You may have read that eBay has declared an end to branded keyword advertising, which has contributed to the newly coined phrase “Google Tax.”
While you may or may not have blindly believed other panic inducing terms, such as “Y2K” and “12/21/12”, it is wise to test whether your branded keywords are profitable before you turn them off.
The goal of this test is to determine the quantity and value of lost branded keyword clicks, conversions, and revenue when branded paid search ads are not presented to prospective customers.
Paid Search Branded Keyword A/B Test steps
1. Determine equivalent test populations
The first step is to segment your branded paid search audience into two reasonably equivalent populations in terms of absolute number of conversions over time. There are several ways to accomplish this segmentation:
- Geolocation — Within a specific country, are there two equivalent sets of states, metro areas, or cities that produce equivalent conversions?
NOTE: When dealing with accounts that are limited to a single metro area, you may be tempted to split between “city name” and “zip codes within that city.” This method tends to skew the data since Google location specificity is inconsistent and probably tied to how the user connects to the internet (i.e. via a static IP at work, a dynamic IP at home, a mobile network, etc.) It is still possible to segment this way, it just takes more test phases to eliminate location/connection type bias.
- Dayparting — Are there two equivalent sets of hours that produce equivalent conversions?
2. Create two new Adwords Campaigns
Next, create two new Adwords campaigns which will cater to each of your equivalent test populations and name them “Brand Test A” and “Brand Test B”. This step is critical to ensure that existing campaign historical data does not artificially skew the results.
Each campaign should be identical in terms of ad groups, keywords, ads, ad scheduling, etc. — the only difference is the population served.
When setting these test populations, make sure that you also add the negatives to each campaign as well. For example, if “Brand Test A” advertises to cities 1, 2, and 3, you will need to add negative locations of cities 4, 5, and 6 to this campaign as well to prevent possible cross-contamination of the data.
3. Establish “Baseline” metrics for the branded keyword test
Upon completion of setup, pause your original branded keyword campaign and run both new campaigns until you have at least 100 conversions for each campaign. This step ensures that the new campaigns are “up to speed” in terms of ad approval and impression share.
4. Run Phase 1 of the branded keyword test
Pause all keywords for “Brand Test A” (not the campaign itself). Run this phase of the test until you have at least 100 conversions for “Brand Test B”.
5. Run Phase 2 of the branded keyword test
Now enable all keywords for “Brand Test A” and pause all keywords for “Brand Test B” (not the campaign itself). Run this phase of the test until you have at least 100 conversions for “Brand Test A”.
6. Analyze the branded keyword test results
The Key Performance Indicators (KPI) for this branded keyword test are:
- Overall Website Revenue
- Natural Search Revenue
- Paid Search Revenue
Since we created two new Adwords campaigns for our test, it is easy to view the results in Google Analytics by creating the following custom segments:
- Brand Test A ( Advanced Segment Link )
- Brand Test B ( Advanced Segment Link )
- Brand Test A & B ( Advanced Segment Link )
- Google Organic ( Advanced Segment Link )
- Google Paid Search ( Advanced Segment Link )
Now, using the “Ecommerce Overview” report in Google Analytics:
- Plot the two test campaign segments, “Brand Test A” and “Brand Test B” for the entire testing period
- Plot the “Brand Test A & B”, “Google Organic”, “Google Paid Search”, and “All Visits” for the entire testing period
Every business situation is different, and the analysis of the test results will vary from industry to industry and company to company. The most important question to answer is:
Is it profitable to advertise on branded keywords?
To answer this most important question, the following questions must be answered first:
- How much revenue was lost during test phases 2 and 3 compared to the baseline phase for Paid Search, Natural Search, and Overall Website visitors?
- How many advertising dollars did you save by not advertising on branded keywords in test phases 2 and 3?
- Given your business’ profit margin ( NI / Gross Revenue ), what profit would your business have made from that lost revenue ( lost profit = lost revenue * profit margin )?
- If your company’s advertising dollars saved is greater than the lost profit, does the lifetime customer value of the lost customers outweigh the difference?
What are the results of your branded keywords test?
If you found this article helpful, please comment below with your results (keeping your company name confidential of course).
Helpful information to include would be industry, pretest branded keyword spend per month, revenue lost during test, and ultimately, whether your company decided to continue advertising on branded keywords or not.