The Iterative Marketing Podcast Ep. 49: Moving From Attribution To Contribution

The Iterative Marketing Podcast Ep. 49: Moving From Attribution To Contribution


Hello, Iterative Marketers! Welcome to the
Iterative Marketing Podcast where each week we give
marketers and entrepreneurs actionable ideas, techniques,
and examples to improve your marketing results. If you want notes and links to
the resources discussed on the show sign up to get them emailed to you
each week at iterativemarketing.net. There you’ll also find
the Iterative Marketing blog and our community LinkedIn group where you can share ideas
and ask questions of your fellow Iterative Marketers. Now let’s dive into the show. Hello everyone and welcome to
the Iterative Marketing podcast. I am your host Steve Robinson and with me as always is the smart
and savvy, Elizabeth Earin. How are you doing today, Elizabeth? I am doing great. Staying really busy with
work this week. That’s right. I don’t think we’ve
told our audience yet that you started a new role,
didn’t you? I did. I did. I started working for a startup
that’s located in my hometown Coeur d’Alene, Idaho,
northern Idaho and they produce
cold brew coffee equipment. That’s different. – It is.
– Do you get free cold brew coffee? I don’t get free cold brew coffee. We are more on the
production side of things. So, some coffee companies
from around the world send us their coffee to
test in the machine so they can see if they like
what the cold brew tastes like and so by that point you
don’t necessarily want to drink it because we’ve run
it through a few times but I have been introduced
and heard about some really interesting
coffee companies again all over the world. So, it kind of encourages
me to get out and not just always
swing by my Starbucks but try the little the mom-and-pop
places on the corners because they’ve got some
pretty cool stuff going on. Are you turning into a coffee snob? Oh! gosh, no. I’ll take coffee any
old way I can get it. We talked about last episode. I had a second child. I have two children now,
a 2-1/2 year old and a 4-month-old and I am not ashamed to say that there are mornings that I simply
reheat yesterday’s leftover coffee because I don’t have the
energy to make a new pot. So, that does not make
me a coffee snob. You’re well in the clear of
becoming a coffee snob. No, no, but I do appreciate
a good cup of coffee. So, if we’re not talking
about coffee snobbery today what are we talking about today? So today we’re talking
about attribution modeling and this is a really hot topic because it’s something
that everyone wants. CEOs are asking CMOs
and directors of marketing and everyone in the organization to be able to say
what it is that ties back, what sales have tied back
directly to their marketing efforts. The problem is that the attribution
model as we know it is flawed. Yeah. So, attribution modeling it does a really good job of giving
management the numbers that they want but as we’ll talk about today those numbers aren’t
necessarily as meaningful as most of the industry
tries to make them seem. And we have a solution to this or at least something
that we found that works and that’s contribution modeling and so today we’re going
to introduce that concept and we’re going to talk about
how you can track contribution versus attribution. And then we’ll end with
a nice little warning because there with every good
solution there are caveats and so we’ll talk about when contribution modeling doesn’t
do itself justice as well, so… Perfect. Well, I think it makes
sense to start by diving into what exactly
is attribution modeling. Yeah. It’s not a really easy kind of easy
thing for people to understand, it’s essentially a measurement for
return on investment for marketing. So, it’s a way to determine
you put this much money in and then how much money do you
get out at the tactical level, right? So to figure it out you’re
going to take your returns or for example your revenue and you are going to divvy that up allocating a slice of the pie to each
of the different tactics that you used that led to that conversion. And there’s a couple different ways
you can divvy up that pie. Your pie is still the total amount
of revenue you got but you can divvy it up
in different ways. So, three common models are what
are called last touch attribution, where the last marketing tactic that
that particular prospect touched before they became a customer gets all of the
money from that sale. There’s first touch attribution where the first marketing tactic that
that person touched before they became a – on their way to becoming a customer that touch gets all of
the money for the sale and then there’s something called
multi-touch attribution where you apply some
sort of statistical model to spread that revenue out
across multiple touch-points based on their weight or how important they may have been
in that particular customer journey. All of these come with
pros and cons and trade-offs but in our mind all of them
are fundamentally flawed. Do you want to get into why? Sure. So, when you take a look at any one
of these on their own it makes sense. You look at just the first touch
what brought them in or you look at the last touch
what is it that converted them or you try and spread
that over all of them and like you said you’re waiting
and trying to figuring stuff out, figure out what contributed
to that sale, but the problem with this is that
that none of these are perfect models and I think you have a really
good analogy to try and explain this. Yeah. I like to relate it back to a car. So, let’s say that your
marketing department is a car and in this case the money you put in
is the gasoline you put in the car and the return
you get on that is the ability for that car to take
you from point A to point B. There’s a lot of different ways
