April 7, 2020
Marketing Channels in Adobe Analytics

Marketing Channels in Adobe Analytics

– [Eric] All right,
thank you for joining us for this Analytics Insider webinar. We’re really excited to
be walking you through some of the basics, some
advanced techniques, and some of the configuration
options that you have for managing an ever-changing
number of marketing channels within the Adobe Analytics Cloud. My name is Eric Matisoff, and at Adobe I’m the analytics
and data science evangelist. And joining me today I have Ben Thompson, who is the director of ecommerce analytics over at Bluestem Brands, who’s gonna be walking you
through some real-world examples of taking the best advantage
of marketing channels within his company. So without further ado
let’s actually jump in. To kick things off, a quick reminder. If you have questions on
the latest and greatest of Adobe Analytics
features and capabilities and tips and tricks, head on over to the Adobe
Analytics YouTube channel, which you can access easily by going to adobe.ly/adobeanalyticsyoutube. Keep in mind that’s all lowercase. In addition, we also have
the Adobe Experience League, which you can access through
adobe.ly/ExperienceLeague, and a capital E, capital L there. It’s a great way to stay in touch with the community at large when it comes to the
Adobe Experience Cloud. And go ahead and click on
that Intelligent Guidance link in order to receive tailored
tutorials and feedback and recommendations associated with your needs within Adobe
experience cloud products. You can also head on over into the analytics forum by
going to Adobe League slash analytics forum all lowercase or just clicking the question mark and clicking into the community. It’s a great way to talk
with all of your peers within the analytics community, as well as interact with the
folks from the product teams here at Adobe. And finally, we have
just a couple more stops left on our Adobe Insider Tour as well. Up coming soon are Denver and Toronto. So be sure to sign up
and register for those. They’re just a couple of weeks away. So you can do that by going to Adobe.ly/InsiderTour, capital I capital T. And while you’re there
you’ll get the opportunity to learn some tips and tricks, get a feel for the future
of Adobe Analytics in our annual sneaks
competition,or sneaks session, as well as be a part of our
Analytics Rockstar Event. Then finally, if you’re
looking to stay on top of the latest and greatest when
it comes to Adobe Analytics, there’s no better way
than leveraging the Adobe Analytics Insider newsletter. You can access it by receiving monthly information from Adobe by going to Adobe.ly/AnalyticsInsider,
capital A, capital I. All right, so from there
let’s actually talk about what we’re all here for, which is to talk about marketing channels. And so throughout the day, or over the next hour
we’re gonna be walking through each of these different topics. Preparing for marketing channels, an overview of course of
those marketing channels, then preparing for them, configuring them, and
some tips and tricks, and then I’ll be handing
it over to Ben Thompson from Bluestem Brants. He’ll be walking you
through a bunch of different tips from the real world, showing you exactly
what he’s been doing at Bluestem to make the most of the marketing channel’s capabilities, and even provide you
with some analyses of how he’s been able to leverage
them within his organization. So let’s talk through an
overview real quickly. First of all, you have a lot of options when
it comes to Adobe Analytics, and that will not be a surprise to anyone, and when it comes to
external traffic dimensions you’ve actually got
three different types of external traffic sources. You’ve got three different
columns across the top. You have traffic sources, which populates via the
paid search detection rule. Then the next column you
have campaign reporting, coming in through s.campaign. And then finally you
have marketing channels, which is the topic of today’s session. There’s a lot of content on this table, so this may actually be
one of those times that you take a screenshot of our webinar, or even hit pause on the YouTube channel, or something like that, in order to really dive
deep into the difference between them, but there’s a couple of
things to point out here. Basically the recommendation
is to take advantage of all three of these, but in terms of setup, but leverage marketing
channels specifically for your analyses. And we’ll get into the reasons why, but the short answer is
that traffic sources is useful for setting up
paid search detection. Campaign reporting is
useful for setting up other paid campaigns, and marketing channels
combines all of that along with your organic and
natural channels as well, into one uber powerful marketing
channel set of dimensions. So what are these marketing
channel dimensions? Marketing channel
dimensions as I mentioned, combine both paid and
organic traffic for all of your incoming traffic to the site. Your channel rule criteria
is customizable and it can take into consideration
any of those items we were just discussing
around tracking codes and search engine detection, as well as referring
domains, entry pages, and query parameters. Then processing rules are
applied to all incoming traffic in sort of a waterfall order, and so you can see that
in that screenshot there on the left of the order
for how natural search is defined. Then if it doesn’t meet
those rules it follows through to email, and finally affiliate, and all of that then
turns into your beautiful marketing channel workspace dimension. A few of those reporting
features that are built into marketing channels
