SMX East 2017 Day 1 – Online-to-Offline, Shopping, Display

SMX East 2017 Day 1 - Online-to-Offline, Shopping, Display

Day 1 of #SMX East offers the keynote address from Google on online-to-offline (o2o), Shopping best practices, Display best practices and more.

SMX East 2017, Day 1
QuanticMind is covering the latest trends and tactics in paid search and digital marketing at SMX East 2017 from search publishers and expert practitioners. Here’s our list of the quick hits:
>> Quick Links to Day 1 Session Recaps

Day 1 Keynote – Google Unveils New Online-To-Offline Ads Innovations
Featuring: Kishore Kanakamedala of @Google

  • Google’s Director of Online-to-Offline Solutions Kishore Kanakamedala opened SMX East 2017 by discussing some of the changing trends in online-to-offline (o2o) conversions for retailers, then highlighted some of the new and upcoming tools that will be available for local search marketing.
  • Some statistics about online-to-offline conversions:
    • 90% of retail sales take place at physical storefronts
    • E-commerce has grown 23% to traditional retail’s 6%
    • 3 in 4 customers who conduct a local search on their smartphone visit a local business in the next 24 hrs – these customers are more purposeful and likely to make a purchase
  • To answer the question, “How do I know people who are visiting my  storefronts are influenced by online ads?” Google ran an independent study with 140+ advertisers across multiple verticals. It found that investing in Google search ads drives an +80% higher rate of incremental store visits
  • Google also found that mobile generates 160% more incremental visits compared to desktop or tablet.
  • On the matter of in-store purchase behavior, since final purchases are what matter to retailers at the end of the day, Google found that customers who click on search ads before visiting a store were 40% more valuable, and spent 10%+ more on average.
  • Toward the goal of continuing to build innovations to drive more foot traffic for stores, Kanakamedala recommends acquiring more omnichannel consumers by:
    • Building engaging ad experiences that reach consumers when they’re looking to buy, when they happen to be in the area.
    • Using cross-channel metrics to measure store visits and sales.
    • Taking action on this data with “actionable optimization” through such channels as search, display, Maps, Shopping and YouTube.
  • These are some of the new tools Google has recently introduced to drive more o2o conversions and sales:
    • Ad units such as Local inventory Ads (LIA) – the “See what’s in your store” module was recently added to Maps and search to help shoppers find what’s in stock, and can browse the most popular products and categories.
    • New local ads for display – These spotlight store-level products and promotions in richer and visually engaging ways to signal upcoming clearance sales and new product launches, similar to a print catalog or circular.
    • Affiliate location extensions for search – These launched last year for the US, and are now available in such territories as the UK, Canada and others. They are also available for display. These extensions give more cross-channel reach to your ad listings.
    • Growth of Google store Visits – It has been about 3 years since this was metric was introduced. It’s now available in 20 countries (including Singapore, Turkey and India). It has also been added to Display and YouTube.
    • Distance reporting – This tool was introduced last year to show how far away searchers are from your storefronts while they search.
    • Geographic reporting – This tool was launched this year, and tracks, down to the ZIP code level, and tracks who is driving the most clicks.
  • Using these tools, advertisers across verticals like such as retail and telecom measured 7 billion store visits in AdWords. Here’s how this was done:
    • Google’s o2o tools include store visits metrics, which track:
      • Location history – Managed by first party opt-in to preserve privacy
      • Mapping – The same technology that powers Maps/Earth/Streetview
      • Deep learning – Upgraded versions of Google’s machine learning models, which track GPS, Wi-Fi strength, bluetooth beacons, location history, search queries and visit duration
      • Survey verification – To verify collected data
  • These are some of the new tools Google has will unveil in the near future drive more o2o conversions and sales:
    • Impression-based store visits for display – In the coming months, this tool will track moments when a customer views a display ad, keeps your products top-of-mind but visits a store later
    • New reporting tools:
      • Demographic reporting – This will show which groups are more likely to visit stores.
      • Time lag reporting – This will show how long it takes users to visit stores after an initial ad click – essentially, how quickly an online ad influences an offline action.
      • New vs. returning customer reporting – This will show how quickly ads as proxies generate visits. If a large number of new customers visit in a few days after a click, this metric will show a higher correlation than visits that take place months after a click.
      • Measuring full-funnel store performance – These will measure the funnel “all the way down to the final sale.”
    • Store Sales Direct (beta) – This tool will measure how ads drive in-store purchases. The Indian retailer Shoppers Stop observed 3x more sales when it included in-store transactions.
  • Google has observed that store visits where the journey starts on Google are 30% more valuable.
  • At AdWeek last week, the company announced improved integrations with Datalogix and Nielsen, expanding Datalogix integration to include 6-second video bumpers on YouTube.

