Measuring and Predicting Marketing Performance

Report as inappropriate Your report has been received

Description

The ability to measure and then predict your marketing effectiveness opens up a new set of opportunities that will not only optimise your business performance but help to cultivate innovation within your marketing activity. Point of View By Azlan Raj, Marketing Strategy & Analysis, SapientNitro

Shared by: sapientnitro

Tags

analytics channel, attribution marketing, performance measurement media, mix, modelling predictive, analysis sapient sapientnitro

Stats

views:: 14
posted:: 10/25/2012
language:: English
pages:: 11

Traditional Copyright:All rights reserved

Document Sample

POINT OF view

Measuring and Predicting
Marketing Performance
By: Azlan Raj, Marketing Strategy & Analysis, SapientNitro

In the current economic climate, which includes a saturated and competitive online market, the
ability to convert customers within a new age of consumer behaviour is becoming increasingly
difficult (and expensive).

There are three ways in which you can increase your profit margin: reduce costs, increase revenue,
or both. To increase your online revenue, you can optimise on-site or you can throw money at your
marketing efforts. However, instead of increasing your marketing spend in a time where businesses
are tightening their budgets, you can also look at optimising your existing budget. Understanding how
consumers engage with your brand touchpoints is the first step to maximising your marketing efforts.

The ability to measure and then predict your marketing effectiveness opens up a new set of
opportunities that will not only optimise your business performance but help to cultivate innovation
within your marketing activity.

INTRODUCTION TO MEASURING AND PREDICTING MARKETING PERFORMANCE
Marketing has evolved over the last 50 years from billboard advertising right through to present day
digital engagement. Whilst walking on the streets of London, I have seen digital billboards, digital
signage, and even digital bins. If you walk into a store, you may see hybrid tills and computers with
store assistants using iPads instead of running upstairs to check stock. We now live in an entirely new
digital age.

Source: Mat Siltala, Dream Systems Media

This evolution has happened as part of a constantly changing society that has quickly adapted to the
new technologies it has been presented with. It shows the adaptability of today’s simplicity-driven
consumer. Children as young as 12 months are now able to use these technologies, and it’s forcing
us to quickly evolve.

Marketing has progressed accordingly, adapting to this new age of technology in order to meet
consumer needs. However, analytics is often overlooked through this growth, and whilst more
advanced techniques in measuring performance (through digital and even traditional channels) have
come to fruition, the ability to effectively measure a multi-channel environment is still in its infancy
in most businesses despite being high on the corporate agenda. Over 50% of C-Suite executives see
data and analytics as a top 10 corporate priority, with almost 10% placing it right at the top of their
agenda. This shows just how much higher a priority analytics should be in profile and efficiency than
it currently is.
DIGITAL BUSINESS PRIORITIES ON CORPORATE AGENDA
(% OF C-SUITE EXECUTIVES) MAY 2012

BIG DATA AND ANALYTICS 91 62 61 4 20 13

DIGITAL MARKETING AND SOCIAL TOOLS 91 62 61 4 20 13

FLEXIBLE DELIVERY PLATFORMS 61 22 31 5 23 17

Top corporate priority Not a top corporate or BU priority

Top priority for 1-2 business units Top 10 corporate priority

Top 3 corporate priority Not on the agenda

Source: McKinsey & Company, Marketingcharts.com

In addition, there is much talk about “Big Data” and the ability to measure every item of data, from
every system, in every channel. However, few companies take the opportunity to effectively measure
the data that is available now.

The ability to track marketing (in most cases) can be completed without the entire headache of getting
legacy systems to talk to each other, yet very few businesses are doing this in a way that will drive
business change. The large majority of businesses don’t effectively attribute their channels to measure
return on investment (ROI).

A recent study from IBM has shown that 78% of CEOs believe that marketers aren’t empowering their
teams to focus more on ROI and are losing sight of their main objectives. This is something that can
be addressed by correctly investing and shifting the culture of a business from the top down in order to
become a performance-driven (data-driven) organisation.

Businesses need to consider their more tactical options as well as their wider strategic initiatives. It’s a
fine balance, but some of the tactical work can feed into the wider data needs of the business.

HOW DO YOU EFFECTIVELY MEASURE AND PREDICT MARKETING PERFORMANCE?
There is a lot of focus on two buzz phrases at the moment: “channel attribution” and “media mix
modelling.” Search for either of these phrases in Google, and you’ll receive a list of blog posts about
what they are and how effective they can be. These articles aren’t wrong. There are some very lucrative
business benefits to using both. However, there is a level of understanding and maturity that goes
alongside using either approach. In order to start using them to efficiently calculate ROI, you must first
understand the differences and how they can be applied within your business.

Channel Attribution
Channel attribution allows businesses to look at the historic performance of marketing activity at a
journey level. Single-touch journeys account for approximately 50% of conversions online. This means
that the remaining 50% contain more than one touchpoint. Allocating performance solely to one
channel can be a misrepresentation of the true effectiveness of your marketing activity.

Key Benefits of Channel Attribution
• Allows you to acquire cost savings by allocating more accurate spend across a multi-channel journey
(e.g., payments to affiliates)
• Allows you to understand where marketing channels have underperformed at channel level,
campaign level, product level, and media level

Media Mix Modelling
Media mix modelling helps to predict the effectiveness of marketing. Using the historic data from
channel attribution, media mix modelling leverages on the power of predictive analysis to forecast
which channels will perform better based on certain criteria.