include dimensions that default to last touch or closers when it comes to attribution, but can be expanded using
attribution IQ to help adjust attribution
on-the-fly for your marketing channel dimensions as well. Each channel can be broken
down by details that are specific and unique to that channel, or even bubbled up a level
using classifications in order to apply additional metadata, like you see in the
screenshot on the right, where we’ve bubbled the
two different types of marketing channels between
paid, earned, and then if we had a longer screen shot, you’d also see owned as well. There’s two very simple
steps to setting up marketing channels. First you create your channels, then you define your rules, and don’t worry. A little later we’ll be
walking you through Ben’s exact examples for how
they set them up over at Bluestem Brands. So once you create and
name your channels you’ll go through a wizard, and that wizard starts with
what we call an auto setup. If you’ve never created
or set up marketing channels for your report suite before, that is the first screen
you’ll see within the admin console when you
head into the marketing channel manager. Beware that once you’ve
then customized the names of your channels, you actually cannot delete
them once they’ve been created. So only create names that
will be used going forward. However, they can be renamed. Finally, you’ll want to
make sure that all the rules that apply do have
that checkbox on the left under enabled in
order to make sure that they appear within your
reporting dimensions. There’s one other item to call-out here, in that there are some override options. So each of these channels
you can manually select whether or not it should
override the previous channel. By default you’re going
to see channels like direct and session,
refresh, or internal, that most likely don’t have
the override checked, and we’ll actually give
you an example of that right now. So here’s an example of
four different visitors that have three different visits each. So that in this first
world where direct override is unchecked, which is recommended, that first visitor that
you see there with the red hair, when she visits on the
third visit and perhaps creates some revenue there, that revenue, instead of
being attributed back to direct or that click-through, would be attributed back
to email since email is set to override direct, because direct is the
override is not checked. Similarly in the second row, you can see that from
paid search on the first visit then direct and
direct on the next two, again that paid search
value is going to override the values collected within direct, because that’s the most
recently used marketing channel that does have the
override checkbox checked. And et cetera, with the third one, if all three are direct
then direct will get the attribution there, and then at the bottom one
if none of them are direct, then it will simply be
the most recently set value will be used within
last touch channel, or can be adjusted of
course using attribution IQ. Now take a look at how
those three items in purple actually change
when I switch from direct override unchecked to
direct override checked. See how all three of
them now are have a click through up associated with each of those different direct visits, which means that that has overridden the previously driven marketing channel. It’s a really important
concept and we want to give you the flexibility
to make the decision on your own who should be
getting that attribution when it comes to setting the
values of marketing channels. The next step after you’ve
defined your channel your channels and naming them is actually defining the processing rules. So the processing rules
that you can see listed out here are pretty standard items, maybe we added a custom one
or two of them in there, but Ben is actually gonna
walk you through their exact detail for how
they’ve defined processing rules within Bluestem. But the idea is that if a
visit comes through on search, first, a user’s referring
domain and paid search detection will try to
identify whether or not that visit should be
attributed to paid search. If it doesn’t meet that
criteria then it’ll go to the next processing rule. However, if it does meet that criteria, it will stay with paid search. That’s terribly important
to understand is how this ordering of processing rules works, because if those were flipped, natural search then paid search, you may actually accidentally receive or attribute traffic to
natural search by accident. The last piece around
marketing channels to go through are around marketing
channel classifications. So this is possibly one
of those items that you may not be aware of, is you have the ability through your admin console to actually apply
classifications to all three of the different types of marketing channel dimensions. That’s marketing channel, first Marketing channel, last touch marketing channel, as well as all three of
those four channel details. So that means marketing
channel, detail, first touch, marketing channel
detail, and last touch marketing channel detail. Now any of those can be used, and when you apply a single
classification to one, then that is actually
flows to all three of them. So you only have to perform that once. Keep in mind, you can
upload classifications through the standard
classification importer, or through using the
classification rule builder as well. Both of those may be useful for you. So wait, wait, wait, wait, wait. This is a animated gif
from one of my favorite movies of all time, The Princess Bride, and a question comes in
wait, just wait, just wait. Why, and the question that we’ll get is, what happens to first touch channel and last touch channel dimensions
