Day 1 of #SMX Advanced includes keynote addresses from Google and Bing, conversion optimization best practices, audience targeting tips and social + PPC discussion.
SMX Advanced 2017, Day 1
QuanticMind is at Search Marketing Expo Advanced in Seattle, WA to pick up on the latest trends and recommended tactics in paid search from industry experts. Here’s our list of the quick hits:
Day 1 Keynote – Welcome & SEM Keynote Conversation: Google & SEM
- Jerry Dischler, AdWords’ VP of Product Management opens the session by observing that “consumer demands and expectations have gone up in the past few years, especially with mobile.” Users expect more assisted experiences within compressing sessions – shorter timeframes in which impatient and busy searchers want to quickly find exactly what they want with a minimum of annoyances.
- Dischler remarks that from his perspective, there are three primary areas of focus for AdWords that are attempting to address these higher expectations:
- Audience targeting – Via methods such as remarketing lists for search ads (RLSA), Customer Match, Similar Audiences and the recently unveiled in-market audience targeting for Search.
- Improved mobile experience – Via methods such as expanding AMP to reduce load times for mobile landing pages and ads on Google Display Network (GDN).
- Measurement & attribution – Via Google’s various attribution models, particularly data-driven attribution (DDA).
- On the topic of Google’s constant promotion of machine learning and automated solutions, Dischler offers that these innovations from Google are being used “everywhere, from automated bidding, targeting, broad match and close match.” Dischler suggests that rather than being a threat to paid search professionals’ livelihood, Google’s intent with machine learning and automation is to make advertisers’ jobs easier by removing the drudgery of manual bid management and allowing advertisers to focus on more-strategic efforts such as focusing on audience targeting and development.
- On the matter of machine learning becoming more of a trend at Google, Dischler offers that this is perhaps simply how things will be. Traditionally, paid search professionals enjoy an enormous amount of control over all variables associated with their work. Dischler suggests that in the future, like with the move to deep neural nets, advertisers will still control the “inputs” of data points for their campaigns, but will have to have faith in the machine learning-powered process that produces the outputs effectively by way of a black box. The upside of this method is to mitigate the impossible task that large-scale advertisers have of managing literally millions of keywords – a job too big for a few human beings to handle.
- In-Market Targeting for Search is apparently the next logical step for audience targeting after RLSA, Customer Match and Similar Audiences, and it the next tool to help advertisers ensure they display the right message to the right customer at the right time.
- Dischler suggests that while AMP is still known to have issues with tracking when compared to responsive formats, the added load speed might still justify the effort. Dischler cites a handful of studies that collectively suggest that 50% of users abandon a website with a loading time greater than 3 seconds, and that each additional second of load time adds a 20% drop in conversion rate.
- In terms of online-to-offline conversions, Google recently announced In-Store Purchases metrics, which anecdotally allow users to research and purchase items in less than 30 seconds, due in part to the fact that in the US, for 70% of all online credit and debit card transactions, purchaser information is already on file with Google. However, Google takes individual user privacy seriously and is unable to see any individual user purchase information or credit card info – only tracking info for aggregate purchasing activity.
- On the matter of Apple’s recent announcement that the publisher would block third-party trackers in Safari, Dischler expressed concern, stating that he prefers cookies as a tracking method because they enable user choice, while other methods, such as fingerprinting, do not give users choice. Dischler suggested that Google will continue to attempt discussions with Apple to come to an understanding that lets advertisers track conversions on different sites.
Day 1 Keynote – Welcome & SEM Keynote Conversation: Bing & SEM
- In response to questions about where LinkedIn will potentially play a role in search engine marketing, Bing Ads general manager Steve Sirich explains that while there are no specific announcements to make at present, the future of Bing and LinkedIn together is bright, and will eventually address the supply constraints that some advertisers are experiencing with Bing Ads.
- Bing recently announced Dynamics 365, its big data CRM service, to begin integrating features across Microsoft’s properties, including LinkedIn.
- In terms of future product and company strategy, Microsoft refers to itself as a cloud-first and mobile-first company that sees a strong future for itself through partnerships and multiple ecosystems. The company continues to pursue partnerships, such as syndication deals with Oath and CBS Interactive, as well as powering Apple products such as Siri and Spotlight. It also continues to explore supporting external ecosystems such as iOS and Android, developing such apps as Cortana for iOS.
- Sirich suggests that artificial intelligence (AI) represents an exciting new frontier of possibilities. According to the 2017 Internet Trends Report by Mary Meeker, by the year 2020, 50% of all searches will be voice-based. Sirich also pointed out that machine learning and AI are already being actively explored, citing Bing’s well-known research into parlaying search results into predictive diagnosis of pancreatic cancer.
- Sirich suggests that Microsoft has donned the mantle of “fast follower” to keep pace with the innovations of Google and other publishers – recently releasing a Dynamic Search Ads beta, continuing to support Enhanced CPC, and now heading into audience targeting tools that will resemble Google’s Custom Audiences, In-Market Audiences and Similar Audiences.
- This panel opens with discussion of exact match close variants – a functionality that debuted on Google in 2012 to allow phrase and/or exact match keywords, became mandatory in 2014 (requiring advertisers to use negative keywords) – and most recently had exact match close variants announced in March 2017.
- The long and the short of exact match close variants is that, for the time being, there don’t seem to be significant performance deltas before and after using them, except possibly on tablets – a less popular platform now that tablet bidding modifiers have been broken out from desktop modifiers whose CPC and conversion rates tend to be far less favorable.
