Brad Geddes / PPC Geek
Official Google Ads Seminar Leader.
Author of Advanced Google AdWords.
Co-Founder, Adalysis.
(703) 828-5811‬
Brad Geddes's Theories on Marketing Case Study: Quadrupling A Small Account’s Conversions in Just 90...

Case Study: Quadrupling A Small Account’s Conversions in Just 90 Day

CPA Bidding in Google AdWords allows you to set a CPA bid and then Google will do all the bidding for you. When it works, its fantastic as you don’t have to spend all day trying to set bids inside the account. Instead, you can focus you time on more permanent improvements, such as testing your ads and landing pages.

When CPA bidding first rolled out in 2007; you needed at least 300 conversions in the past month for it to work. Over the years, Google has improved the algorithms to the point in 2010 that you only needed 15 conversions in a campaign over the past 30 days to enable this setting.

However, I always found that no matter Google’s minimum conversions, the tool failed more often that not if you didn’t have roughly 150-200% of the minimums needed – until very recently. Now, I’m seeing a different story with CPA bidding.


Small Account CPA Bidding Case Study

I was working on an account where most campaigns were averaging 15-20  conversions per month. The data was sporadic and inconsistent. One month certain words would get a conversion and the next month a completely different set of words would get a conversion. There were very few words that would receive a conversion every single month; and the words that did have these consistent conversions were often higher than the target CPA for the account.

Typical Month


The target CPA was $65 per conversions. Only one campaign had exceeded this CPA. Every single other campaign was significantly higher than this goal.

To turn this account around, I went through several steps:

Step 1: Build a New ‘Optimized’ Campaign

The first step is to take all the data and create a very granular organized campaign that used the best features of the data from the past several years. This was to be a benchmark campaign of what was possible.

Due to some very high bids that were risky to try in a new campaign and items working ‘well enough’ in the other campaigns, some of the other campaigns were tweaked a little to show the difference between well organized and the typical campaigns; but were left to continue running to mitigate any transition risk.

This is why many columns talk about organization and they are right to do so, just going through this step increased the total conversions significantly (50 to 112).

Month 1 Data:


Tweaking some settings, such as adding negative keywords, removing poor keywords, finishing some ad tests, moving some budgets around, etc resulted in the account more than doubling it’s conversions; and lowering the CPAs. However, the ‘new campaign’ with the ‘proper’ organization immediately hit the CPA goals while brining in an additional 18 conversions, which is a large increase since the account only had 50 conversions in the previous month.

Step 2: CPA Bidding

In the second month, some minor tweaks were made; but the biggest change was turning on CPA bidding for a few campaigns.

Month 2 Data:


Once CPA bidding was enabled, the conversion rates went up by 0.7%; which is a nice; but not huge increase. The CTR dropped a touch, which is not uncommon to see. However, the total conversions increased by 40 with a $10 reduction in CPA. That’s what CPA bidding can often do for smaller accounts. No longer are the conversions hovering at the 15-20 range where you might not be able to use CPA bidding, but they have increased enough wh

make money online

ere Google does have more data to work with.

Step 3: Ad Rotate: Optimize For Conversions

I’m a huge fan of ad testing, and usually use one of the ‘rotate’ features for testing purposes. However, this was a small account and I wanted to see how good the ‘optimize for conversions’ would be in a small account. In addition, I had consistently heard from people who had tried these small CPA tests before that it was often working for a few weeks; but then the volume dropped so much they had to stop using the biding option. In order to keep building the conversion numbers so that wouldn’t be a trap; I then enabled ‘Optimize for Conversions’ in the ad settings:


Now beyond that; other things were being done inside the campaigns, such as adding negatives, removing some very very poor performing words, testing extensions, and so forth; but the largest change was the ad serving.

The ‘optimize for conversion’ setting had an unexpected result – it not only increased the conversion rates, it increased the CTR in a few campaigns:

Month 3 Data:


The ‘well organized’ campaign amazingly went from a 5.64% CTR to a 10% CTR with optimize for conversions. Other campaigns increased in CTRs as well; however, not to quite the same levels. This seems so strange to me that I’m still testing this setting in some other campaigns to see if this is somehow a  common side effect. As CPA bidding is suppose to just set bids on keywords, and ‘optimize for conversions’ is suppose to just serve ads; it seems very odd to see a CTR difference of this magnitude.

The overall conversion rates went from 3.24% to 5.09%; with the ‘well organized campaign’ going from a 2.97% to a 4.99% conversion rate.  That’s a very nice increase since it only takes 30 seconds to change the ad rotation setting from one option to another one.

The other dramatic change was the PLA campaign went from a 5.77% conversion rate to a 11.30% conversion rate. That’s an amazing difference in a PLA campaign where very little work is being done on the management side.

The Results

This is the overall results of one good campaign (the other campaigns weren’t touched outside of some basic items such as adding negative keywords and manipulating the budgets) and conversion optimizer during a 90 day period:

  • Conversion rates: Increase from 1.94% to 5.09%
  • CPA: Decreased from $101 to $44
  • CTR: Increase from 3.21% to 3.46%
  • Total budget: Increased $4,600
  • Total conversions: Increased from 50 to 216 per month

Most of these campaigns are also limited by budget, so the next step is to open up the budgets and see with just budget manipulation if the total conversions can more than double again (at the same CPA). After that’s done; it’ll be time to reorganize some of the older campaigns, test serving the ads on more hours of the day, and of course, do some landing page testing.