in workspace in this new world where attribution
IQ can effectively change the attribution of any
dimension on-the-fly? Well, let me explain. The answer is you actually
have a lot of flexibility. We’ve created a brand new
dimension that you may or may not be aware of, simply called marketing channel. The goal of that is simply
to help avoid confusion. Any attribution model
that is applied to the previous dimensions ignores
the settings that are applied to those dimensions. So if I were to effectively apply a linear attribution to a metric, even though I’m looking at the first touch channel dimension, then linear overrides the
first touch channel concept. Similarly I could even apply a last touch attribution model to
the first touch channel. My head will actually not explode. Instead we’ll see the
last touch attribution model in that instance. So that confusion is why
we’ve created this new concept of a marketing channel dimension. So that it’s just simply
no attribution is applied. If you’re curious how it works, that second paragraph, or second set of sentences
will be useful for you. Marketing channel actually
duplicates the values in last touch channel, and classifications are
applied to all three. All right, so what do
you need to think about when it comes to preparing
for marketing channels? First of all, you need to think there’s some
tracking code best practices, and there is a again, a lot of useful information on this slide, but it’s not worth going
through every single item. This may be a useful one again, to take a screenshot of, or leverage the recording to hit pause. Now we have some
recommendations for when it comes to assigning prefixes
to your tracking code campaign ID. So that may be useful there on the right, but a good tracking code, generally they’re going to be shorter, they’re going to be granular, and they’re going to be
potentially cryptic in order to keep your customers
comfortable and your competitors at bay. They should also be unique,
lowercase, and consistent. I can’t stress consistency enough. So be sure to work with your agencies, work with your advertising teams, to ensure that your best
practices are being followed, and Ben even has some
good tips later that he’ll walk you through associated
with these as well. Also, we highly recommend
leveraging ad Analytics for paid search to expand
the high quality data that you’re able to capture
within Adobe Analytics. This is where data from
each of the paid search engines Google, Bing, and
Yahoo is automatically importing data when it
comes to the dimensions as well as the metrics
associated with your paid search data. Plus we’ve even built
a search engine focused template to make it easy to analyze. That additional integration
comes to you at no additional cost for any of
the current Analytics queues. Select prime or ultimate. So how about the importance
of tracking codes and marketing channels? Keep in mind that we’re
talking earlier about that really consistent
set of tracking codes, and this shows exactly why it’s valuable. If those tracking codes are not useful, then that could potentially
throw off all of the data within the marketing
channel dimensions. Whereas in this example, you can see that that
prefix of DA means it’s a display ad with a unique
identifier after it. That helps me know two things. First of all, how to properly define
the marketing channel, and then second of all
that I can include that and define that within
my marketing channel detail within my marketing
channel processing rules. Next up, be sure to
still set up paid search detection within your report suite. Here in bold we’ve identified
where this actually exists, and it works very simply. The first thing that paid
search detection does, is it looks to see is
this visit coming from a recognized search engine. And that list of search engines, we keep updated all the time. Then when traffic comes to the site, paid search detection
looks for a query string parameter that you’ve
customized here within the paid search detection settings
within the admin console. If it exists, then the visit will be
bucketed as paid search. If that query string
parameter QSP does not exist, instead we’re labeling that
traffic as natural search. Beware that almost
definitely the keywords will not be collected automatically
by paid search detection. As you may be aware, natural search keywords
were removed from refers for all Analytics platforms
and almost all search engines. So rarely do we ever see
a natural search keyword come in. And then most likely that same keyword was removed from the refer
for paid search now even as well. So again, you’ll have to
think about how in the landing page URL, or as a query parameter, or as part of ad
Analytics for paid search. You can actually automatically
capture that data within a dimension, or variable within Adobe Analytics. The next step you want
to do is configure your internal URL filters. Beware that the goal of
doing this is to ensure that your own domains and
subdomains are filtered out when Adobe Analytics
is understanding what a refer is. They are different from the
URL filters in the s_code or at measurement library, because that simply
says that differentiates between an exit link
versus a non-exit link. So beware of using these properly, and you can check them
to confirm that they’re set up correctly by
heading into your admin console for your report suite. One of the last pieces to define is your engagement window. Your engagement window is
by default set to 30 days, but you can reset that
to any other look-back window or engagement window that you like. And so keep in mind that
