September 2017 Coffee Chat: Let's Optimize your Buyer's Journey

Keeping your site up and running is just the beginning when it comes to e-Commerce. If you want to keep a customer from first click to final checkout, you’ve got to dig deep into the customer experience and make sure that it’s optimized. Every. Step. Of. The. Way.

Fortunately, New Relic has an amazing suite of tools to help you do just that. Join us for our September Coffee Chat where New Relic’s @awetzel will share all our secrets for creating and sustaining an amazing buying journey.

When: Wednesday, September 27 at 9:00 am Pacific / Noon Eastern.

Register Today!

What’s a Coffee Chat?
We’re so glad you asked. Grab a coffee (or tea, or energy drink), pull up a chair and join us in our Slack space for an hour of conversation. You ask the questions, we supply the answers. Not in our Slack space yet? Register for the Coffee Chat and we’ll send you an invite the day before.


:coffee: :coffee: :coffee:

E-Commerce Coffee Chat with @awetzel

September 27, 2017 Conversation from #coffee-chat Slack Channel

@hross [8:52 AM]
Everyone have a cup of :coffee: ? We’ll be getting started with our Coffee Chat about your buyer’s journey in just about 10 minutes!

@linds [9:01 AM]
@awetzel Is in the house! Andy is the Director of Industry Solutions here at New Relic and he is helping us understand our Buyer’s Journey today! Welcome!

@awetzel [9:01 AM]
Morning all :slight_smile:

@linds [9:01 AM]
We are officially taking any and all e-commerce questions!

@hross [9:01 AM]

Hey @awetzel

Maybe you can start us off by sharing some of the things that you think are most important to look at in the buyer’s journey through your site.

@awetzel [9:02 AM]

It might seem obvious, but it’s important to understand the steps of the buying journey itself

where are users coming from?

do they land on your A site or come to a purchase page from a social media referral?

common steps are



Item Select

Add to Cart



after you know the “happy path” for your customers, you want to think about which data attributes to capture from the process and how visualize/measure/monitor

@ebodor [9:08 AM]
We are primarily an e-commerce development company and this is a big concern for us. We need a way to monitor that exact flow, look at performance, bottlenecks, any errors, and abandonment. We have yet to figure out how to do this in New Relic as all the data is aggregated and does not really help with visualizing the flow. We have tried with Google Analytics but thats more analytics instead of the lower level that APM provides. So my question is how do you recommend mapping this out in New Relic?

@linds [9:08 AM]
set the channel topic: Sept 27 E-Commerce Chat

@awetzel [9:09 AM]
great question @ebodor

i’d recommend using Insights to do this type of analysis

for one, the data is not aggregated, so that’s great

@ebodor have you had the pleasure of using Insights?

@ebodor [9:15 AM]
I have but have never really figured out how to embrace it into our work flow for ecommerce. I have seen some examples here What's on Your E-commerce Dashboard?, but a lot of the examples are more dashboard specific. I am trying to drill down into where pages are not performing during the process, are they running into errors, is that causing abandonment. So while conversion rate by page load time seems really useful it seems too high level.

We also have the challenge of I think we need to figure out custom instrumentation to even get that kind of data into New Relic, I assume ecommerce related data would not be very specific in New Relic.

@awetzel [9:16 AM]
here’s something I’ve seen work at other companies. lmk what you think

create a funnel in Insights for the buying process. Start with “bottom of the funnel” analysis… in other words, the “pay” step. measure your “payment success rate” then optimize it for availability, correctness, and performance

let me provide some more detail

@awetzel [9:19 AM]*
uploaded and commented on this image: Pasted image at 2017-09-27, 9:19 AM

Insights funnel for buying process ^

I recommend starting at the bottom of the funnel for a few reasons

  1. if you’re having problems there, you are leaking revenue

  2. there is no ambiguity about the user’s intentions. Higher up in the funnel you might be hard pressed to understand why a user doesn’t click on an item they searched for or why they don’t pay for items in their cart. But if they click on “Complete Purchase” you know their intention. So there is a certainty to it

  3. it connects concretely to the backend fulfillment… in other words all the infra, apps, and services you are monitoring with new relic

  4. it provides a derived KPI that might cause some discomfort, which I think is good. What do I mean by that? If I ask a customer “What is your Payment Success Rate?” they usually have no idea. So first you need to know what that number is … and then make it better.

on that last point, i’ve worked with customers who had an 89% success rate. that’s an optimistic way of saying 11% error rate on purchase

@awetzel [9:27 AM]
uploaded this image: in our demo data, it’s even worse…


@ebodor [9:27 AM]
That is very helpful thank you and a great way to look at it. So if we discover for example payment success rate not being good, how do you then create insights that help identify root cause?

This has been the biggest challenge, is finding actionable data out of the reporting.

@awetzel [9:28 AM]
@ebodor you’d use Insights plus APM, Browser etc.

