What does the rise of voice mean for paid search?

But what does the explosion in voice search mean for PPC?

From a paid search perspective, at first glance, it's a little bit hamstrung. There isn't a report in AdWords or Bing that says "this search was typed; this search was voice".

At Croud we're increasingly seeing search queries that look like voice search. One obvious giveaway is "ok google" - where someone has attempted to invoke google's magical assistant, got frustrated and said it again. On the second attempt, google picks this up as a query - for example, “ok google - where is my nearest KFC".

We're also seeing the volume of keywords within a search query grow - another indicator of how users are shifting to voice search.

In addition, with many voice searches happening on smartphones, search queries involving “near me” are increasingly popping up; similarly it’s worth looking into queries like "when does", "how much does", "what kind of" or "how to" being used across searches.

Attention then needs to be turned to the device. It might not be widely known; but it is possible to segment search query data by device in adwords.

This then gives a wider range of interesting data to work with. But interesting information can be a killer in this industry. "That's cool" doesn't sell more or make things better. "That's cool because I can do X with it and that means Y for my business", is where we need to be at.

Steps to be taken

Firstly, understand whether a match is made to these queries. If not, there are things that can be done, for example, looking at the account structure. Are broad match modifier keywords being leveraged properly? Are dynamic search ads being run?

Examine the type of queries that are appearing (impressions); what is driving engagement (clicks) and what is driving conversions (sales, leads, or enquiries)? Then start to look at the intent behind the search queries. Search terms such as "when does X shut" suggests the search will drive an in-store visit. If things are optimising towards that - using AdWords Store Visits - it could be worth bidding on.

Search terms such as "how much does X cost?" might be one to let the SEO team tackle. These searches are in the research phase and might not be worth paying for, unless other areas are exhausted and aren't ranking well.

"Near me" is another term that might drive great conversion, even online. If they can get something delivered and the conversion process is short, it can be more attractive to a searcher than actually going to the store.

Going through this process can help dictate which terms that can be paid to appear, which terms can be ranked organically, and what to ignore (for now).

The next step is understanding what to do with this from a paid search perspective. If it is a pure-play online, matching these terms and working well to a target cost-per-action, it might not be worth breaking out. The only other reason to break it out is ad copy - and for that there needs a reason to be different. If not, keep the structure simple and improve reporting and forecasting flows!

If there is a real reason to differentiate at a local level, then look to break this out, in order to improve engagement with the ad (manifested as click-through rate) and the conversion rate by providing a local relevant landing page.

Keep an eye on voice

At Croud, we've been testing the theory of targeting these types of queries in a specific campaign. We've closely reviewed search query reports over a period of time and identified what we believe to be voice-specific. We've then isolated these terms into their own campaign and given a heavy weighting to mobile bid adjustments.

For one client over the last quarter, we saw cost-per-action reduced by 24% for these campaigns versus standard search, along with a 1% uplift in conversions.

It’s a tiny volume at this stage, and difficult to pinpoint as precisely voice search, but it’s a promising area and something we’re continuing to keep an eye on for our clients.

Martin Reed is a PPC account director at Croud