CPA bidding system. Setting bids is necessary; but merely setting them only leads to short-term progress. Your bidding work is only useful until the data changes, and then you have to set bids again.

The maxim, “your data are only good until the data changes,” can be applied to ads, landing pages, placements and any data point within your PPC campaign.

The issue that needs to be examined, here, is the frequency of data changes and how that affects your strategy. Typically, ads do not need to be changed every few days. An ad or a landing page can perform well for long periods of time. The same cannot be said of bids.

Rarely do you have an optimal bid set for a keyword for a month straight. So, when you consider areas of your campaign where you can make long-term gains, they are not in bidding. CPA bidding is a short-term gain, but a necessary action; and when done right, it leaves the PPC manager free to spend more time focusing on these long-term gains and less time on bidding.



However, I often find a simple mistake with CPA bidding, or even with 3rd-party bid systems. Once CPA bidding (or a 3rd-party system) has been enabled, the PPC manager sits back and thinks his or her work is done.

For the purpose of this article, we will leave out all of the testing you should be doing regardless of your bidding methodology, and focus on making CPA bidding work for you.

I find that when CPA bidding fails, it does so for one of three reasons:

  1. The campaign’s conversion data is sporadic by keywords and ads. An example is when you have thousands of keywords; but only 20% of them have received a conversion in the past 30 days; however, since all contribute to total conversions over the course of a year, you can’t really delete any.
  2. “Best Practices” of account management are ignored because the bidding is being taken care of by Google.
  3. It just doesn’t work for totally unknown reasons. I do see this happen on occasion, where everything is set up and managed perfectly, but for unknown reasons, the CPA bidding just can’t seem to get the bids correct.

Let’s look at four scenarios in which CPA bidding initially failed, but ultimately succeeded after applying some minor changes.

Scenario 1: Data Problems

This first example is one that amazed me. I was auditing an account that was using CPA bidding and discovered that the company had failed to set up conversion tracking on their mobile site. However, the campaign was set to all devices. It wasn’t immediately noticeable because desktop performance was high enough to ensure that CPA bidding had enough conversion data to keep running.

After using CPA bidding for three months, 25% of all traffic was still coming from mobile devices.

CPA bidding does take devices into account when setting bids. However, it rarely ‘gives up’ on a device; instead, it keeps trying to find a bid that will work. By just adding the conversion code to the mobile site, CPA bidding becomes much more effective.

The Lesson: Make sure all your tracking is set up correctly.

Scenario 2: Using Call Extensions To Create Goals

The second example features an e-commerce site. They were B2B e-commerce, so they did use the phone extension, as calls often converted into sales; however, they were all sales on the phone that were not put back into analytics to see the actual revenue per conversion.

When this company upgraded to an enhanced campaign, they liked the fact that they could count calls as conversions and thus, used the ‘report phone call conversions’ option in their account.


They continued to receive phone calls, but their CPAs climbed considerably for all e-commerce goals and were well above their target CPAs. While CPA bidding didn’t technically fail in this case (they were getting the calls, after all), the e-commerce manager was quite unhappy as the overall site e-commerce was declining, and there wasn’t any data to show them exactly what data points were generating the calls.

They disabled the option to report calls as goals. After the disabling, they still received calls (as they did before going to enhanced campaigns); however, their CPAs went back to their target goals, and all the conversions were actual e-commerce checkouts

The Lesson: If you are going to add additional conversions for CPA bidding, make sure you really want the optimizer working off of those goals.

Scenario 3: Keyword Expansion

The next lesson comes from a company that used CPA bidding for months. They were very much enjoying the bid system, and they had put so much faith into it that they just kept adding keywords and thought Google would figure it all out.

Every month, their CPAs went up; but not by enough that anyone was motivated to investigate. They just assumed it was bid pressure and kept adding more keywords.

After several months of expansion, it was time to give their quarterly report to the VP of Marketing. The CPA trend worried her, so she asked for a larger time frame for the CPA trend. Once she saw the CPA climbing for several months, she asked for a 3rd-party investigation.

The answer was quite simple. They were adding keywords, but they were not paying attention to the search queries of those keywords. Just by adding a few hundred negative keywords, the CPAs quickly returned to an acceptable amount.