you’ll want to do that, and when you do update
that you’ll most likely want to hit an expire all so that you can completely reset the
data associated with it. Again, think about your business. If you’re selling socks or t-shirts, then maybe a smaller
engagement period makes sense. If you’re selling cars
or big-ticket items, then perhaps a longer
engagement period makes more sense. In fact, here’s even some
details around how to determine the right engagement window. You can consider using
the visit level segment focused on return frequency
to get a better feel for how that expiration could be set. Alright, so coming up
next we have some details on configuring your marketing channels. So as I mentioned earlier, by default the first
time you click to enable marketing channels in your report suite, you’ll have this auto setup wizard. This can be a useful way
to pre-populate a bunch of individual marketing
channel processing rules that are automatically ordered for you, just based on the email
identifier at the display query string parameter, or any affiliate domains as well. This is a really useful
way to kind of kickstart your marketing channel setup. Then you’ll want to
consider creating a custom channel as well, and Ben has some great
ideas for some other custom channels you could create. For example, a couple of
them that we’ve seen are associated with internal, and direct, and referring domains. So this concept of internal
or session refresh is perhaps a little bit of a confusing one. And by default that’s going
to be listed as internal, but we actually think
it makes more sense to rename it session refresh in
order to prevent any confusion. So generally, in terms of order, we like to keep these
almost all the way at the bottom in terms of your
priority within your marketing channel processing rules. So that concept of a session refresh, what actually is it? It consists of visits
to the site where the referring URL matches
the internal URL filters that we were talking
about just a minute ago. And what that means is
the visitor came from somewhere within the
site in order to start a brand new visit. Both direct and session
refresh only get last touch credit if they were
also the first touch channel. So the thinking there
is that you should not choose an override associated with session refresh direct or internal, however you have them defined. So here’s some scenarios
where session refresh may occur. I’m not going to go through
all six of them in detail, but probably the most
common one is associated with session timeout. When the engagement
period expires they open a tab to visit the website
again the referring URL is going to show as an internal domain, and therefore should be
classified as session refresh. Another scenario is if
not all site pages are properly tagged, or if there is a redirect
where the referer is getting lost, perhaps some cross domain
traffic where a user is heading from one report suite to another, or one domain to another. Finally, some long entry page load times, or if a user for some
reason decides to clear their cookies mid site, the referer remains, but the cookie does not, and therefore the visit will look new. So that final channel order. Here’s again that list
of processing orders that we saw earlier, and Ben will walk us
through exactly how he has set his up so that you
have a good feel for each of those as well. So let’s talk through
a few different tips, and this is where we’re
gonna get a little bit more advanced, and how you can take
advantage of marketing channels within your organization. First of all, let the attribution panel
do the heavy lifting. When you’re comparing and
contrasting models and marketing channels, it can be really useful to have these pre-built visualizations. And Ben will walk us
through exactly what the results of one of his
recent analyses using this panel actually looks like. So what you can do is actually compare attribution models on
the fly by right-clicking on almost any metric
within analysis workspace, and applying a different
attribution model. Then automatically we’re
going to create a third column that shows exactly
the difference between the attribution model that is the default, and the new one that you’ve
set as your comparison. Another useful tip is
to consider just using a custom model in order to
show assist focused orders. So this is where you can
apply 0% to the starter, 100% to the players, and 0% to the closer. And this custom model shows you all of the marketing channels that
didn’t create awareness or actually close the the conversion, but instead we’re key
to helping the purchase along the way. Next up you can consider
still leveraging those last touch channel and first touch channel dimensions in order to
create this cross tab view of your metrics, and so here across the
Left we have last touch, and across the top we have
first touch dragged in as a dimension, and then we’re quickly
able to see how first touch and last touch are
working together where they are exactly the same. Generally, this means
that they were either the only visit that drove that conversion, or it was both the
originating and the closing attribution marketing channel as well. Then you have the opportunity to analyze, for example, all of the
different last touch channels that were supported based on any specific first touch channel, as you can see here. So I can see that the top
different channels that were helping to drive
email driven first touches, or actually aligned
with referring domains, display, and social networks. So that can be really
valuable in seeing the difference between how those two different attribution models work together. Now this is one location
