Enhanced CPC vs. Manual Bidding Test
Many search marketers may be afraid to admit it, but we aren’t always perfect. There are always some things that can be done better but great search marketers know when to admit these faults and allow for a little help. Recently we learned that Google AdWords had recently adjusted their enhanced CPC bidding strategy to be more effective. For those who aren’t aware, AdWords describes enhanced CPC (or ECPC) as “a form of AdWords Smart Bidding that uses a wide range of auction-time signals such as device, browser, location, and time of day to tailor bids to someone’s unique context.” ECPC uses about half your campaigns’ traffic to make bid adjustments of up to 30% of your Max. CPC bid. All of this is done with the purpose of driving more conversions and/or improve efficiency.
Our initial thoughts when hearing more about the ECPC bidding strategy were very positive. The fact that AdWords would leverage many signals (like browsing history and relevancy) to make real-time adjustments seemed like a perfect fit for many of our client accounts. We decided to test ECPC bidding with one of our e-commerce client accounts, where we optimize the campaigns using first-click attribution from Google Analytics (currently ECPC works off of AdWords data only).. Below we discuss how we set up the test and our initial results and findings.
How to Set up Your Test
Currently we handle most of our bidding efforts through manual adjustments. We leverage historical, seasonal, and many other data points to help make effective bidding decisions. So we decided to test the enhanced CPC bidding strategy directly against our standard manual bidding. This required us to set up a test in AdWords leveraging the Drafts & Experiments tool to set up an A/B test. The goal was to allow half of our traffic to be only affected by ECPC adjustments and the other half to only be affected by manual bidding. Results for each traffic segment would then be reviewed against one another.
To get statistically significant data, we ran this test for 4 weeks. The length of time will vary client to client, but based on our client’s traffic volume we felt 4 weeks was a sufficient length of time. We also decided to test in various markets, both domestically and internationally.. Our hypothesis was that because of its ability to leverage numerous data points, the ECPC bidding would outperform manual bidding.
Initial Results & Findings
While the ECPC bidding strategy can be set to optimize for either AdWords conversions or Google Analytics conversions, we decided to have ECPC optimize for AdWords conversions. This is because we currently optimize our campaigns for first-click interaction from GA which can’t be imported to AdWords. We were interested in seeing what type of results we’d get so we moved forward with the test anyways. Please find our test results below:
Domestic Market Results:
International Market Results:
When looking at Google Analytics data, the manual bidding strategy significantly outperformed ECPC bidding. For the domestic market, the ECPC segment conversion rate was 41% lower than manual bidding and the international market’s conversion rate was 30% lower than manual bidding.
When looking at Google AdWords conversion data, performance varied between markets and bidding strategies. For domestic, conversion rate was mostly flat but international was oddly up 39%.
What was most interesting was the fact that front-end metrics like impressions, clicks, and CPC were very even between the two bidding strategies. These were very questionable results because the front-end data was so even between the two testing segments, but there was significant variation when referencing the eCommerce data.
Since our test was relatively inconclusive, we decided to end the test for the holiday season to maximize this account’s performance. We plan to relaunch this test in more domestic and international markets in Q1 and update our findings. We will also be opting into cross device conversions and extending the period we run the test for. After internal and external discussions, these factors along with seasonality could have led to unfavorable results for the ECPC bidding strategy.
Another important consideration when evaluating ECPC, is the attribution model the conversion data is leveraging. Currently, ECPC only has the capability to optimize using AdWords’ conversion attribution or GA’s last non-direct click attribution. Both attribution models are limited and don’t value the source of discovery. Therefore we are focused more on GA’s first-interaction, which ECPC can’t leverage. This is a major reason why we have a hard time trusting the results of the ECPC test.
If you were interested in learning more about the enhanced CPC bidding strategy, please look out for our next blog scheduled to be published next month. For all other inquiries, please contact us by email at firstname.lastname@example.org or by phone at 781-591-0752.