SMX West 2017 Day 3 - Audience Targeting, PPC for Brands, Testing and Optimization

SMX West 2017 Day 3 – Audience Targeting, PPC for Brands, Testing and Optimization

Day 3 of #SMX West brings with it new tactics for audience targeting, insights on PPC for brands and important reminders on testing best practices.

SMX West 2017 Day 3 - Audience Targeting, PPC for Brands, Testing and Optimization(pictured: @michellemsem)

SMX West 2017, Day 3
QuanticMind wraps up Day 3 of one of the biggest search marketing conferences of the quarter with more sessions coverage from #SMX West, including targeting, brand SEM insights and testing best practices.

Advanced Audience Targeting
Featuring: Austin Denny, @juliavyse@michellemsem

  • The ad industry is changing – every major data management platform (DMP) launched since 2004 has been acquired.
  • Ads themselves appear to be trending away from ubiquity (being present everywhere) and towards granularity, but in many cases, still lack specific understanding of a prospect’s full range of interests – the kinds of things that drive prospect along their increasingly convoluted customer journey.
  • Austin Denny: “Because Google and Facebook control so much of the advertising landscape, In the future, the biggest wins will not be from SEO or individual channel tactics, but realized by advertising teams that are able to access rich, clean user data to strip out the noise of multiple data sources and figure out how to deliver meaningful experiences to the most relevant prospects. For this, the roles of information security and legal will become increasingly important to advertisers.”
  • Customer match lists have traditionally been considered to be remarketing tools, but they are also a potentially powerful negative exclusion list that can be excluded from top-funnel user acquisition prospect lists to narrow down potential customers and save precious spend.
  • Other tactics, such as YouTube remarketing, are also highly underused. YouTube offers powerful and impressive demographic and engagement statistics:
    • More 18-34 and 18-49 year olds are reached by YouTube than any cable network in the US
    • YouTube has more than 1 billion active users
    • The average time per session for YouTube users is 40 minutes
    • YouTube also has the power of Google research and development behind it, including support for virtual reality, Google Analytics 360 and live streaming
  • YouTube offers plentiful segmentation options, including targeting users that have viewed specific videos, visited a specific page, liked a video, subscribed to a playlist, and many others. These options let advertisers push viewers into remarketing lists. When combined with keyword intent, this can be a powerful user targeting tool.
  • Gmail Promotions ads offer additional targeting options for competitor conquest, including not only the ability to find users that have received emails from competitors, but also users that have responded to competitor emails with “unsubscribe,” “unsatisfactory,” and other keywords that suggest a negative experience with the competitor’s product or service. This offers a potential “in” to aggressive marketers who feel confident they can speak directly to the pain points of churned competitor customers/prospects.
  • Household income, an advanced geo targeting indicator which is currently buried in the AdWords interface and must be tested manually, has been available since 2014 but is still rarely used. However, layering demographic information can provide powerful segmentation insights.
    • Example: An advertiser ran a campaign in New York City, finding that its brand terms were converting for households in the top 10% income brackets and also the bottom 50% income brackets. However, it noted that its non-brand keywords converted more strongly for households in the 20-50% income brackets – leading to the important insight that, depending on the household income, different ad copy and messaging would be needed.
  • There are two main reasons to exclude prospects from your campaigns – either a specific prospect is an unwanted target, or is someone who is better targeted with a more-specific strategy that can be formulated using audience shaping (query mapping).
  • There are several reasons a prospect might be unwanted – they might be already known and in your funnel; they might show poor engagement (such as high bounce rate from your website); or they may not fit your desired customer persona.
  • Using publisher audience pattern tools is recommended to help create positive and negative audiences and compare differences: AdWords’ and Facebook’s tools share the name “Audience Insights,” while LinkedIn’s tool is Demographics.
  • Once audiences are identified, it’s important to perform audience shaping by using query mapping to ensure your messaging matches the right audience, that you’re targeting effectively, that you can manage and optimize your marketing, and also to manage frequency control (to avoid bombarding audiences).
  • Unfortunately, while it’s theoretically possible to segment audiences, it’s quite common to see different audiences overlap, and without a meticulous segmentation approach, it’s also quite common to end up sending more than one type of messaging to the same prospect; for instance, your most relevant prospects might be part of your retargeting list, but you may also have messaging specific to certain job titles – but due to audience overlap, you may end up sending both messages to the same prospect, which can be a confusing and non-ideal customer experience.
  • To avoid this type of overlap, it is recommended that marketers create a hierarchy of audiences in order of importance by means of audience size, highest value targets and most-specific messaging, and “exclude down” that hierarchy. Example:Segmented for retargeting — Segmented by industry — Segmented by job title — Lookalikes
    <– HIGHEST PRIORITY                                                                                           LOWEST PRIORITY –>

    In this case, a marketer would start by putting together its retargeting list, then exclude the retargeting list from his/her industry segmentation, job title segmentation and lookalike list. Next, the marketer would take the industry segmentation list and exclude from the job title and lookalike list – and so on.



How Brands Test Their PPC Ads
Featuring: @bgtheory

  • For some extremely large brands, some search queries are brand names but not search-specific; thought of as a product or action, but not a true brand term. Example – using a “Kleenex” rather than a “tissue,” or going to “Google something” rather than searching for it.
  • Especially large brands often contend with other logistical problems, such as keeping its legal, brand and design teams in sync. 
  • It should be noted that extremely large brands tend to be slow to test for numerous reasons not simply limited to a lack of agility – in many cases, testing may require the approval of many stakeholders and unseemly search query results may go so far as to affect the share price of a publicly traded company.
  • As such, when a large brand does agree to testing, it’s important to make sure test metrics are aligned with business goals and will genuinely provide usable insights. 
  • Other important considerations are confidence factors for term types to determine whether performance changes are due to actual test factors or random chance; as well as ideally using single ad group tests as opposed to multi-ad group tests, as single ad group tests let marketers zero in on one specific factor and test it extensively. 
  • As you might imagine, taking an organized approach to testing is also important. At minimum, it’s a good idea to create a spreadsheet that documents key information such as keyword terms, term types (brand or non-brand), where the ads appeared, ad types, specific metrics, teams worked with, flight dates and resulting insights.


Maximizing Performance By Testing And Optimizing Paid Search Ads
Featuring: @pedropar

  • Understanding that SEM relies on testing dynamic data, unlike laboratory research, which tests static data, it’s important to begin testing with a clear hypothesis that outlines exactly what will be tested, what the expected result is, and why – this will inform the best, clearest results.
  • It’s also important to get as granular and specific as possible when testing – for instance, when trying out a new call-to-action in a headline, it’s important to test either the copy or the placement, but not both at the same time. Introducing as little variance into tests as possible makes it easier to glean clearer insights more quickly.
  • Segmentation for audience types and funnel stage is also important.
    • Example – for travel ad copy,  using “Book the best vacations” vs. “Book top-rated vacations” may appeal to different audiences in different stages of their customer journey. Top-funnel prospects may simply be looking for whatever is deemed “best,” while lower-funnel prospects who are actively researching purchases may be more interested in spending the time to read user reviews. However, top-funnel copy may bring in more clicks from top-funnel prospects but fewer conversions, since top-funnel prospects may simply not be prepared to buy.

For more in-depth session coverage of SMX West 2017, please check our previous coverage:

Andrew Park

Andrew Park is a content marketing manager at QuanticMind. A UC Berkeley graduate and lifelong Bay Area resident, Andrew has done tours of duty in editorial, PR and marketing, and now works with the QuanticMind team to communicate the importance of data science and machine learning in digital advertising.