where the new marketing channel dimension actually won’t help you. So consider leveraging
last touch across the left and first touch across the top. So with that I’d like to
hand things out over to Ben, who’s going to talk us
through some marketing channels in the real world, and how Ben is using them
over at Bluestem Brands. So Ben, why don’t you take it away. – [Ben] Thanks Eric. Hi everybody, this is Ben Thompson. I’m the director of
Ecommerce Analytics over at Bluestem Brands. With our agenda today
we’re going to talk a little bit about who I am and Bluestem is. We’re going to walk through
defining and setting up some channels, and the processing rules
that go into those. Then we’ll take a look
at some reporting on some channels and how channels
interact with tracking codes, and that can help you
QA your channel setups, and finally we’ll play a little bit in the workspace in the new
attribution panel that’ll help us see some of those
models that Eric was talking about. So who am I? I’m Ben Thompson. As I mentioned previously, I work for Bluestem Brands. We’re a multi-channel
retailer based on Eden Prairie, Minnesota. It’s about 20 minutes
outside of Minneapolis. We’ve got 13 different retail brands. You can see the logos of
some of them off to the right there. We do about two billion
dollars in revenue. The vast majority of that online. Historically we’re a kind
of a catalog and direct marketing company, but now with the move to
web by basically everybody, we’re very web-centric
and actually do quite a bit of mobile sales, which is interesting. We play in a unique spot
with a credit and retail mix, where a lot of our sales are done on our proprietary credit products. So let’s talk a little
bit more about channels. So first off, you’ll want to prep a
little bit internally to define your channels. There’s some key decisions
that are going to go into that. One thing I found here
at Bluestem is we really need to define what is our channel, and what’s some detail
or a sub channel about that channel? For example we’ve had
some internal discussions around our branded and
non-branded search channels, or are they just a detail
or a sub channel of paid search? It’s really important to
discuss these with your marketing team on how
they’re pulling levers and aligning their spend for these channels, and if your reporting
is going to match up for their needs. Next do you have the
correct identifiers in place for your channels and those details? Usually these are URL parameter. We use one called CTIV internally. Often it’s CID, or some other companies
have kind of their own code that goes there, but this is how Adobe’s
gonna grab that detail for you, and it’s really important that your URL parameters are A, present, but B, consistent. Third, we really want to
check for some gotchas. I’ll walk you through a
little bit more detail on some of these, but some that can come
up are if your site uses redirects you may lose some tracking. Google Display Network, because it runs through
Google can be detected as a paid search referral. Your paid versus organic social. You may have tagged some
of your paid social, but how are we doing that organic? And finally, there’s
some tricky things with email as well. Finally on this slide, we really want to
document what the channel, the details, and the parameters are. This isn’t the most
exciting part of our jobs, or of Adobe Analytics, but it’s really, really important. So on the next slide you’ll
see if a sample of my SBR. This can really, really handy, especially for users who aren’t admins, and they can’t go in and
check how you set things up as Eric showed. They need a document
they can go reference. So this is ours. We show what the source is, what our CCID is, and if there’s a sub
campaign we put that into a CC campaign parameter, and then we check off
whether or not we’ve got an override on those channels. Finally, as we look at
collecting those marketing channel detail or values, we mark that down on
here so people know which piece of information from
our URL we’re collecting. So let’s dive a little bit into my setup. So here’s a screenshot of our
marketing channel manager. This is where we set up
our channels initially, and you can enable and
disable channels here, as well as choose the
overrides, as Eric showed. You touched on using the wizard, but what if you already
have a setup that you want to make some changes to? Here’s the same place we can do that. So I’ll touch first a
little bit just on override as Eric did. An override set to true
is going to overwrite that previous Channel as last touch. That’s a noble nod. So for instance, if I have
a direct set to override, and I visit by a paid
search then direct and convert direct is gonna get credit. If that override is set
to no then paid search is gonna get credit. Most often direct, internal,
and other refers are set to not override. Some companies choose to
override using any paid media, and not paid media to not. This one’s really up to you. As a best practice, direct, internal, and
other should be set to not, but where you go from there
is sort of up to your company. Next let’s take a look at the fun stuff, the processing rules. These help Adobe identify
which channel a visit came from. These rules will happen
in a waterfall style, and you set the processing
order however you like. One of the things to
remember here is that these are not a priority, but these are how you bucket
your marketing channels. It can be a little tricky
because when most of the time, a channel is clearly
identified via your URL parameter or our CCID. Sometimes a visit could
look like it came from more than one. So the order becomes