you could measure that. For the sake of this conversation
will measure it with horsepower. So if that’s the case then we should be able to look
at every component of that car and be able to attribute how many
horsepower we get from each part but that doesn’t really work out
if for example if I take the spark plugs out of the car
I have zero horsepower. Car does not go. Does not get me from
point A to point B but if I put the spark plugs back in
I’ve got full horsepower. So, how many horsepower do I give
the spark plugs exactly in this model? That’s where it starts to get tricky because if you give them everything
then your model is flawed because it’s not just the
spark plugs that drive the car, so what about all the other
things that make it work. Right and if you give them a portion, well, that’s not telling
the truth at all because it’s really a zero-sum game, you take them out
and they don’t work at all and I think a lot of marketing
tactics are like this. Wouldn’t you agree? Yeah. So we start spending our marketing
dollars on these different things, very rarely are we just doing
one thing in marketing, we’re doing multiple things. So, how do we know how much money we got back from
any one of those individual tactics? There are synergies, there are different tactics
support other tactics and not in obvious ways and we as marketers
are really fairly in the dark on the interplay between the
different tactics that we put out there and so if we start trying to divvy up
the revenue that we’re getting back it doesn’t work out because if you pull
out any one tactic it’s going to take away more
than its fair share of revenue because it’s going to hurt the
other tactics in your program. So when your CEO or your CMO says
how much did that video cost produce and you give them an answer and they say, okay, let’s cut that. It’s not as simple as that. It’s because you’re not necessarily
you might be cutting just that expense but you’re losing everything
that comes with it because to your point
everything is working together. Right. Am I understanding that right? Yeah that video could very
well be the key to the sale because it educates the
consumer on something that if the consumer
isn’t educated on they can’t continue
that journey later and so when you
take out that video it means that they
don’t know this term, that they could be Googling later
that leads them to an AdWords ad that leads them to a conversion and it’s not obvious nor is there any good way to research
what the relationships are between your various
marketing tactics. All you can know is that
when I put them all together I get 220 horsepower. And so that’s why these
attribution models don’t work because when we’re
looking at last touch it’s saying if our
last touch is an email, then it makes it
look like this email is what converted the sale and we could say everything
before that’s unimportant and you can just get rid of it but that’s not actually the case. Right. And we look at first touch we have
the same problem there and even if we look
at multi-touch it implies that if the – if I got $200 of revenue out of
this particular touch point that if I take out that touch point that my revenue is only
going to drop by $200 but that’s not true because that touch point
is probably supporting some other touch points
in that customer journey and if you take it out it’s going
to have detrimental effects, that the effects are going to cascade through the rest of the marketing
touch points in some way and it could be in a
very insignificant way and it could be in a
very significant way and you could have just pulled
out the keystone to the arch and the whole thing crumbles and no multi-touch attribution model
can accurately depict that because you’re trying to spread
a finite amount of revenue across a complex set
of marketing tactics. So that makes sense the question is
what do we do about it. Well, I think we should talk about that
after we get back from a break. So, let’s go help some people. Before we continue I’d
like to take a quick moment to ask you Iterative Marketers
a small but meaningful favor and ask that you give a
few dollars to a charity that’s important to
one of our own. This week we are asking that you make
a donation to Operation Homefront. Operation Homefront assists
military families during difficult financial times by providing food assistance,
auto and home repair, vision care, travel and transportation, moving assistance,
essential home items, and rent-free transitional housing
for wounded vets and their families. To find out how you can help please visit operationhomefront.net or visit the link in the show notes. And we are back. So we beat up on the attribution model
pretty badly before the break. Let’s talk about what
we can do instead. What’s our alternative, Elizabeth? So we call this the contribution model and what we want to know
when we’re talking about this is what extent in using your example, what extent did the spark plugs
contribute to our horsepower. We want to know did the spark plugs
make a significant contribution or an insignificant contribution. Are they the keystone to the arch or are they just window dressing? And the way we can get
at that is to understand what percentage of
successful customer journeys have that touch point in them. So in this context how often do
you drive your car without your spark plugs
and get where you need to go. Never. Not very often, yeah. I don’t know a lot about cars but I know that’s not
going to really work. Now that’s a little different though when you look at the windshield wipers, how often do you get in your car
and drive where you need to go without the windshield wipers. 17%. Yeah. We probably don’t know but the fact of the matter is you
can get where you need to go, you can get from point A
to point B in your car without using windshield wipers, maybe even if it’s raining and you don’t
really care about pedestrian safety but the point is that we can look at the
contribution of the windshield wipers versus the spark plugs in getting you where you need to go and there is a significant contribution
made by the spark plugs, the windshield wipers help and the radio is probably
not all that important at all. So I just want to sum this up. So in using your example
we’ve determined, the spark plugs have 100% correlation
with getting where I need to go in my car whereas the windshield wipers only have 17% or whatever number
it was that I made up a few minutes ago. So, when I’m tasked
with making a decision if I have a limited budget
and I can only replace one thing I now know that I should
replace the spark plugs because without the spark plugs
I can’t get there but without the windshield wipers
there’s still a chance I can get there. Exactly, exactly. So if we bring this back in
the marketing world you can now say well that
particular video that we have, well, that’s really really important and if we’re going to redo anything
with this new branding campaign we need to make sure that
when we refresh the brand we refresh that video because we can’t kill it and at the same
time we might say, well, you know that email
drip sequence that we send when somebody signs up
for the newsletter and the people that open that aren’t
necessarily the same people that buy and it’s not really very
strongly correlated and we can do without that
for a while or maybe forever. But to be able to do that
we have to have some data. Yeah, we have to be able
to track correlations, so we have to have the data in
some sort of a third party system to be able to track what our marketing inputs to a
particular individual’s customer journey or an account’s journey in the
case of account based marketing or a lot of B2B organizations, what were the inputs and then what were the outputs and be able to mathematically determine
whether there is that correlation. Now some marketing automation
systems can do this for you or do at least a decent job of it. You may have to do some
trickery on the back end, appending tactics to contacts
and account levels and kind of figuring out where
which touch points impacted or were included in
the customer journey but that’s as simple as downloading the
data and doing some analytics on it. Yeah. It’s a fair amount of
work to get set up but you can get this data out of a
marketing automation system. There are other analytics
packages that’ll help and one will probably announce
or talk about in a future podcast that we’re working on our side but the point is that you have
to get the data into a system that’s being able to
follow individuals and follow accounts
through their journey so that you know which
touch points occurred and then download it and run an
analysis on it outside of that. Which this isn’t too far off of what you had to do with your
attribution tracking anyway because you had to have
something in place with traditional attribution
tracking, correct? Correct, correct. It’s a little bit more advanced than
a lot of last touch attribution because in that case
you just needed to know okay, where did they click on before
they clicked buy or submitted a lead. That was a lot easier but it’s no more complicated
than a multi-touch attribution, it just isn’t baked into a lot of
attribution modeling systems so you have to go outside of that. So it’s more complicated
than first or last touch but not as complicated or equally
complicated as multi-touch. Yeah, it’s actually less
complicated than multi-touch because in multi-touch attribution you either have to have
a machine determining how to weight the individual touches or you have to go in and this is
more common as the marketer and make guesses at how important
this touch is compared to that touch and really you don’t necessarily know. Well and again back to the point of one individual piece on its own
might not do the job but when you combine
two or three pieces together it tells this story that leads to the ultimate result
of someone making a purchase. So it gives you its work but it gives you some really
great data that you can use to help steer your future
marketing activities. Yeah and I think will write
a blog post in the future that will kind of detail exactly how to get this data out of
a marketing automation system because we’ve done it,
it’s not easy but it’s doable. Now when we first started
talking about this we mentioned that there
is a little bit of a catch. Yeah. No solution is going to be perfect and well we believe that this gives
you much more actionable data than attribution modeling does. It does come with an issue and we describe it
as a feedback loop or kind of an echo chamber that
you can create with your marketing. So to paint a better
picture of this, let’s look, sorry, I am laughing
at my own joke, let’s take a look at the
paint job on our car analogy. Right. So when we looked at the radio
or the windshield wipers we could prove that there was
no significant correlation between the radio and getting
from point A to point B and we could take a look at
the windshield wipers and say, well, what percentage of journeys
did you use windshield wipers or need to use windshield wipers, well, it’s pretty low but we look at the paint
on the car and we go, well, when was the last time you drove
the car without the paint on it. Most people not very often. Yeah, I haven’t really
done that lately. So, we don’t really know
how important the paint is just observing to getting
from point A to point B, for all we know we take the paint
off the car and it stops working and this problem can happen when you are too good at your
job as a marketer essentially, right? So the better marketer you are the more consistent of an experience that you’re going to be
creating for your audience and then the more
consistent the experience the more likely that every
single completed sale is going to have
touched your marketing, all of your intended
marketing touch points and as a result that
creates a correlation. It’s kind of this loop, right? Yeah or a self-fulfilling prophecy. I mean you made sure that every
prospect hit the intended tactics that you wanted to make
fall along their journey and in doing so you made sure
that every sale hit those tactics and now you don’t
have any actionable data because when you go
to look at all of your sales they all contain the same tactics, the ones that you set out to
make sure they all contained. It paints a picture that
correlation equals causation which isn’t necessarily true, right? Correct. So we had had a conversation
about this prior to the podcast and I think you found
a really interesting story but really what it comes down to is when we start to see something where it looks like two things are
related but they’re not really related. Right. And the example that statisticians
use all the time is ice-cream sales. So when I say that ice-cream
controls the weather that’s not true even though I can say that when
ice-cream sales go up it gets hot outside. Mm-hmm. The problem is that
logic is exactly backwards and that’s obvious in the case of
ice-cream sales and temperature but it’s not always
obvious in other instances and then you also have crazy correlations
that happen for no particular reason. I think you would found
one on a website, I don’t remember the name of it, will link to it. I don’t know if you remember it. correlated.org I think, yeah correlated.org
– Yeah, I think that sounds right, but there’s a correlation
between people who like candy corn and those who are not concerned
about the privacy aspect of their home automation. So looking at that statistic you would think that people
that like eating candy corn don’t really care about home security but that doesn’t necessarily make
sense there’s not a correlation there. Right, right. So the bottom-line is correlation
does not equal causation, so just because there’s
a strong correlation between this particular marketing tactic
and sales that come out the other end doesn’t always mean that there’s a
causation that those sales came out because that marketing
tactic fell on that path. The only way you can combat this is to purposefully after you
get really good at your job and every sale has all the
same touch points on it purposely introduce noise. The good news is from
the work that we’ve done no marketer is perfect, ourselves included, and we end up introducing this noise
whether we mean to or not but it is a danger particularly with things
that happen for towards the very end of the sales funnel where you are able to really control
that buyers’ journey in more detail, you can create a feedback loop, an echo chamber where we’re
feeding in the experience and then we’re measuring that the
experience correlated to the outcome and it is a danger and there are ways you can get right
around it with statistical models, you have to go kind
of one step above but as long as we’re
aware of this as marketers and we check it and check the data then I think that we can
work around that and I still think that this is a far
better model than attribution modeling. So, I think now is a great time to sort
of sum up what we’ve talked about today and that really starts with attribution and the fact that it’s a non-starter. Unfortunately it gives us the
illusion of understanding on our return on investment
at a tactical level but unfortunately that’s
just that’s not true, it’s giving us some
incomplete information. And every marketing plan much like a car has
critical components to it that if you take them out
the plan just doesn’t work anymore and if you put them
back in the plan does and it’s important that any analysis we
do takes that idea into consideration. And that’s why modeling
the contribution or more precisely the correlation of marketing tactics to
successful program outcomes is much more realistic and accurate
of a picture to report on to executives. There is the one chink however
in contribution modeling’s armor and that is that
without any noise, without any inconsistency
in your buyer’s journeys every buyer’s journey looks the same and every single touch point is strongly
correlated to successful outcomes because it’s strongly correlated
to every outcome and so that’s something that
we need to be mindful of and aware of and account for
when we’re doing this sort of analysis. And with that I’d like to thank
everybody for your time today and until next week
onward and upward. If you haven’t already, be sure to
subscribe to the podcast on YouTube on your favorite podcast directory. If you want notes and links to
resources discussed on the show sign up to get them emailed to you
each week at iterativemarketing.net. There you’ll also find
the Iterative Marketing blog and our community LinkedIn group where you can share
ideas and ask questions of your fellow Iterative Marketers. You can also follow us on Twitter. Our user name is @iter8ive or email us at
[email protected] The Iterative Marketing Podcast is
a production of Brilliant Metrics a consultancy helping
brands and agencies rid the world of marketing waste. Our producer is Heather Ohlman with transcription assistance
from Emily Bechtel. Our music is by SeaStock Audio,
Music Production and Sound Design. You can check them out
at seastockaudio.com. We will see you next week. Until then onward and upward!

1 thought on “The Iterative Marketing Podcast Ep. 49: Moving From Attribution To Contribution

Leave a Reply

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