@ebodor that is the art of the science (is that a thing?) of creating dashboards…

you have to ask… Who is the audience of the dashboard? How is the data actionable? What value will be derived from the actions taken"

let me give you a specific example of bridging from Insights to APM

@ebodor [9:30 AM]
So if we have people abandoning during the payment step, is there a way to filter down APM/Browser to just those specific transactions? Like looking at slow transactions thats a great list but with a large app its hard to find the specific users that are getting those.

Awesome thank you

@awetzel [9:32 AM]
one thing I skipped over is that to do this type of granular analysis, you’ll want to capture specific attributes from the buying process

like this ^

@awetzel [9:33 AM]
once you have them you can do a much more specific analysis

let’s take the example of a user getting an error in the process

you can create a list of users with errors. and then see the transactions, apps, error messages, etc.

@awetzel [9:35 AM]

@awetzel [9:35 AM]
here’s an example

it’s demo data, so we have customer names, but probably you want to make that something that’s not PII

in this example, Kristen had errors

we can see the transactions she ran, the results, which backend apps they hit, etc.

great news, we’re getting errors from Storefront, Ordering AND Inventory services :slightly_smiling_face:

@ebodor [9:38 AM]*
Thats awesome! Yes we were thinking of potentially using Insights as well to do admin ecommerce dashboards for our customers. Running dashboard and reports on our system is rather taxing on the system so we were hoping that Insights would be a good fit to get that stress off our system. But we have to figure out exactly the PII side for sure.

@linds [9:38 AM]
:tada: Insights FTW!

@ebodor [9:39 AM]
Is there any demo dashboards we can use like the one you sent where we just change the data to our attributes?

@awetzel [9:41 AM]
not exactly, but we could have someone on our side work with you to get these dashboard examples into your account

i think that’s the best way

@ebodor [9:41 AM]
Awesome thank you

@awetzel [9:41 AM]

@awetzel [9:42 AM]
uploaded this image: Errror1.png

@awetzel [9:43 AM]
just to finish how you might some actual, fixing of stuff… the visibility from Insights would let you know which apps/services to look at, which error messages to look for etc.

@awetzel [9:44 AM]
uploaded this image: all the way down to the line of code

@awetzel [9:45 AM]
people think about needle in the haystack troubleshooting, but I think that skips the biggest challenge… you have a field full of 10,000 haystacks. which one or ones do you look in?

Identify, Isolate, Fix

Insights and Alerting can help with identification and isolation (as can other NR products)… but to really fix stuff you’ll need to use APM, Browser, etc.

@awetzel [9:49 AM]
uploaded and commented on this image: Pasted image at 2017-09-27, 9:49 AM

oh, and i’d be remiss not to mention that you could do a quick ROI on fixing errors :slight_smile:

@ebodor [9:49 AM]
Thats very helpful thank you. You are right about the needle in a haystack. Its so much that dev’s don’t even want to look at the interface. But yes if we can drive it to this is where its broken and a bug ticket can be created out of that data it would be incredibly useful!

@awetzel [9:50 AM]
@ebodor great!

the visuals might not be for the devs, but the prod mgrs and app owners should eat this up

I’ve found it really helps drive collaboration

you take a concept like conversion rate and apply it to perf/avail of the code

@ebodor [9:52 AM]
Building those insight metrics would be fantastic for proving the value of spending time on this. The business always want new features and its hard to get them to stop and say we need to spend time on improving things in general. New features just add complexity and potential performance impact. But if you can prove it with numbers, that its time well spent, thats the real winner. The beauty of ecommerce is everything you do can impact the sales performance.

@awetzel [9:52 AM]
@ebodor so true.

you hit on what I think is biggest unsung value of Insights and NR in general… giving you the evidence to identify and justify optimization activities

one way you might be able to get marketing interested into a payment step optimization is this…

say you have a 1% conversion rate

for every 10 users that you lose at the payment step due to errors, you’d need to replace them with 990 users at the top of the funnel to make up for it

might even be the full 1000 users (need to think it through more)

but regardless, what does it take to drive that many more users into the top of the funnel?

I’d sum up this strategy with 3 rules of my own creation

  1. The Anna Karinina Rule

  2. The GI Joe Rule

  3. The Voltron Rule

Not sure we have time to go through them, but maybe that’s a discussion for next time

@linds [9:59 AM]
Ha so intriguing! @awetzel

but I think my favorite quote of the day is still: “that is the art of the science of creating dashboards”


@awetzel [9:59 AM]
my pleasure

@linds [9:59 AM]
and thank you @ebodor for sharing your use cases and questions with us!

@ebodor [9:59 AM]
Yes thank you so much, this has been very helpful. Its always hard to figure out where to start and this is a very good roadmap.

@awetzel [9:59 AM]
i’m going to get a hand massage now

@hross [10:00 AM]
you deserve it Andy!

@linds [10:00 AM]
Haha excellent work!

@awetzel [10:00 AM]
thansks @ebodor! glad this was helpful for you

@linds [10:00 AM]
and @ebodor we will be looping around with you about those dashboards! :thumbsup:

@ebodor [10:00 AM]
Sounds great thank you

@linds [10:00 AM]
That is a wrap! Thanks again everyone!

See you next time! :wave: