Help with understanding how to use EVAR properly on a conversion report

[VirginieRe2]

I am struggling with building a report based on an evar as I dont fully comprehend the nunances of evar.

On my website, we have product pages. On those product pages, we use and evar to capture the product brand (evar10).

The configurations of that evar are:

merchandising : conversion variable

binding events: purchase event (order, revenue, units) checkout, cart additions, remove from cart, add to wishlist, remove from wishlist, product view, cart view, and cart open.

allocation most recent : last

reset: do not reset

expire after: purchase

enable merchandising: enabled.

Now, i want to know conversion rate for evar10 = brand_A
Specifically, I want to know 2 things :

QUESTION 1. of all visits to products page where evar10='brand_A', what percentage of those visits purchased 'brand_A' products?

QUESTION 2. of all visits to products page where evar10='brand_A', what percentage of those visits purchased a product (of any brand, does not have to be a brand_A product).

I created a segment based on HITS where evar10='brand_A' (attribution = default) and i applied that segment to a freeform table where I put the name of the region as rows, visits and conversion rate (out of the box Adobe CVR) as rows. I believe this report will answer report 2 above, but not report 1. Is that correct?
How do i set up my report to answer question 1 or 2 above?

[Vinay_Chauhan]

Hi @VirginieRe2 

You are thinking about this the right way and your current setup for eVar10 looks solid. Since it’s a merchandising eVar (last-touch allocation, no reset, expires after purchase), it binds to individual products and attributes success events like orders, revenue, or units based on that binding.

and here are my comments to your two questions -

Firstly For Your Question 2 : “Of all visits to product pages where eVar10 = brand_A, what percentage of those visits resulted in any purchase (any brand)?”
Yes, your current approach using a hit-level segment (eVar10 = brand_A) and applying it to a Freeform table with Visits and Conversion Rate is correct. The standard CVR metric will calculate how many of those visits included any purchase event, regardless of brand.

and, For Your For Question 1 : “Of all visits to product pages where eVar10 = brand_A, what percentage purchased brand_A products?”
For this, you’ll need to narrow the numerator to purchases that were actually tied to brand_A. Because your eVar10 is merchandising-enabled, success metrics like Orders or Revenue will automatically attribute to the correct brand binding.
You can get this view by:

• Using eVar10 = brand_A as a row in your Freeform table and applying Orders or Revenue as the metric — that will show only purchases attributed to brand_A.

• Then, divide by total Visits where eVar10 = brand_A to calculate your brand-specific conversion rate.

So according to me your current setup correctly answers Q2 (any purchase), while a merchandising-based attribution on Orders or Revenue gives you Q1 (brand-specific purchase rate).

Let me know if that helps.

[VirginieRe2]

Very helpful! Thank you so much.

[Jennifer_Dungan]

Hi @VirginieRe2,

Merchandising eVars can be tricky to wrap one's head around, and "conversion syntax" even more so.

The thing about conversion syntax is that it won't appear in your data except for on the binding events. This is a way to set a value earlier in the visit (such as a search term captured during a search, where you don't yet have product information), and stitching it to the product later on based on events such as prodView, scAdd, purchase, etc.

On those binding events, the value will be available to you like any other merchandising eVars.

Note that with Products, segments currently don't distinguish between just the specified product, and all products on the same hit..

So if you have a purchase of three items (Item 1, Item 2 and Item 3) and you create a segment for "Item 2", you will still get the other items returned, even if you are using HIT (the segment logic returns the entire hit and all the products from that Hit.

---

QUESTION 1. of all visits to products page where evar10='brand_A', what percentage of those visits purchased 'brand_A' products?

---

Assuming you are using prodView on your products page, and prodView is a binding event, then you should be able to look at a segment like:

HIT [

    eVar10 equals brand_A

    AND

    prodView exists

]

and another segment like:

HIT [

    eVar10 equals brand_A

    AND

    orders exists

]

NOTE: Yes, I am specifically looking for Hits here... you want to be sure that the HIT had both the specified product and the event. You don't want Visits that had brand_A in one hit and prodView in another... they must be on the same hit.

Pair each of these in columns with the Visit metric (this will get you the number of visits that looked at the brand_A product page, and the number of visits that purchased brand_A... then you can create a calculated metric using these segments with Visit, to create a percentage calculation

---

QUESTION 2. of all visits to products page where evar10='brand_A', what percentage of those visits purchased a product (of any brand, does not have to be a brand_A product).

---

For this, you would use the Product View segment from above, and create new segment for any product purchased:

HIT [

    Order exists

]

(Assuming all Orders are products, which in this context I feel is likely the case)

And like before, pair these with the Visits metric and create a calculated metric to get the percentage.

@Vinay_Chauhan is right, you are definitely on the correct path.

[VirginieRe2]

Thank you very much!
What is the effect of having the evar configured to "allocation most recent (last)?"

For example, the user sees product A with evar10="brand A" ,then product B with evar10="brand B". User purchases product B.

Isn't it so that if I create a report with visits and orders for product A ="brand A", it will show 'zero' visits for evar10="brand A" because the evar configuration is set to "last", so evar10 = brand A is overridden with the most recent value of evar10, which is "brand B"?

[Jennifer_Dungan]

The other interesting thing about Merchandising eVars is that the allocation is per product.

But even still, "allocation most recent (last)" does not re-write the history of hits... the hits that were made with a value are still available in the data... "Last" just means that the last value will persist on hits until it's overwritten... and again, products are slightly different where each product maintains it's own "last touch" value. So even when using a normal eVar, the value will still be in your data:

• Page 1
  • eVar set to "value 1"
• Page 2
  • eVar not set, but "value 1" persists
• Page 3
  • eVar set to "value 2"
• Page 4
  • ...

Pages 1 and 2 will always have "value 1" in your reporting, and Page 3 and beyond will have "value 2" until it's overwritten, or it expires.

I am also trying to figure out if you really need conversion syntax... if you know the brand on the product page, a normal product syntax merchandising eVar would work just as well... you don't need to bind it, but you also don't have to set it on each use, the attribution will keep the value.

Either way, let's look at an example, I will use eVar1 (conversion bound on prodView) for "search query", and eVar2 (product syntax) for "brand":

• Home Page
• Search for "shoes"
• Search Result Page
  • Set eVar1 to "shoes"
• Product Page Product A
  • Set Products to Product A and eVar2 to "brand A"
• Search for "t-shirts"
• Search Result Page
  • Set eVar1 to "t-shirt"
• Product Page Product B
  • Set Products to Product B and eVar2 to "brand B" and prodView event
• Add Product B to Cart
  • Set Product B (you don't need to set eVar2, this will carry forward with attribution, I believe that eVar2 even though not explicitly bound will also persist, double check this) and scCart event
• Go to Cart
  • Set Product B (again without eVar1 or 2 which should still persist) and scView event
• Start Checkout
  • Set Product B (again without eVar1 or 2 which should still persist) and scCheckout event
• Purchase
  • Set Product B (again without eVar1 or 2 which should still persist) and purchase event

When you look at your data, the Product View for Product A will be set to have "shoes" and "brand a"

The Product View, Add to Cart, Cart View, Checkout, and Order for Product B will have "t-shirt" and "brand B"

I suggest using a QA or Dev suite to test these scenarios yourself... checking an isolate flow and looking at each step is the best way to visualize what is happening.