really important here. For example, if you have
a redirect that carries through URL parameters, it could look like an
internal referral or some other paid channel. So we put internal really
low on our order in the waterfall here. Also as mentioned before, in paid search, there’s
an option for Adobe to detect using their rules of the visit was problems say Google. It’s great but Google
Display Network also comes from Google, and might look like it was a paid search. So what we do is we put
display at the very top because we’re heavy users of GDM. It’s also really good to create that other campaigns bucket to
catch any missed channels toward the bottom. We’ll touch more on that in a minute. Let’s look at some of
the detail behind some of our processing rules. So for this one we’ve set up for display. On our SDR we said display
to be from a couple of CCIDs for 471, or 1364. So we have set the rules
to look at any of the following criteria to be true. You also have the option
of all if you need them all to match. So if our query string
parameter is 471 or 1364, we’re going to set that
channel for display. Then we get to choose our channel detail. This is a secondary piece
of data we can store about the visit, which will still follow
the attribution rules and overrides that we set
on the previous screen. In this case because display is two CIDs, we actually are storing
the CTID value in our channel value or detail. As a side note we could
have just called these two different channels, but for internal reporting
purposes we wanted them lump together. Setting that value or
detail lets us break them back apart for reporting purposes. If you’re not sure what to
put as your channel’s value, a good rule of thumb is just to use its identifier or URL parameter. That can help you troubleshoot and do data quality checks on what is
ending up in your channel, and you can always change it later. Our next rule is paid search. We mentioned earlier
about the paid and natural search detection rules, and this is where they come into play. They can be really handy
to help catch those paid search visits, even when you may have or
altered an old parameter. If they’re still in your
paid search detection rules you can still catch those. When the URL has one of our tracking IDs, then we’re going to set
that channel to paid search. Also if it has those protection rules met, and we’re going to set
the channel value here, to our CT campaign, which helps us know which
paid search campaign it is, such as branded, non-branded, or PLAs. This details becomes
really useful later as we breakout reporting, and I’ll show you that. Next we’re going to look at
some of our other refers. So we’ll check out social
network and kind of our other bucket of referring domains. So far we’ve looked mainly
at the URL parameters for our rules, but there’s a lot of other options. So for instance, on social we use the
referring domain to check social networks, as they might come from Facebook, Twitter, Instagram, et cetera, but might not have our
URL parameters if somebody posted a link to our site via organic. We also have our URL
parameters identified in this rule as well. Finally we have a rule
looking just for all other referring domains, lower in the waterfall, just to check if some
other website has linked to ours without our tracking IDs on them, we like to capture that and
see where they came from. Next, I highly recommend
that you create an other campaigns bucket. This one should just
have a rule that if your query parameter is present, then we’re going to grab it
and put it in other campaigns, and I definitely recommend
setting the detail value to that URL parameter. This helps a ton in troubleshooting. If you see that this bucket is growing, or you think there are
visits and orders missing from another bucket, digging into the detail through your other campaigns bucket can help you a ton. I would also definitely
recommend setting alerts around here to notify
you if that bucket is growing unexpectedly. Finally a little lower
in the waterfall we’re going to want rules that
capture direct traffic and internal traffic. So for direct it’s simply, if it refers it does not exist, and it’s the first page of the visit, and in our case we set
the value or detail for this to the page. Next is internal. You might have other
subdomains that you track, or have some redirects
that move traffic around from their original landing page, or have some of those cases
that Eric laid out before, where an internal or a refresh
rule become really handy. So in our case we’ve
got a lot of redirects, and we want to make sure
that we’re capturing those correctly as the
Adobe Analytics tag might not have time to fire
on the very first page. One word of caution on these, if you do have redirects, sometimes they drop URL parameters. You’ll want to work with
your engineering team to make sure that those
keep the URL parameters, because a redirect from say, a paid search might look like
it’s an internal referral, just because it drops the
paid search URL parameter. They often happen so
fast that the Adobe tags can’t capture them. So now just a few other notes on set up. First of all, it’s really important to
remember that changing these rules doesn’t reset history. The rules you put in
place will only start to process new data from the
point at which you hit save. If you do make some changes
make note of what and when, and put that in your SDR so
people know why data change. We had an instance a few
months ago where we saw one of our channels leap
in the year-over-year view, and that was because a
year ago we hadn’t made some changes that nobody documented, and we didn’t know why that