I’ve now tested CPA bidding in a few smaller accounts; and I’m seeing similar results. Even with 15-20 conversions, conversion optimizer is working more often than not; and if it works early on, then the total number of conversions often goes up causing it to be even better in later months as it has more data in which to work.

CPA bidding being successful for small accounts is a huge change from last year when it failed more often than not . No longer do the advanced tools such as CPA bidding have to be relegated to just large accounts (where it almost always consistently performs well) – small accounts can try some of these more advanced tools as well.

The total amount of time put into this account’s rise to success is roughly 20 hours over 90 days. Even if no other changes were made in the account to help conversions, the account is looking to spend about double in 2013 than in 2012; but with over quadruple the conversions. That’s a pretty good 20 hours of work; and its mostly due to proper organization, an execution and testing strategy, and Google’s CPA Bidding option.

If you have struggling small accounts, test CPA bidding and ‘optimize for conversions’ ad serving, the results might surprise you.



  1. giorgiagatti86
    February 8, 2013 at 3:30 am ·

    hi brad
    can you explain more in detail what it means in practice:
    “Step 1: Build a New ‘Optimized’ Campaign”

    for you what is the best structure for an account accoun medium / small ?

    divide adgroup for match type keywords or not

    it is better to join Exact and Phrase match type – and instead put in its own right Broad Modifcate to discover new query?

    is a better adgroup for each keyword or keyword about how many in each adgroup?

    obviously the question referred in the best ratio quality/time for an account of about $ 1,000 / month

  2. brad
    February 11, 2013 at 8:33 am ·

    Hi @giorgiagatti86- I think I answered some of this in the premium members group.

    However, for this account – what we did was think about the ad and the landing page as usual; but then go one more step in the segmentation so the ad was very relevant.

    For instance (and this is not the vertical type; just trying to show how this worked); lets say your an accountant. Now, you often find queries like: Chicago accountant, b2b accounting, accounting services, etc.

    What we did was go one more level down in the ads and keywords. So instead of ‘Chicago Accountant’; we had that query plus all the neighborhood and other regional terms. So if you searched for ‘Rogers Park Accountant’ – you saw an ad that mentioned the geography. For words like ‘accounting services’ or ‘tax prep’, etc – we didn’t stop there, we did it by vertical; so like ‘restaurant accounting’; ‘plumber tax prep’, etc.

    Now that ads were just one step more relevant than most of the other ads on the page; and they were very specific,; and this lead to higher CTRs and conversion rates.

  3. brad
    February 20, 2013 at 7:56 am ·

    Hi @williamvicary

    In this case I didn’t because the average position was very high and the CPAs were lower than the target; so I didn’t see how CPA bidding could help out so its the only campaign in this particular account that is not using CPA bidding.

    I have done CPA bidding with PLAs and in most of the cases, I haven’t seen an issue; but I have seen a few where the PLAs were doing great and CPA bidding made the campaigns worse (like you’re seeing).

  4. williamvicary
    February 20, 2013 at 7:45 am ·

    Brad, did you use CPA bidding with the PLA group? We enabled this for a client a little while ago and found traffic stopped rolling through the account!

  5. jelle
    May 5, 2013 at 11:32 am ·

    Hi Brad,

    I’m in the process of testing a small enhanced campaign with cpa bidding.
    It has accumulated two months of data with 25 conversions per month on average with manual bidding and “rotate indefinitely” ad settings.

    Am I right that the longer a campaign is live the better cpa bidding should work because there is more (normalized) data for Google to work with?

    Should I simultaneously turn on “optimize for conversions” in the ad settings like you did or should I wait for a month to see if cpa bidding works with the “rotate ad indefinitely” setting and if it works then turn it on “optimize for conversions”?

    What should be best practice? (for a small campaign/account)


    • brad
      May 6, 2013 at 5:59 am ·

      The longer a campaign is running; to a point, the better CPA bidding is. After 2-3 months; I rarely see it improve a lot (often small improvements, but not huge ones) unless you make adjustments. Adjustments could be ad testing, adding negative keywords, removing poor performing areas, removing bad keywords, testing landing pages, adding new products, etc.

      There are two ways I start CPA bidding.

      If I have lots of historical data and have done some ad tests, then I’ll turn it on with ‘optimize for conversions’ to see if its going to work. This is more a ‘proof of concept’ test to see if CPA bidding will work.

      If I don’t have lots of historical data, then I need to do ad tests and try CPA bidding, so I’ll start with rotate indefinitely and do an ad test. Then, for small campaigns, end the ad test after a month or two (depending when you have enough data) and use ‘optimize for conversions’.

      On occasion, I’ll do another big ad test and switch back to ‘rotate indefinitely’ and repeat the cycle depending on how aggressive I am about testing the ads in a campaign.

      Hope that helps.