Optimising PPC for Voice Search: A Thought Experiment
Is 2017 the year of voice search? I would argue that dates are as relevant as they were for the ‘year of mobile’, but facts are a plenty, and undoubtedly voice search will become increasingly prevalent as tech giants develop their virtual assistants and increase integration with relevant devices.
Here are just a few stats regarding voice search:
- According to a 2014 study commissioned by Google, 55% of teens and 41% of adults use voice search more than once a day
- By 2020, 30 percent of web browsing sessions will be done without a screen (Gartner)
- 37% use Siri, 23% use Microsoft’s Cortana AI, and 19% use Amazon’s Alexa AI at least monthly (HubSpot)
So, if voice search is arguably becoming the new norm, I can’t help but wonder what could be done today (using the current tools of the trade), to create and optimise a paid search campaign exclusively for voice search.
The Thought Experiment
First of all, it’s essential to understand that voice search queries are inherently different from typed search. They are more conversational and tend to be structured as a full sentence, often shaped in the form of a question, using words like: why, who, what, which, where, when, and how. This form calls for the optimisation of long-tail keywords.
While I’m not sure that voice search was something the creators of Answer the Public had in mind, this free visual keyword research tool is the ideal tool to help build up a keyword list in this context. The company produces a report of what people are searching in Google and YouTube and displays the results in the form of questions, prepositions and alphabetical listings. Here’s an example for ‘smartwatch’:
It provides an extremely comprehensive list of keywords, that at first glance (ignoring match types for now), could be used to create a campaign structure such as the one below:
However, this may not be the best structure solution in regards to budget management and/or return on investment.
When diving deeper into the results, it becomes apparent that the type of question asked is generally indicative of the stage the consumer is at in the search funnel:
With this knowledge in mind, a more sensible approach would be to structure campaigns and ad groups according to the customer journey and, obviously, the overall marketing objectives. Awareness campaigns will probably include the “why” questions, with “where” questions asked closer to the bottom of the funnel. As such, budgets and keyword bids will then need to be adjusted accordingly, with a higher ROI for “where” queries.
However, not all voice searches are questions, and so the structure highlighted will not cover all relevant traffic. Still, the fact remains that most searches will be closer to natural speech, and full sentences should occur more frequently. As a result, this is where DSA campaigns have the greatest opportunity to perform.
Dynamic Search Ads, instead of using keywords to target ads to searches, use content from your website to help fill in keyword gaps not included in already built campaigns. When a match is found between the content and the search query, it automatically generates an ad headline with an appropriate landing page. When going through their search query report, I found plenty of long-tail queries, probably more than those found in ‘regular’ search campaigns. If you have a content rich website, DSAs can perform quite well when structured correctly…but I’ll have to dive deeper into DSAs in a future post.
So, thus far, my though experiment tells me that as a starting point, I would create campaigns in a long-tail-question-type-keywords structure, grouped according to the different stages in the search funnel, with the DSA campaign(s) capturing the traffic not covered by the Answer the Public initial keyword research.
Following best practice, the ad copy should answer the concerns of the search; something that remains true whether you’re optimising for voice or for typed queries. So, depending on the stage of the search funnel, the ad copy (continuing to use the smartwatch example), would be created to state the following:
Needless to say, the corresponding landing pages should properly address all possible questions.
What other targeting options are relevant for voice search?
Not all voice searches happen on mobile, but possibly the largest share do. This highlights the importance of ad extensions (particularly message and call extensions) that offer users the appropriate action for the device they’re using. It also hints at making bid adjustments at device level from the get go. Although device bid adjustments should be done based on statistically significant data, I wouldn’t be too shy of raising bids for mobile/tablet when initially setting up campaigns. And, although I wouldn’t go as far as to set bid adjustments according to time of day, it makes sense to assume that voice search volume is higher during ‘on the move moments’, such as commuting hours.
Caveats and conclusion
I haven’t gotten into many other considerations that could render all of the above extraneous – their impact on any digital marketing strategy is too large to touch. There are many factors that make voice search focused PPC campaign impracticable to a great extent: a lack of a mobile-friendly website, a lack of a content rich website (particularly in the case of DSAs), a lack of cross-device conversion tracking, as well as the fact that it’s not yet possible to know which queries in the search query report are actual voice searches.
Nonetheless, the objective of this post is not to linger on the realistic plausibility of every aspect of PPC for voice search, but to enable and encourage a mindset that accommodates and anticipates upcoming changes in the world of search marketing. Hopefully I have achieved this.