In the fall of 2019, the PPC Twitter verse was overwhelmed with a question that has been around since the industry existed: Is it worth bidding on branded keywords? At that time we wrote extensively on this topic. For a detailed summary, see Matt Umbro's “The Great Brand Keyword Debate: What You Need to Know”.
As with many questions, the answer to the question of whether or not to bid on branded keywords can be frustrated in two words: "It depends". For some brands, bidding on their brand keywords can be an integral part of their marketing strategy. For others, spending on these keywords may be as frivolous as investing in Juicero packages. In this post, I will outline a strategy that advertisers can use to help themselves and their customers isolate and measure the value of brand advertising: geo-exclusions.
How the analysis works (and an example)
I will describe how this test is set up in Google Ads and Google Analytics. However, we can imagine how the basic process works for all advertising platforms that you can use to exclude and measure specific geographic areas. If you work in Google Ads, you can set up the experiment by excluding a representative geographic area from the brand campaign whose value you want to measure. If you advertise in the United States, I would recommend dividing the states into two groups of 25, so that their population is the same and a group of states is randomly selected to exclude them from the future campaign.
If you've excluded this geographic area from your campaign, create two segments in the associated Google Analytics account: one for the excluded states and one for the remaining states. In the future, you can now measure the relative increase / decrease in your KPI for each group for combined organic and paid searches. The important thing is that you want to make sure that you measure the increase / decrease from time to time rather than just the total number of conversions for each channel. This should help control confusing variables that can affect total paid or organic traffic.
For example, suppose you run the campaign with geo-exclusions for 30 days, with a group of 25 states eligible to see your trademark ads, a group of 25 states that were not eligible. In this example, we can imagine that there is only one goal that the advertiser is interested in. To give this hypothesis a certain texture, read the table below with the results that were created:
These data suggest that for those countries that have been excluded from the branding campaign, the increase in organic traffic conversions almost completely compensates for the loss of goal achievement through paid traffic. In practice, this would mean that the vast majority of users who clicked on an ad and converted it would have converted through organic channels anyway. In such a case, one might reasonably conclude that launching the brand campaign was probably not worth it, as the impact on net conversions appears to be negligible.
Other factors to consider
The above example has been simplified to the essentials for illustration. In practice, there are many factors that should be considered when setting up this analysis and evaluating your data. These factors include:
Can you afford a 50% geo-exclusion? Perhaps you strongly suspect that paid traffic is a strong driver for branded traffic that you would otherwise miss. If so, I would recommend starting with a much smaller geo-exclusion. As long as you measure the net increase / decrease, you should still be able to gain knowledge from measurement groups of different sizes. If the results of the smaller exclusion indicate that branded traffic is not as valuable as you imagined, you can always switch to a larger geo-exclusion to get a larger sample.
Pro tip: If you want to start with a single state that is excluded from your analysis, Illinois has been named the most representative state in terms of demographic composition. When it plays in Peoria …
How much profit is associated with each brand conversion? Perhaps your analysis suggests that most (but not all) paid brand conversions would occur organically if your paid campaign stopped running. Whether you want to end the campaign as a whole or not depends on the profit associated with each conversion. For example, if each conversion brings a profit of $ 1,000 and the CPL is only $ 0.05, it can be extremely effective to miss a few conversions yourself, and it may make sense to keep running the campaign at full steam.
What confusing variables might have sneaked through? However, this type of analysis will never be a perfect scientific experiment. Therefore, when interpreting the results, you want to consider as many confusing variables as possible. For example,
- The brand advertisements may have contained a special that was not yet reflected in the organic result for your customer. If this were the case, it could affect the relative profitability of organic conversions and paid conversions.
- Perhaps there were offline advertising campaigns that affected different geos. For example, if what you've promoted has been the subject of a regional television campaign, it could juice up the Bio-Lift in certain areas and affect your results.
- Even non-geospecific campaigns on other channels can influence the interpretation of the results. If there were a huge mailer campaign that affects both your included and your excluded geos, that campaign could create a brand elevator so large that it drowns out the change you actually wanted to measure.
Ultimately, this type of analysis is unlikely to answer the question once and for all of whether brand campaign spending is actually creating value or not. But hopefully it can bring some data and nuance into your research into this question. Every advertiser faces different challenges and opportunities. Therefore, it is time to end dogmatic platitudes such as "bidding on brand terms is worth it because they are cheap anyway and competitors may do it if we don't" or "never bid on a brand because you are for them anyway." should have a high rank ”. Instead, be open about each case and try to collect data to support your later strategy. Geo-exclusion analysis is one way to do just that.