- Discussion continues by once again debunking the frequently assumed link between causation and correlation, in this case, with a look back at the massive drop in life expectancy in the US circa 1918. While two significant social events – an outbreak of the Spanish Influenza and World War I – both seem like likely culprits, a closer inspection of the data reveals that men did not disproportionately die younger in larger numbers than women – which would suggest that the war was the primary cause – so the epidemic likely was a stronger contributing factor here. In the same way, the behavior of customers – and of SEM campaigns – should not simply be attributed to educated guesswork, but rather, should be scrutinized scientifically.
- The panel’s discussion turned toward scientific testing across large sample sizes for Google Shopping campaigns, testing thousands of products and queries over controlled time periods.
- These scientific tests led to conclusions such as:
- Across 2,000 products, more than 60% of the time, among competitive Product Listing Ads (PLA) listing, the product with the lowest total list price appeared in position 1.
- Not having a seller rating can preclude a seller from taking position 1, even if its product has the lowest total list price.
- Endlessly raising bids on Shopping campaigns eventually leads to huge, wasteful traffic spikes for irrelevant queries and matches that convert poorly.
- This panel opens with discussion on the potential power of optimization, which arguably fuels such powerhouse companies as Google, Netflix, Facebook and Amazon.
- The process suggested involves six steps to turn optimized conversions into explosive growth:
- Strategic business alignment
- Having a “culture of optimization”
- Customer-focused approach
- Data driven research
- Scientific testing process
- High-velocity testing
- Effectively, the suggested methodology requires forming hypotheses, granular testing of individual factors in isolation, data-driven quantitative research and conceptual qualitative research to understand “why” things happen.
- The session continued by citing studies that suggest that only 13% of all A/B tests succeed – a poor track record that’s even worse than a 50/50 coin flip.
- A suggested methodology for improving conversion optimization is a multi-step testing process that involves the following:
- Heuristic analysis
- Qualitative analysis
- Quantitative research
- Competitive analysis
- Identifying problems
- Conversion framework analysis, such as the LIFT model
- Prioritizing problems
- Creating a conversion roadmap
- Creating a hypothesis
- Creating new website designs
- Conducting A/B testing
- Post-test analysis
- This panel opens with discussion of strategies to effectively use GDN ads to target the right people. While a more-traditional method is site-based targeting (showing GDN ads on websites that appear to have relevant keywords), this method is arguably inferior to audience-based targeting, which attempts to “follow” people who show possible purchasing behavior.
- Different types of audience targeting that can help your GDN ads reach relevant audiences include placement targeting (pre-loaded with websites you already know are relevant to your products and services); topic targeting and affinity targeting to zero in on relevant areas of interest; audience targeting to “follow” prospects across different websites; and in-market targeting to find prospects likely to buy.
- Different tools that are relevant here include AdWords Display Planner to compile lists for placement targeting; Automated Placement reports to exclude irrelevant websites for topic targeting; and Target Exclusions to avoid showing ads on any webpages featuring specific keywords or categories.
- Despite its low volume, Customer Match has anecdotally shown to reach highly relevant customers and deliver stronger clickthrough rates (CTR), conversion rates (CVR) and better overall return on investment (ROI).
- Specific tactics for Customer Match include targeting unsubscribers with Google Sponsored Promotions emails, or cross-selling opportunities to sell add-ons or accessories to prospects who are known to have purchased the base product (such as selling a phone charger to a prospect who has purchased a new phone). These campaigns tend to show higher CTR and average CVR.
- When running Customer Match campaigns, it’s recommended to adjust messaging to relevant audiences; avoid stereotyping audiences (a recent study suggests that 40% of women don’t relate to the women they see in ads); and to follow all Google guidelines, such as refraining from explicitly calling out factors such as religious beliefs or marital status in ad copy.
- Customer Match’s reach is sadly still limited, accounting for only 5% of all advertiser clicks. While Google has been observed to be able to match Gmail addresses to searchers at a rate of 90%, its match rate for Yahoo or AOL addresses is closer to 50%, and its match rate with .EDU or corporate emails is closer to 17%. Those businesses lucky enough to have significantly large Customer Match lists are encouraged to also experiment with layering Customer Match onto RLSA to find exceptionally relevant leads.
- This session included discussion on how integrating social campaigns with paid search campaigns can lift performance in terms of traffic, CVR and the ever-important social metric of engagement.
- Social and paid social can have cross-channel impacts; a massive shock in Facebook impressions has typically been observed to lead to an increase in Google brand search impressions, peaking 3-4 days after the initial Facebook activity.
- It’s worth noting that Facebook’s bidding and available testing methods differ greatly from AdWords. For instance, Facebook lacks auction dynamics that can be predictably manipulated with bid changes – in other words, bidding up simply to gain scale at the cost of efficiency no longer works. This is why audience segmentation is exceptionally important on Facebook.
- Currently, the most conversion-focused bidding strategy for Facebook seems to be Optimized Cost Per Mille (oCPM), which tends to result in higher CPC and lower CTR, but focuses on conversions.
- Tests have shown that while mobile tends to account for about 1/3 of spend for many advertisers’ paid search and social accounts, mobile paid search tends to evince an 11% lower return on ad spend (ROAS) from desktop, while mobile Facebook campaigns show a massive 2.3x higher ROAS than desktop.