Yehoshua Coren - Analytics Ninja (All Things Data 2015)

What to expect in this presentation

• An in-depth, technical review of the core features of the Enhance Ecommerce platform

• Tactical implementation examples and how to use the resulting data

• My current take on the platform.

@analyticsninja

Product Lists

Product List Performance

• Category Pages• Search Results Pages• Cross Sells• Up Sells• Promoted products (like homepage

splashes)

Merchandising

Product Scope

Impression list #1 / product impression #9 / ID, Name, Position

Product Scope• With the product scope, you can send

data for multiple products on a single hit.

• For example, the pageview will now contain data that allows you to answer the question show all products that were displayed on a category page that are almost out of stock and what position were they in.

Import dimensions to enhance data

Custom dimensions to enhance data• Product freshness – season / added to inventory date

• Product level promotional descriptions– On sale, featured item, etc.

• Current product stock amount• Product reviews• Product level profit margins• Product attributes such as color, size, weight,

manufacturer, distributor, warranty

Cross sells

Cross sells

Up sells

Product List Name

Impressions and Actions

• Impressions are meant to model the choices the user can make.

• Actions are meant to model the decisions the user made.

Product clicked

Product list position

What do we do with it?

• Cross Sell / Up Sell performance the most exciting aspect of product list “impression” data– Recommendation engine optimization (@fastbloke)

• Insights regarding merchandising placement within category pages can be gleamed, though sales data is a stronger indicator of when to highlight products (as we shall see later).

• Revenue from a product list view is reported via a “last action” attribution.

Product Detail Views

Product Detail Views

Product Performance Reports

Product Performance Standard Metrics

• Good use of standard dimensions that include 5 levels of product taxonomy and brand.

• Big benefit of providing Cart to Detail and Buy to Detail Rates – many businesses wouldn’t be accessing those metrics. Propensity to purchase!

• For now, those metrics are based on VIEWS instead of unique views, which I believe doesn’t model shopping behavior in the optimal way.

Calculated Cart to Detail

Calculated Buy to Detail Rate

Mix in Product Scoped Custom Metrics

Dimensions vs. Metrics

• Show me all product views that for products that have a profit margin of 40% or higher– CUSTOM DIMENSIONS

• Show me the top 10 profit generating products– CUSTOM METRICS

Enhanced Product Performance Reports

What do we do with it?

• Advertise. Duh!– Product Listing Ads– Keyword bidding

• Product placement on category pages and homepage splashes.

• Help guide merchandising decisions

Overall Shopping / Checkout Behavior

Overall Shopping / Checkout Behavior

Not really different than horizontal funnelsbut DOES make these setups available to more users

Quick segmentation capability from within the interface

Additional Standard Dimension ofCheckout Options

Overall Shopping / Checkout Behavior

Same basic funnel in KissMetrics, but the denominator is “people”

Current issues with checkout funnel

• Rigid – Not all sites follow the same checkout funnel, this just

models the most common use case.– Different payment options have different checkout process

that happen offsite (PayPal).– Users may log in an automatically bypass many of the

predefined steps.

• Sessionized– Purchases process may not be within a 30 minute window

• :-/

Refunds – I don’t have data, but it looks like it’s good!

Internal Promotions

Internal Promotions

Discounts

Discounts

Discounts

Discounts• Track total value of discounts on order and

product level using custom metrics.• Discounts may increase demand, but may not be

increasing profitability.• Make sure to apply the order level discount across

the Product Price for all products being purchased. A common error is to leave the product price as a static value at point of sale, and not apply the discount to the Product Revenue.

• Follow @minethatdata (Kevin Hillstrom)

Now what? Segmentation

Now What? Segmentation

Now What? Segmentation

Now What? Segmentation

Now What? Segmentation

• Prioritize owned, earned, and paid marketing efforts by page type and product type.

• Create meaningful GDN retargeting audiences in GA based on segments (winners and losers)

• RLSAs need to be served via pixel to include the “conversion label”. Armed with a strong GTM implementation and GA data, create awesome RLSA lists with a few clicks.

Final thoughts• Being able to send a large amount of impression

data (including promotions) on a single hit is a huge improvement in GA architecture.

• Product lists, especially for cross sells and upsells, is a great feature. Internal promotion tracking also very good.

• Much of the merchandising related data model in Enhanced Ecommerce was “doable” before product launch, but is a very welcome addition to the platform now that it is built in. User adoption.

Final thoughts

• Session based data is still “meh”, same with Buy to Detail rates being based on views.

• Vocabulary of product related “verbs” (like add to cart, purchase) is solid, but a bit rigid (no native way to track ‘add to wishlist’, ‘social share’, ‘video play’.

• Product scoped custom metrics for profit metrics are relatively easy to implement and major in terms of data value.

Final thoughts

• Generally speaking, I think that there are lots of “actionable items” that come out of the new platform.– Product level advertising insights• especially good for PLAs

– Smart Remarketing– On site merchandising decisions– On site promotional decisions– Improve performance of recommendation engines

Current @AnalyticsNinja Rating