Search Audiences

It surprises me when I run across digital advertisers that aren’t utilizing paid search audience targeting, particularly those spending very size-able budgets. Audience targeting within search offers endless possibilities to prioritize reaching the most opportune consumers, and truly cater a journey specifically to them. Most search campaign strategies I’ve worked with were built from an initial audience strategy. 

First, Look at it From a Consumer’s Perspective

The first step though is to step into the shoes of the consumer, and truly understand what it is they are seeking. Keyword spraying campaigns are inefficient, and the best way to lower CPCs, higher CTRs and efficient CPAs is being surgical with who we show ads to and making sure we make the most of that exposure. The technology available to us today, as digital marketers, offers endless possibilities and strategies. But without a solid strategy – it’s just technology with endless possibilities. 

An easy first approach is simply segmenting our audiences. Start at higher levels, and segment deeper within the segments once we have more results and when it makes sense. But to begin, simple segments could be existing/returning customer vs new customers, geo segments, gender segments (particularly important in retail based on different products).

For example: perhaps your existing customers return purchase at a high average rate. You could use existing customers as a separate audience target in your campaigns, and make higher bid adjustments against that audience segment. Or, perhaps you want to bid on more expensive head terms for different products only with this existing customer audience, since you know they’re more likely to convert. 

It all starts with understanding your audience though. Map out your audiences, this is particularly important with vast ecomm sites that have multiple different product categories and items. 

Adding in Other Tech to Cater Journeys

DSA & Inventory Management (SA360)

Dynamic Search Ads allows Google to utilize landing page content to generate new keywords your campaigns bid on. This can be a powerful tool that can drastically reduce the work needed to search for new keywords to bid on. It can be a little scary though just allowing DSA to run free on generating keywords to bid on. So it’s important to mine your DSA campaigns, and eliminate under-performing keywords DSA has generated.

Inventory Feed Management (SA360) is a very powerful tool, similar to DSA. Except, instead of scraping your landing pages to generate keywords, this tool scrapes inventory feeds you supply to generate relevant keywords. (You must have SA360 to use this feature.) This is incredibly powerful for online retailers and listing sites – keeping your keywords campaigns up to date based on inventory availability, and allowing your search copy to reflect inventory descriptions and pricing. But this isn’t just for retailers – it can be creatively applied in other industries to. For example, imagine a subscription streaming service – if they load their series and movie catalog into SA360 as an inventory feed, they can generate ads to those looking for titles searchers are looking to watch.

Keyword-Less Campaigns

By combining DSA or Inventory Feed Management with Search Audiences, you can develop truly keyword-less search campaigns. While DSA and Inventory Feeds are generating keywords (and you’re mining under-performing keywords out), you can focus on which audiences to show ads to. My team did this for a client a few years back. They were an auto listing site, and the plethora of year/make/model keyword combinations was infinite. So we loaded their auto inventory feed into SA360, and let it generate relevant keywords. Then we optimized bidding at a geo-audience level, and inventory level. 

So for example, if someone in Boston was searching for a 2013 BMW X3, but we only had 3 available in that market – we wouldn’t serve an a. (We know less than 5 listings wasn’t enough to generate a lead.) Yet when someone in San Diego was searching for 2015 Hyundai Sonatas, and we had 9 in listings in that market, we would serve an ad. 

We also layered on ad customizers to make the ad copy much more specific “9 Hyundai Sonatas available in San Diego starting at $8,000”. 

GA 360 Audiences

In a recent post I talked about the power of GA Audiences. These are easy and efficient audiences to implement within search. Understanding your audience and creativity are both important here:

  • If you’ve integrated DCM data into your GA 360 instance – you can adjust search bids on display exposure before someone visited your site (post-view)
  • Elevate bids on re-targeting audiences. Go deeper, elevate bids higher on abandons who had something in cart vs those who didn’t.
  • For brand terms, suppress search ads on those who have been to the site – or maybe only those who have come to the site previously via organic search.
  • Suppress purchasers from future served ads, or suppress those who have come to the site 3 or more times through paid search. (Suppression of under-performing search audiences in often overlooked easy efficiency tactic.)

These are just a few examples. The audience feature within GA is a mini-DMP, and really allows audience segmentation at very granular site behavior levels. The possibilities are endless, so it’s important to start with larger more general segments, test and learn, then segment deeper. 

Audiences from Offline Data

Ingestion of offline data, particularly modeled data, is a more advanced, but incredibly effective, way to leverage Search Audiences. I spoke about Segmenting LTV for Better efficiency in a previous post, and these LTV segments can certainly be applied as separate search audiences to adjust bidding against. The same could be done with Propensity Scoring, where you’d bid higher against those more likely to convert. 

Adwords offers it’s own version of Look-a-Like model targeting, called similar audiences for search. But if you have the in-house capability, building a prospect database and doing your own in-house LAL modeling would give you more control over how you’re tweaking the LAL model. It would also give you more insight into which similar attributes are the most predictive of likelihood to purchase. Combined with DSA or Inventory Feed campaigns, this can be an incredibly powerful use of offline data with search audiences.

Google’s Own Audience Targeting

Google offers it’s own audience targeting capabilities based on it’s own data. I already mentioned similar audiences, but Google also offers search targeting capabilities based on demo, affinity (which I’ve usually found too broad), custom-affinity (ability to layer affinities for deeper custom segments), and in-market audiences (based on search history). 

Pulling it All Together

The multitude of targeting capabilities can be overwhelming. Which is why you should never start with the how. Start with who you’re trying to reach, and then figure out which of the targeting capabilities is the best “how” to reach your intended audience. Whiteboard out your target audiences, what you think they’d be searching for and what steps they take to purchase – and begin segmenting audience targets from there. 

Couple last bits:

  • Often times I’ll simply segment audiences, and run them without bid modifiers for a time – to prove out my theory is accurate. If one audience outperforms the other as expected, I’ll then apply the bid modifiers. 
  • Audience layering is IMPERATIVE. These capabilities aren’t exclusive of one another, so by combining audience tactics and layers you can more effectively reach granular targets. Start at the high level with the most important segmentation, then segment underneath to layer.
  • Constantly test and learn. Form hypotheses, test them and then adjust as necessary.

Hopefully I’ve sparked some ideas in you with search audiences. If you have anything to add, or anything I missed, please leave a comment!