all of a sudden changed. You can also add classification
hierarchies to channels, just like you can to llevarse. You can also change
the expiration setting, so if you’d like to end
the engagement period at 30 days of inactivity or maybe seven, you can do that, and your own business model
might kind of fit in here. Finally, if you want to
bundle up some channels together for certain types of reporting, you can quickly create some rule builder reporting roll-ups. For example, we often
view our internal and direct as part of one bucket, because we know he has some
redirects on that homepage. So now that you’ve
created your channels and collected the details about them, you can start using the data. A really great place to
start is in workspace, and just searching for channel. In my workspace I’ve got a lot of segments created off of these channel, and classifications, and
some roll-ups as I mentioned. Your data now really
becomes flexible to use however you like. Next let’s bring in some metrics to start analyzing the marketing
channels and look at trends over time. Remember, your marketing
channels don’t equal your tracking codes, but can be really useful
for QA when used together. So we can start to
build out some reporting channel by channel based on the rules and overrides we’ve set up, but you might be thinking, yeah I’ve already done
that with tracking codes. So let’s talk about some
of the advantages we’ve got here. For instance, tracking codes
only give a single allocation, where marketing channels
can give us first and last touch. Marketing channels also
extends beyond tracking codes with channels like internal, direct, and organic search, where you might not already
be setting tracking codes, and you get the advantage
of choosing overrides and marketing channels, which you can’t do in tracking codes. So for all of these reasons, when you compare your
tracking codes and campaign to marketing channels, you’re likely to see some differences, but for QA, looking for general
alignment in these points towards a good setup, and if you crosshatch
your IDs in the detail and the values that we
collected on our processing rules setup, you can make sure those
all match up and make sure you’re collecting the
right IDs in the right channels. I’d recommend highly on
landing on a source of truth within your organization
that everybody can agree upon using. For my own organization
we use marketing channels, and for many of the reasons
that Eric and I have laid out here I feel that’s right, but because of the flexibility, I think it’s overall the best, but for your organization
that might be different. Let’s look a little bit more at reporting. I can also use my channel
detail variables that I saved in my setup to look at
the detail I’ve collected. If you remember, I wanted to lump my
branded, non-branded, and PLAs into a paid search bucket, but I still want to be able
to break those three up. That’s where looking at
that last such channel detail comes in. Can now perform all of these
breakouts in workspace, report builder, et cetera. This is also another great
way to start checking that your channels are working properly. If you’re breaking down the
channels by your detail, and some of the things
you’re seeing in there don’t make sense, or you see data that should
be in other channels, you might have a setup problem. Again, this is where that
other bucket comes into play. When we break down the
other bucket channel by a detail we can see the
others IDs that might have been missed. On the next slide I’ll show you some mine. Here you can see in the
third row of my last touch channel detail under bucket, I’ve got a campaign or
query parameter that got messed up and lumped together. So because this this got
malformed in the URL, all of these were falling
into my other bucket, where these were actually
SMS text messages we sent up, and should have been in another bucket. So this is a really great
way to catch some of those gotchas, and make sure you’ve got
everything flowing into the right channel. Again, it’s a really good
idea to create an alert to proactively inform
you of any increases in this other bucket. So now that you’ve got some
beautiful channel data, what else can we do with it? Well right here is an
example of a report that I created in Excel using report builder, breaking out some of our programs, and looking at year-over-year
changes for our visits for some of my stakeholders. For people who might not have access, or really understand Adobe Analytics, this can be super powerful
for them to still get that same view of the
data that you’ve got. Now a quick wrap-up before
we go play with some of this data in workspace. So clearly define your
channels and how you identify them. Do the work upfront and
document this stuff in your SDR. You’ll save yourself a
lot of headaches later. Get those channels and
processing rules set up. Really watch for some
of those gotchas like Google Display, the order of the buckets, URL parameters getting messed up, or dropping on redirects, and make sure that that
other bucket is there to save you. And then do a data quality
check and document any changes that you might make along the way. Finally, start some great
reporting and data exploration. Break down those channels, roll them back up, try some attribution
models like we’ll do next, and use what you already
know about your media to find the best model fit. So next, let’s hop over into workspace. So now that we’re in
workspace what we can do is we can go explore the new
attribution channel, or panel. This panel, and I’ll just drag it in here, lets us compare attribution models for any dimension or metric we’d like. Right now we’re just going