>>Back to top

Bing in the Holidays
Featuring: @cadycondyles,

  • Cady Condyles of Bing Ads recommends that merchants consider budgeting beyond Cyber Monday/Black Friday, since 40% of shoppers have been observed to do the bulk of their shopping after Black Friday.
  • In fact, product ads on Bing still show decent traffic post-Christmas. This may be a good time to run ads for accessories to top products sold that holiday season.
  • Bing’s findings also indicate that retailers may wish to ramp spend starting in early November to capture early-season shoppers.
  • The US Department of Commerce found that for Q1 2017, online sales growth was 5.7x that of offline sales, but 87% of sales still took place at physical stores.
  • To drive more sales at physical stores and convert more online activity into profitable store visits, Bing recommends observing the differences in purchase patterns between online and physical store shoppers.
  • Spikes of online and offline store spend have been observed to converge on major shopping days such as Black Friday, Veteran’s Day and Green Monday. In-store sales – but not necessarily e-commerce sales – have been observed to spike on the Saturdays in December leading up to Christmas Eve, with an especially big spike on December 23 for last-minute shoppers.
  • To gauge how paid search impacts offline sales, Bing recommends tracking store transactions tied back to the date of the click that originated them. Some Bing tools to measure this include:
    • Campaign planner – For seasonal traffic spikes
    • Automated Rules / Email Alerts – To track ad spend and ensure budgets don’t get exhausted before high-traffic days (leading to the undesirable result of “going dark” on Black Friday or Cyber Monday)
    • Shared Budgets – To redistribute budgets across campaigns
  • Bing recommends using both text ads and Bing Shopping Campaigns (BSC) together. A recent case study showed one merchant recorded +71% clickthrough rate (CTR), +76% conversion rate (CVR) and +22% return on ad spend (ROAS).
  • Bing also has several newly-released, or soon-to-be-released, features to aid merchants in driving o2o conversions:
    • Local Inventory Ads – These are in a pilot stage for the US only, and like with their Google counterparts, promote products available in stores to nearby shoppers.
    • Shopping tab – This currently US-only feature displays product ads with product filters – GTIN, MPN, brand and other unique product attributes, which increase the likelihood of ads appearing in search results.
    • Flyer Extension and Flyer Carousel – This currently US-only feature is in pilot phase. Last year, the Carousel extension went through a limited holiday beta, and participating merchants saw a 10-13% CTR lift with no observed cost-per-click (CPC) increase. A Bing Merchant Center account is required to use this feature.
    • Product ads in Bing Image Search – This feature was recently released in the US only. Like on Google Shopping, ads now appear on Bing image search, leading to an estimated click increase of +7% to +8%.
    • Product ads in Bing Snapshots – This feature is available in the US and is coming to the UK, and surfaces product ads on Bing Snapshots for product searches done on desktop or mobile (queries that include model number, specific product names, etc. ).
    • Merchant promotions – This feature was launched last year, and is in pilot in the UK, and includes a “special offer” icon at bottom of product ads.
    • Testing Merchant Promotions beyond Product Ads – Merchant promotions are eligible to appear in offer widgets within the Bing Entity Pane for a particular brand or advertiser search.
    • Enhanced CPC for Bing Shopping – Available now in the US, coming soon to the UK.
    • Google Feed Import – This feature is in open pilot in all BSC markets.

>>Back to top

Shopping Strategies That Never Go Out of Style 
Featuring: Jake Favaro of @3QDigital

  • Google Shopping merchants should have largely consistent key performance indicators (KPIs), including cutting CPCs, increasing CVR and increasing average order value (AOV)
  • To achieve these results, best practices include:
    • Using the “Product Type” ” in your data feed – This is because “the front end will only be as powerful as data in your feed.”
    • Arranging account structure to categorize similar products – It’s a good idea to try to match products in similar groups that will have similar performance, such as similar CPC, CVR and AOV.
    • Applying one bid to a categorized set of products – Keeping bids as granular as possible, it’s a good idea to match bids for products with comparable CPC, CVR and AOV – but for products whose CPC and CVR are comparable but have wildly different AOV due to the prices of individual stock-keeping units (SKUs), consider adding SKU value as a fourth granular category. This will keep your products in a similar price range .
    • Account product segmentation vs. data density – In your account structure, it’s a good idea to segment according to click volume, not the number of products in a category, to ensure sufficient data for smart bidding decisions. Merchants that sell very few individual SKUs (100 or less) might even consider SKU-level bids for their highest-volume products in terms of clicks and CVR.
    • Pull SKU level reports – These find the most wasteful products with high spend and low CVR. You can then exclude these troublesome products. These reports can also find VIP SKUs with the highest clicks and CVR. Make sure to exclude any SKUs that were separated out in this process – getting those structural negatives in place will ensure that your poor performers don’t get the traffic your VIP SKUs should be.
  • Brand vs. Non-brand campaigns – Setting brand and non-brand campaigns in the same Shopping campaign will yield a blended efficiency number that likely means you’re overbidding (wasting spend) for non-brand. To avoid this problem:
    • Create two versions of your campaign – First, the non-brand campaign set to high priority, for which you should exclude brand negatives extensively, and which will likely have higher bids than brand. Your second campaign, the brand campaign, should be set to medium priority and will catch any spillover traffic. To avoid overspending, it’s a good idea to scrub for pesky nonbrand queries to exclude, and bid below your non-brand as a safeguard.
    • Brand campaigns sit at the bottom of the funnel. They can almost be thought of as an “attribution piece,” since it’s very common to see non-brand clicks at top of funnel drive clicks to brand.
    • Accounting for non-brand-to-brand conversions can be done with a remarketing list of users that clicked your non-brand Shopping campaign. By creating two duplicate campaigns, the first of which targets and bids on non-brand clickers audience, the second of which excludes non-brand clickers, you can effectively start reattributing some credit from brand clicks back to non-brand and better assign bids. Net-net, this should also let you bid more aggressively on non-brand campaigns.
    • Caveat for Showcase Shopping Ads – Google has verified a bug which will destroy a brand/non-brand split when using Showcase Shopping Ads. Opting out of Showcase means you can’t get your split campaign back – which means that Showcase Shopping Ads should be used with caution.
  • Bidding on Google Shopping – Once your account structure is set, bidding should be simple – bid higher on VIP SKUs, and bid lower on inefficient SKUs. Some additional tips on bidding:
    • Budget caps – Favaro recommends avoiding these in general. Uncapped budgets lead to cheaper CPCs and more clicks, which drive higher CVR at lower CPA. In a non-brand/brand split campaign structure, a capped budget will cause non-brand queries to eventually waterfall into your brand campaigns, which defeats the purpose of making the initial split.
    • Avoid standard delivery – Favaro also recommends not using this to rely on AdWords to tell you how and when a campaign is limited by budget, and instead recommends sticking to accelerated delivery.

>>Back to top

Tailoring Your Shopping Campaign Programs to Your Customers
Featuring: Jared DeSisto of @merkleCRM,

  • Merkle has observed that clickshare for Product Listing Ads (PLAs) has increased across both Google and Bing:
    • Google has seen PLA clickshare rise to 75% of non-brand clicks for retailers, and 53% of clicks  overall
    • Bing has seen Shopping ads rise to 34% of non-brand clicks to retailers, and 20% of clicks overall
    • These increases seem to be due simply to the strong performance of PLAs compared to text ads.
  • Amazon – It should be noted that Amazon is exploring PLAs, and has periodically done so by maintaining a narrow category focus, most recently on home goods in Spring 2017. Amazon clickshare jumped 15% and maintained its share throughout Q3, though the retail giant now seems to be sitting on the sidelines.
  • Showcase Shopping Ads – These image-focused units, which appear in response to a generic product query, are also gaining traction, with traffic increasing by a factor of 5x-10x for some advertisers since January.
  • Optimizing PLAs to capture user intent – DeSisto recommends attempting to capture user intent for PLA campaigns by bidding to query value – a method that is similar to AdWords text ads’ system of bidding by individual keyword. (Note: Keyword-level bidding is not available in Shopping.) To perform this successfully:
    • Bid by groups of queries – Treat bid management for Shopping primarily as bidding by grouping of search queries rather than products/groups.
    • Negatives and campaign priorities – Leverage negative keywords and campaign priorities to do to further segment by query.
    • Look for important trends in Shopping query behavior – Look through top text keywords and Shopping search terms to spot important trends in user search behavior for your store, such as designer names, product attributes, head-heavy terms…and brand names. Query segmentation can maximize exposure and significantly decrease cost for brand terms.
    • Strong and consistent account structure – Solid account structure enables you to scale your query segmentation. DeSisto recommends organizing campaigns similar to retail store layout or website navigation.
  • Mobile and PLAs – PLA spend growth has been observed to outpace other devices: In Q1 of 2017, mobile PLAs saw +62% growth, compared to +31% growth on desktop, and a -23% decline on tablet.
  • Absolute Top Impression – This new metric shows how often your ad appears in Position #1 of the newly-expanded PLA carousel. This position has been observed to have the highest CTR, and can account for 20% to 33% of all traffic for mobile PLAs.
  • Local Inventory Ads – These capitalize on local intent to drive more o2o and more in-store sales. LIA ads drove 40% more in-store visits while being 42% more efficient year-over-year. They also saw a 28% savings in CPC versus traditional Shopping ads and 34% higher CTR than standard PLAs.
  • Audience segmentation – Consider using tools such as Remarketing Lists for Search Ads (RLSA), Customer Match (using opted-in Gmail-based customer databases), Similar Audiences, geo-location and household income (HHI) targeting. It’s a good idea to differentiate bidding strategy to account for funnel stage: site visitor versus product page visitor versus cart abandoners versus actual people who convert.

>>Back to top

Building a Solid Foundation for GDN Success 
Featuring: @michellemsem

  • The foundations of Google Display Network (GDN) ads: There are two sides of the relationship we can target: onscreen content, and audience.
  • To target content, there are a number of contextual tactics that can be used for segmentation:
    • Placements – Morgan indicates that SEM managers can choose individual websites, should keep this list fairly short and can also “dumpster dive” to find placements to target by manually searching for queries and seeing whether resulting pages serve GDN ads.
    • Keywords – This is the “next level up” from Placements, and works by Google finding placements that match keywords you provide. Because this method isn’t very granular, it’s recommended that you treat it like a broad match query, kept to 2-3 words and 5-15 keywords per ad group. Negatives should be used sparingly.
    • Topics – This is the broadest type of contextual targeting. This feature lists large, general groups with related content in preset lists. This highly unspecific method is for getting additional traffic or to reach a broad audience quickly at scale, not to target with any specificity. It’s recommended that this targeting type be combined with other targeting types for those with CPA or budget goals.
  • To target GDN campaigns for audiences, there are also several methods:
    • Demographics – These tap into general user info based on Google profiles and browsing behavior – though due to potentially false entries (such as underage users claiming to be 18 years of age), this style of targeting should be taken “with a grain of salt” and layered with other methods
    • Retargeting – This builds audiences through cookies and email uploads, and targets users as they browse online. It’s recommended to start with an “All Visitors” list, then segment further for different funnel stages, as well as to create lists and structure based on how you plan to message people – new messages should have a new list; new lists should have a new ad group; new calls to action (CTAs) should have a new campaign.
    • Similar Audiences – These are auto-created when Google identifies a persona pattern, but because they’re based on limited retargeting or customer email lists, they won’t provide good results without significant volume.
    • In-market audiences – These are audience lists created by Google based on behavior that suggests they actively researching and preparing to purchase.
    • Similar & In-market Audiences – Using these targeting methods together, they can be treated as highly general retargeting lists. Therefore, they aren’t ideal for campaigns with strict CPA goals.
    • Affinity – These are essentially “in-market audiences, but bigger.” They can be compared to TV ads, which are poorly targeted to big audiences around big topics. They tend to have fewer subset options available than In-Market targeting.
    • Custom Affinity – These work the same as Affinity audiences, but you can customize your audience triggers. They are based on interests or URLs, and are prompted by factors such as brand, competitors, industry publications, interest points and potential search terms. It’s recommended that a minimum of 10 interests are used. This targeting method works best for branding and awareness.
  • Building GDN campaign structure – There are several factors to consider when structuring a GDN campaign:
    • Segmentation of target types by campaign – It’s recommended that targeting types be segmented by campaign, with each topic placed in its own ad group, and image ads separated from text ads in different ad groups to avoid messy crossover.
    • Layered targeting – Combining targets can also lead to more-specific reach, since this method uses an “and” operator to combine context and audience. It’s recommended to enable custom bids for modifier layers and utilize Google Display Planner to surface insights.
    • Smart Campaigns vs. manual management – Google claims its machine learning algorithms power Smart Campaign management, which automatically mixes and matches different creative types to deliver the best results. Morgan recommends taking advantage of the time-saving testing ability of Smart Campaigns to determine best-performing creatives and to consider leaving strong-performing campaigns to run.

>>Back to top

How to Maximize & Measure Performance on the Google Display Network
Featuring: @tedives

  • There are three ways to maximize and measure performance on GDN:
    1. Get reach
    2. Be selective
    3. Attribute value
  • For this, you need lots of creatives and messages, and you need to be in lots of places using lots of targeting methods.
  • Ideally, you’ll have a variety of ad types, including text, static, animated, video ads. It should be noted that video ads on GDN are typically 10% cost of YouTube.
  • For image ads, it’s a good idea to prioritize creative sizes to maximize inventory: such as 300×250 mid-page units and 728×90 leaderboards.
  • Getting reach – There are several strategies to increase your GDN reach. They include:
    • Placements – Getting placed on a specific website, for which you should:
      • Have an idea of which sites to target (such as news sites)
      • Have an evergreen marketing campaign to analyze placements and see where site visitors like to “hang out”
    • Parked domains – Misspelled versions of your domain can act as auxiliary traffic sources; when properly used, they can essentially be considered alternative brand campaigns.
    • Targeting methods – A variety of targeting methods can be used here. Ives recommends:
      • RLSA
      • Placements
      • Topics used with placements
      • In-market audiences
      • Affinity audience + Topics
      • Custom affinity audiences + keywords
      • Similar audiences + keywords
  • Be selective – In order to “be selective” about where you’re placed, one of the most important considerations is brand protection – avoiding an accidental placement on a controversial or explicit website. To do this, make use of Google’s Site Categories to ensure you exclude your ads from unwanted locations.
  • Attribute value – Ives recommends considering not focusing on attribution – which remains an unsolved problem in digital marketing – and instead focusing on “assisted conversions.”

>>Back to top

New Ways to Test Using GDN Traffic
Featuring: Cara DeBeer of @CatalystSEM

  • GDN reaches 90% of users worldwide across 2M websites, yet can still drive CPCs as low as $0.25. Its layered targeting options provide considerable flexibility.
  • While GDN is traditionally used for brand awareness, it can also be used to drive cheap, highly targeted traffic at high volumes with built-in reporting and the ability to grow your remarketing pool.
  • Consider using popular creatives to drive the most exposure – this will collect larger amounts of data and further expand remarketing lists. Text ads remain the most popular type of ad on GDN and an inexpensive way to start for advertisers not ready to invest resources on in-depth creative. Static or animated ads can be very eye-catching, and 15-second videos on GDN can make for extremely cheap clicks.
  • Attribution: GDN has multi-channel impressions reports that can also provide conversion pathing and even multiple attribution models, though many advertisers stick to last-click to focus on the clicks that actually drive sales.
  • GDN and Google Analytics (GA) together – It should be noted that GDN works well with GA, though third-party metrics should also work well enough.
  • When deploying GDN ads, it’s recommended to immediately QA using the Ad Preview tool, then enact website changes as needed, add winning ads to rotation and cull losers, and prepare for the next test.

>>Back to top

What’s Your Audience Thinking? 8 Hidden PPC Reports to Find Out 
Featuring: @bigalittlea

  • To get a better sense of your audience, it’s important to determine demographics,  actigraphics (a picture of their behavior) and psychographics (a picture of their mentality and emotions).
  • Basic AdWords demographic information, such as age and gender, was in beta last year and is now available in the Audiences tab.
  • Checking the Audiences tab to view affinities and targeting types such as in-market, device, location, gender and age can provide valuable insights into your audiences.
  • Geo-targeting can also provide unique insights, such as how far different audiences are willing to travel to certain stores. By mapping SEM performance of geo-targeted ads in specific areas, brick-and-mortar merchants can come away knowing the approximate radius around their stores for which they should bid aggressively, and at which points they should place low or no bids at all.
  • Location of interest – This still seems to be a targeting metric that doesn’t work well; its CPCs and CPAs still seem to be consistently higher, so geo-targeting for audiences in-area may be more efficient than geo-targeting for audiences who express interest in a different area they’re in presently.
  • Day of month – While you’d expect shoppers to be most likely to be searching for things to buy around the typical paydays of the first, 15th or 30th of the month, research revealed that, surprisingly, these periods tend to be among the most expensive for CPA and overall ad spend, so it’s recommended to avoid heavy spends on these days. It should be noted that Google’s new 2x AdWords budget cap will backload your budget if the first half of the month has seen low traffic – if the back half of the month ends up having traffic spikes, there’s a potential risk of having your budget drained.
  • Hour of day – Desktop searchers who ran searches at off-hours seemed to show roughly normal conversion patterns, while mobile searches performed after midnight showed significantly worse conversion rates. This stands to reason if you assume that late-night desktop searches are being performed by searchers with serious purchase intent – serious enough to turn on their computers – while mobile searchers at odd hours were simply bored and looking at their phones.
  • Voice search – Voice search remains data-poor and because there simply isn’t an overwhelming majority of natural language search queries overall, it’s not necessarily an important concern. Yet.
  • Auction insights: Impression share over time – Checking impression share over time, particularly for retail advertisers, can be an exercise in futility, as big-name retailers such as Amazon and eBay will always dominate traffic results.  Segmenting for specific terms and on specific timeframes can sometimes reveal more-interesting and more-useful insights.

>>Back to top

Effective Data Storytelling: The Key to Search Reporting By Michael Barthalow
Featuring: @michaelb3600

  • When presenting analytics to advertisers, it’s important to have effective data storytelling – “stories are just data with soul.” By presenting data in the form of “Problem -> Tension -> Resolution,” it’s possible to impress upon your audience the importance of what you’ve achieved for them.
  • The most important thing to keep in mind for your data story are your audience and the medium through which you’re delivering the story, whether that be through dashboard, presentation deck or video.

>>Back to top

Must-Have Reports for Search Advertisers
Featuring: @DgtlMktgGeek

  • While SEO and SEM are definitely different disciplines, there exists some crossover metrics and insights and suggest they should not be siloed from each other. For example, paid search campaigns might provide insights on which keywords to prioritize in SEO.
  • Cutting CPC with Quality Score (QS) – More importantly, it’s possible to cut your paid search CPCs by improving your website’s QS for important  SEO keywords. These performance lifts can be significant compared to simply focusing on SEM campaigns.
  • While AdWords offers a “Paid & Organic” tab, this report may not give all the insights you need. It does not, for instance, clearly indicate which of your URLs ranks organically and which is ranking in paid. Ideally, the results would align, but this particular tab doesn’t reveal this information.
  • Giddens recommends tying together paid and organic by viewing keyword ranking, ad copy and URL components with conventional metrics tools (such as MOZ, SEMRush, AdWords and others) to determine which keywords offer the most opportunity and immediate upside. By using these metrics to calculate estimated QS improvement savings, you can identify where you’ll get the most bang for your buck – if bumping a specific keyword’s QS from 6 to 7 will give you the most savings, you can begin there and then proceed in the most effective manner, optimizing your websites content and landing pages for that specific keyword.

>>Back to top

Online-To-Offline Attribution: Challenges and Opportunities 
Featuring: @spoulton

  • Poulton notes anecdotally that the huge Nordstorm department store in his neighborhood – a longtime fixture – closed its doors, and was replaced with a tiny “Nordstrom Local” outlet, which catered specifically to local needs. While the new store was immediately ridiculed and not well understood, it’s quite possible that smaller, custom experiences could be the future of brick-and-mortar retail.
  • It should also be noted that while e-commerce is continuing to grow, the overwhelming majority of commerce still takes place at physical storefronts.
  • While o2o is an increasingly important consideration for retail, unfortunately, most publishers, including Google, Facebook, SnapChat and Amazon, remain walled gardens with little transparency into success metrics and hardly any cross-platform analytics to speak of:
    • AdWords – AdWords launched store visits tracking in 2014 to use Wi-Fi, beacons and GPS, but it’s unfortunately limited in several ways, such as offering clear attribution between store visits and purchase and clear strategies to engage visitors that did not make purchases. Google’s metrics are also limited only to the Google ecoysystem – just as Amazon is also limited to its own ecosystem.
    • Facebook – Facebook’s store visits tracking is a close facsimile of what Google offers and shows visits from people who paid attention to ads in last 7 days, or clicked within the last 28 days. This is a long window of attribution, but again, works only for Facebook. The publisher recommends focusing on reach as a key success metric rather than than store visits.
    • Snapchat – The app has seen significant user dropoff despite launching Snap to Store next year, which tracks incremental visitors from app views.
  • The rise of store visit tracking also gave rise to privacy issues, particularly credit card tracking.
  • So how can retail merchants effectively drive o2o across multiple channels? Here are some considerations and strategies.
    • 2015 Nielsen and Facebook study – The media research firm Nielsen found no appreciable correlation between clicks and offline metrics such as ad recall, brand awareness or purchase intent.
    • KPIs – Since the goal is to drive incremental performance lift, important KPIs must necessarily include:
      • New customer acquisition
      • Store traffic
      • Store revenue
    • Testing store locations – It’s important to carefully select small control groups of comparable stores against which to test – by then testing against such controllable factors as average age group of an area and household income, retailers can then perform “stratified sampling” of audiences, ending up with tiered lists of stores of approximately comparable performance. For example, top department stores in high-density, high-income areas such as New York and Los Angeles (“A-tier stores”) might have roughly comparable foot traffic and revenue, while a store in Cleveland might have significantly less traffic and revenue (“B-tier store”).

>>Back to top

Offline Attribution in SEM: Getting From Good To Great 
Featuring: @bigalittlea

  • Attribution is arguably a “pretend” concept, especially online, since assigning click credit to different channels can be arbitrary.  This is why it’s important to not “steal” traffic from one source to another simply to pad out impressive-looking numbers. It’s more important to grow a business than simply pad out a channel – and “directional” attribution is better than nothing.
  • Considering the example of retail merchants, there are different actions that can be used as an attribution proxy, including store visits, coupon codes, keywords to customers, and 360 data pass plus real time LTV modeling.
  • Use case 1: An apparel merchant selling low-priced, impulse-purchase garments.  Driving o2o is relatively easy in this case with proximity metrics and mapping queries to purchase intent (a searcher looking for prices or sales might be more likely to purchase than someone generally hunting for coupons of all types), as well as a scannable landing page and offer button to avoid uer error by store associates.
  • Use case 2: A family photographer selling baby photo shoots, for which no online component exists, even though coupons abound for such services on Groupon and LivingSocial. By partnering with credit companies to pull credit and income information, while also building up relevant CTAs for ad extensions such as “schedule an appointment,” it’s possible to drive more o2o even for such a traditionally “offline” business case.
  • Use case 3: An online mattress seller looking to build overall margins as well as to lift margins on big-ticket mattresses. This could be done by capturing click IDs from Google, gathering relevant info and optimizing bids based on that information. This, combined with “directional” offers such as coupons and flyers and aggressive attribution ever quarter or so could drive the incremental gain needed.
  • Use case 4: A pest control company with 4 distinct user paths: calls from a landing page; an online lead form; calls or text messages from ad extensions; or use of an online scheduler. In a case like this, aggregating all relevant data from all channels is exceptionally important – and might facilitate the need for a call tracking solution in addition to data warehouse functionality.

>>Back to top of Day 1 session recaps

>>To SMX East Day 2 Recap

>>To SMX East Day 3 Recap