The Lesson: You must still follow best practices of account organization, match type selection, query analysis, and adding negative keywords — even when using CPA bidding.

Scenario 4: Ad Copy Testing

The next example comes from a company that is great at landing page testing, but decided it was time to start doing more ad tests. So, they created a program for testing their ads, wrote lots of new ads, and put them live into their account.

Their CTRs almost doubled. But their conversion rates dropped nearly by half, and their CPAs rose more than 30%.

The problem? They were using Google’s default ad serving option: optimize for clicks.


If you are going to test ads in CPA bidding campaigns, you have two options:

  • Control everything: use rotate indefinitely, watch the data, pick winners, delete losers, repeat.
  • Know you’ll forget to end tests: in this case, if you are going to create multiple ads and then forget about them, use ‘Optimize for Conversions’ for your ad testing. With this method, Google will pick the ad with the best conversion rate and show it more frequently.

The Lesson: CPA bidding does not serve ads; it sets bids. If you are going to test ads — and you should — make sure you are using the correct ad rotation settings.


I find that more often than not, CPA bidding is highly effective. There are times when it fails, but that now seems to be the exception, even for low-conversion accounts.

Now, CPA bidding is great when you have a static CPA target for all keywords in each ad group; however, many e-commerce sites have a target ROAS instead of a target CPA. In that case, CPA bidding is rarely the best bid method to use.

Regardless, no matter how good CPA bidding is for you, if you don’t continue to follow best practices for optimizing your account, CPA bidding can often become ineffective.

Just because you have an automatic bidding system — either Google’s CPA bidding or a 3rd-party bid management system — that doesn’t mean you can stop working on your account. Those systems change bids based upon the system inputs. If you give them bad data, they will make bad decisions.

Using automated bid management is great. It gives you back the time you would have spent setting bids so you can make sure your account is optimized. However, you can’t abandon your account when you use such a system — you must still continue to follow best practices.

Opinions expressed in the article are those of the guest author and not necessarily Search Engine Land.

Related Topics: Channel: SEM | Google: AdWords | Google: AdWords: Enhanced Campaigns | Paid Search Column


About The Author: is the Founder of Certified Knowledge, a company dedicated to PPC education & training; fficial Google AdWords Seminar Leader, and author of Advanced Google AdWords.

Connect with the author via: Email | Twitter | Google+ | LinkedIn


Get all the top search stories emailed daily!  


Other ways to share:

Read before commenting! We welcome constructive comments and allow any that meet our common sense criteria. This means being respectful and polite to others. It means providing helpful information that contributes to a story or discussion. It means leaving links only that substantially add further to a discussion. Comments using foul language, being disrespectful to others or otherwise violating what we believe are common sense standards of discussion will be deleted. Comments may also be removed if they are posted from anonymous accounts. You can read more about our comments policy here.
  • Pat Grady

    “however, many e-commerce sites have a target ROAS instead of a target
    CPA. In that case, CPA bidding is rarely the best bid method to use.” — Brad, you build segmented, granular Ad Group PPC like we do, and it’s easy to set the CPA there to hit the desired ROAS. So this sentence confused me. Elaborate?

  • http://twitter.com/bgtheory Brad Geddes

    What if keyword 1 brings in $10,000 revenue from 1000 clicks in month 1, but the average order value changes in month 2, and the keyword brings in $20,000 in revenue from 1000 clicks, then in month 3 it returns to $10,000. In month 2; you’d want to double your CPA or use ROAS bidding.

    Or, you have times there are high seasonality changes. Consider flowers when the average order value sometimes drops (depending on the florist type) during mothers day/valentine’s day – but the conversion rate is much higher. So, now you don’t want to bid by CPA but by average order value connected to conversion rates; or you can constantly take ROAS and turn it into CPA bids, but its extra steps.

    As seasonality affects cross sales, upsells, and average order values – setting a static CPA often doesn’t work month-over-month due to these other influences.

    In some ecommerce situations, average order values are consistent as is traffic, and CPA bidding works just fine. However, there are times that it doesn’t work as the values can change dramatically week over week or month over month.