to use the marketing channel, because that’s what we’re talking about, but you can actually use any evar in here. Let’s go ahead and select the channel. I’m gonna grab my marketing channel, and drag it in as my channel here, and for my success metric
I’m gonna use orders. Then we get to pick which
models we’d like to include. I can just hover over
this and do a drop down to see the… They’ve automatically
included last touch, first touch, and linear. There are many other
attribution models included. So same touch, some j-curve,
or even a custom one. I’m going to just leave
it what’s pre-selected. Next we have to pick a look-back window. So this matters when a
user may have more than one conversion, or the visit look-back window, each visit has its conversion distributed according to the model
independently on that visit. For visitors, all the
conversions are distributed according to the model against
that visitor over time. So as a rule of thumb, I like to say, for a low frequency business, a visit is best, for high frequencies, and people are buying
from you quite often, a visitor might be best
to spread those orders over many channels. I’m going to look at the last 30 days, and that of course can be changed as well. And let’s go ahead and
build and see what it looks like. There we go. Now we get this great visual of all of our channels and how revenue
looks depending on the model. I’m going to go ahead and
drill into our affiliates. So if I click on affiliate here, you can see my affiliate
program gets a lot of last touch credit in green, but not a ton of first touch, and a little lower on linear as well. This is probably because
some people might be ready to purchase, but search affiliates for
some sort of promotion, right before they purchase, so then that last touch is an affiliate. Let’s look at paid search. Paid search on the other hand, is often high on the first touch. So there might be starting
that shopping journey with us. And it’s a little earlier
than expected actually. So when we’re doing some
paid search bidding, or keyword tuning, we might want to zoom in
a little bit on people who were early in the
funnel and capture them. This is one of those
things that our marketing team learned by doing these exercises. Let’s go ahead and scroll down and look at channel overlap as well. I’m going to go ahead and look at direct, and paid search, and email, and then it’s going to create a great Venn diagram here of those three programs. There we go. This can help me understand
the incrementality. So we can see in here that
there’s not a ton of overlap. What you might see in some
programs is a lot of overlap, and you might want to
change some of your spend, or the way that you are
addressing some of these media pieces. Let’s go ahead and unselect those three, and you can see a great
histogram here as well. The scatter plot is going
to give us all of my channels by attribution method, and we can see how far
the direct is up here in internal as well. So it’s far and away the
biggest regardless of my method. I also get a trended report, and because of my short
period of time this doesn’t look too exciting by model. I also get a great flow
report to see my channels over time model by model. So I can see the cross-channel
jumps that people are making across my channels. So you might be wondering
after looking at all of this, which model should I use? Well that one’s really up to
you and your organization. One way that we’ve looked
at models is to look at which ones we already know about and incrementality thereof through
testing that we’ve done, and to determine which
model is going to fit that the best. So for instance, we’ve done a lot of
holdout testing to help to determine how effective
are paid searches right? When we know the
incrementality of that from our holdout testing, and we find a model that fits
those beliefs really well, we don’t have to keep
holding out and we can just follow the model. So we get a lot of value
by acting off of the modeling and the data that we
get through these workspaces. That does it for the workspace. – [Eric] Great, thank you Ben. One question I have for
you is whether you’re using the same model for
marketing channel all the time, or are you aligning a
different attribution model per marketing channel, or how are you handling that? Is it based on metric or dimension? What’s the you know, best practice that you’ve
defined when it comes to attribution and marketing channels? – [Ben] That’s a great question. So one of the things that
we found is during the holiday season when a lot
of our sales are occurring, some of the marketing
channel’s modeling that we may have done during the rest of the year, sort of goes out the window. So we see a lot more paid
search actually closing those orders than it does opening them. So from time to time
throughout the year we’ll look at different models. Also we use a lot of the attribution IQ technology on some of our other llevarse, such as where we capture our search terms, so across the site we
can see what people are searching for in their shopping journey, and what sort of terms might be opening or closing those orders. – [Eric] Yeah that’s great. So you’re using a different
model depending on the channel, depending on the dimension, and depending on the time of year. That’s really great. Good thinking. – [Ben] Exactly. – [Eric] Awesome. Alright, well with that, that’s everything that
we had to show when it comes to marketing channels. Thank you so much for joining us for this Analytics Insider webinar. Don’t forget to hit those polls. Thanks everyone.

Leave a Reply

Your email address will not be published. Required fields are marked *