  • http://twitter.com/ppcstrategies Barb Young

    Brad, thank you for such specific information. We’ve been gradually changing an ecommerce client’s campaigns a few at a time to CPA bidding, and it’s been quite successful at gradually hitting our goal. However, I do notice that if we’re going to make adjustments to the bid, that unless we keep changes to less than 5% of the prior bid, impressions, clicks and conversion really come to a halt. We’ve also found that we get better results using the “target” CPA vs. “max” CPA. Do you have any thoughts on the amount of change and whether target or max is easier to control? Thanks.

  • jeroenmaljers

    Great article Brad. I once mistakenly used a target CPA amount as Max CPA bid. The whole campaign collapsed in impressions/click and eventually conversions and needed some time to recover when when changed back to target CPA. When do you think we should use MAX CPA?

  • http://twitter.com/bgtheory Brad Geddes

    I have a tendency to use max when I’m reselling leads, have a hard cap on conversions, or the average order doesn’t vary too much and I have a true CPA I want to hit.

    I use target CPA when I want the most conversions possible. This is common is b2b ecommerce where most conversions are $500; but the random and unpredictable one will come in at $10,000. So you know there are a few huge orders out there; so we have an overall target, but we also want to capture the big orders but there’s no predictable way to do it.

    If I’m dealing with a sales force where they want the most leads possible to keep agents busy on the phone, then I find target is better than max.

    So, usually if I want the most possible, and I’m willing to pay for conversions both above and below my targets – I’ll use target – its more of a ‘portfolio’ bid method though. If I have hard caps, or static order values; I’m more likely to use max.

  • http://twitter.com/bgtheory Brad Geddes

    Hi Barb,

    My reply to Jeron probably answers a lot about the max/target CPA.

    For Search:
    I haven’t seen issues with traffic in raising CPAs, in fact, I often find that raising my CPA can get me a lot more traffic (of course, this is based on some impression share metrics as well and not having 90%+ IS).

    I do find that lowering my CPAs can get me less traffic, but its usually not dramatic. Although, I do rarely lower my CPAs more than 10% at a time.

    For Display:
    As your CPA dictates your average CPC, which then Google transforms into eCPM; I find that raising/lowering CPAs on GDN can dramatically change my overall conversions as you will often end up switching the sites where your ads are being shown. So, for display, before changing CPAs, I’ll often add all the best placements as managed placements so I don’t lose that traffic before I change the bids.

  • http://www.LeadDiscovery.com/ Jerry Nordstrom

    Brad – I really enjoy your articles because you use real world examples that we all can learn from.

    What are your thoughts on CPA bidding when the conversion event is not entirely online? Many marketing campaigns are designed for service based businesses that seek to generate “prospects” that turn into sales and revenues offline.

  • Pat Grady

    We’re saying the same thing Brad. :-) It’s easy for an experienced manager to see the CPA, AOV, and ROAS columns and make adjustments with aplomb, including switching modes from CPA to ROAS when AOV or CR (or other factors) change. I had thought you were making a distinction ONLY between CPA and CPC bidding (the mechanical choices available) – you were not. :-)

  • http://www.swydo.com/ jeroen maljers

    Thanks. I can add that if you had a cost per conversion of let say $ 15, and you turn on CPA bidding, do not set MAX CPA at $ 15 but a little bit higher otherwise you will see a decrease in traffic.

  • http://twitter.com/bgtheory Brad Geddes

    I still like to use them in that case and just back into the numbers.

    The small business makes X per sale. Their close rate is Y. They want to make at least Z per sale/hour/job; so then your target CPA is $Q.

    That’s a very simplistic look; you can get more complex with call tracking, instore coupons, etc; but at a high level – I still like doing CPA bidding with SMB lead gen.

  • http://twitter.com/bgtheory Brad Geddes

    Perfect – glad we’re on the same page :)

  • http://twitter.com/PPCNI Jordan McClements

    Hi Brad.

    I always use Target CPA.

    But for 1 client recently (well, 2 actually, but I’ll stick with one to simplify), the CPC and costs went *crazy* one day then declined quickly back to normal levels. Conversions went up but nowhere near enough to justify the extra spend over a couple of days (sort of hard to explain to client). There were no changes made on the account that could have caused this.

    Speaking with Google (of course they would never admit there was a possible problem even if there was), apparently this is normal behavior etc. etc. etc. though apparently this sort of thing can’t happen if I’m using Max CPA bidding instead of Target CPA bidding.

    So my question is (I’ll get there eventually), if I change from Target to Max, and accept the default Max CPA bids to start with, do you reckon that the number of conversions will remain more or less the same as with target CPA but without the very scary occasional spike in costs and CPCs?

    If this is the case, then what possible reason could there be for using target CPA bidding?

    (It would be nice if you could run an AdWords experiment to see what actual difference this makes..)

  • http://twitter.com/PPCNI Jordan McClements

    By the way, apparently you are still not guaranteed no scary spikes with Max CPA bidding but it should still minmize the likelihood (somewhat, maybe) :-

    ” For either Max CPA
    or Target CPA there is not a set amount of time that the system can go
    above the amounts set. The goal is to get you a conversion within the
    CPA that you set, not over a certain period of time.

    Your actual CPA depends on factors outside Google’s control so it’s possible that your actual CPA may exceed your maximum CPA bid on campaigns using Conversion Optimizer. The
    Conversion Optimizer uses historical conversion data to predict the
    likelihood your ads will convert. However, your actual conversion rate
    can be affected by changes to your website and ads, or external factors
    such as increased competition. If your actual conversion rate turns out
    to be lower than the predicted conversion rate, your actual CPA may
    exceed your maximum CPA bid.

    tend to recommend setting your Max CPA higher than the historical cost
    per conversion so that the system has room to adjust your bids in order
    to try and get you the desired conversion rate.”

  • http://www.facebook.com/tombutlin Tom Butlin

    Hi Brad
    I have a client who runs large public venues and wants to target event managers and conference bookings. I have a massive negative keyword list and very specific ads to minimize consumer enquiries. When I moved to CPA the conversions plummeted for 3 weeks as the traffic was mostly consumers wanting to find out the opening hours, parking, prices etc. I was unable to adjust enough within CPA to compensate and reverted to CPC, with an immediate return to normal conversion rate.
    I have just moved another client to CPA and after 6 days, the conversions have dropped from an average of 1.2 per day to exactly 0. Same ads and keywords, optimised for conversions. I’m seeing a similar number of impressions & clicks per day.
    How long do you give Google to find the sweet spot? I’m running out of patience and will revert to manual in another day.
    Many thanks

  • http://twitter.com/bgtheory Brad Geddes

    Hi Tom,

    If CPA is tanking, I usually won’t give it more than 2 weeks before I turn it off.

    If its a smaller account, I might give it up to a month assuming its doing OK (not great; but also not gettig 0 conversions) just to give it some time to work.

    I do find that with sparse data if the same queries receive the conversions (or some small subset) month over month that it usually works.

    If the queries that get the conversion change every month (not uncommon with small accounts) that sometimes CPA bidding just never works as Google can’t find repeatable conversion information to work from.

    If its a large account (so CPA bidding will be really useful); and the CPA are OK, maybe even a bit higher than target, but it would be a huge win to get it to work – then I might give it up to a month.

    So, no more than a month, but no less than a week is usually the overall guideline.

  • http://twitter.com/tombutlin Tom Butlin

    Excellent, thanks Brad.
    My client has around 40 conversions per month, so that should be more than adequate, given that the threshold is 15. I have returned to manual bids. If a different client has 50 or ideally 100 conversions a month, I will look into it again.
    If you’re ever in New Zealand, I’ll buy you a coffee.


Get Our News, Everywhere!

Daily Email:

Follow Search Engine Land on Twitter @sengineland Like Search Engine Land on Facebook Follow Search Engine Land on Google+ Get the Search Engine Land Feed Connect with Search Engine Land on LinkedIn Check out our Tumblr! See us on Pinterest


Click to watch SMX conference video

Join us at one of our SMX or MarTech events:

United States


Australia & China

Learn more about: SMX | MarTech

Free Daily Search News Recap!

SearchCap is a once-per-day newsletter update - sign up below and get the news delivered to you!



Search Engine Land Periodic Table of SEO Success Factors

Get Your Copy
Read The Full SEO Guide