Key Benefits of Media Mix Modelling
• Allows you to use channel attribution data to predict future performance. It can predict the likelihood
that a channel such as natural search will be more or less likely to convert when preceded by display
or social
• Allows you to look at permutations and combinations to calculate which channels work most
effectively together
• Allows you to stay ahead of your competitors by optimising your spend based on journeys that work
for your customers
• Allows you to reduce future spend whilst potentially increasing conversion (and ultimately ROI)

WHY USE PREDICTIVE ANALYTICS?

Media mix modelling combines the data obtained from channel “Predictive modeling is the area
attribution with the insight of predictive analysis to help define how of data analytics concerned
marketing budgets can be spent in the future. with forecasting probabilities
and trends”
Predictive modelling should only be addressed once you reach a
certain level of confidence with your attribution modelling. However, Predictive Modeling Resources, 2012

there are many questions that your channel attribution model should
leave you with, which predictive analysis could help answer:

• Why is this channel underperforming, and will it work better in any other scenario?
• Why does it (or doesn’t it) perform better?
• Is there a tactical system that can be used to maximise marketing efforts?
• How likely is my campaign to succeed based on this past experience, and what impact will it have?

Predictive analytics will allow you to understand the likelihood that a channel will further perform
within a particular scenario, and the deeper impact that may not be initially noticed by merely looking
at the attribution alone.

But why do you need predictive analysis if you already use advanced attribution modelling? Predictive
analytics delivers information outside of channel attribution. As mentioned previously, media mix
modelling aims to deliver future insights rather than reflect on the historic data. The difference here is
that media mix modelling (predictive analytics) uses the channel attribution data as a springboard to
deliver further insight. This is why it is imperative to have a robust attribution model in place prior to
introducing predictive modelling.

Once in place, predictive analytics will allow you to address some of the key questions above. For
example, if Channel A is performing well and we increase or re-allocate budget:

a) Will it improve its performance in the short term?
b) Will it perform better under different journeys? (e.g., before or after Channel B)
c) What is the likelihood that it will improve performance longer term?
d) What effect could this change have on other channels? (e.g., Channel B)

In order to be at a level to deliver the answers to the above questions, you need to evolve your
marketing analytics so that you can confidently understand your marketing performance.

THE MARKETING ANALYTICS MATURITY MODEL
Assessing your maturity within marketing analytics is the first step in understanding how to develop
your marketing efforts beyond basic tracking. SapientNitro’s Marketing Analytics Maturity Model
outlines the five key stages in developing your marketing analytics maturity in order to ensure that you
maximise your spend.

Source: SapientNitro,2012

1. Reporters
Reporters look at historical data and produce flat content reports to which they can monitor what
is happening at the most basic level. Reporting is about reviewing through hindsight. The reporting
culture is focused on “sense checking” that the data matches the expectations of a particular
campaign.

2. Measurers
Measurers will take into account success criteria and ensure that valuable key performance indicators
(KPIs) are in place, whilst aiming to maintain data integrity. Measurers will also aim to evolve analytical
data into something that can drive actionable business change. Measurers start to look at multiple data
sources and more dynamic reporting.

3. Optimisers
Optimisers look at the present and make changes accordingly. Optimising companies tend to be more
mature with regards to their infrastructure and culture. As a result, they will be more strategically
evolved with their approach to marketing analytics despite the reactive nature of optimising. Optimisers
will also start to look at individual channel performances in silos to help deliver agile improvement.

4. Analysers
Analysers ask the question “why?” in order to deliver deeper insights that can influence the strategic
direction of the business. Analysts in more analytically mature companies are valued because of a
culture that lives and breathes data–driven change. Analysing companies aren’t solely reactive but they
will take a holistic view of optimisation and targeting, whilst also beginning to deliver a multi-channel
approach to analysis.

5. Futurists
Based on the art movement, this culture of analytics maturity oozes innovation by using historic data
to forecast and predict the performance of marketing. By effectively strategising and operationally
implementing changes in the business, the futurist culture will be innovative by aiming to deliver the
ultimate performance in a multi-channel environment in order to ensure that a wider picture is always
accounted for.

WHAT ATTRIBUTION MODELS ARE THERE?
A number of different attribution models exist; all of which are named depending on the author. Each
model has its pros and cons, but the appropriate application of these are dependent on your business.
Exploring a range of models can help you to understand where marketing channels are delivering (or
under-delivering) and give you a clearer picture of the end customer.

Last Touch Allocation
Last Touch Allocation is the most common method of attribution and is used in the majority of
businesses today. Last Touch measures the final channel in a customer journey and allocates all
purchases and conversions against this final channel. Therefore, all success metrics are given to the
channel that “closes” the sale.

First Touch Allocation
First Touch Allocation measures the first channel in a customer journey and allocates all purchases
and conversions against this channel. Therefore, all success metrics are given to the channel that
initially engages the consumer.

In this example, all revenue would be attributed to the display channel:

Last Touch Weighted Allocation
Last Touch Weighted Allocation measures all the channels in a customer journey and allocates
all purchases and conversions against touchpoints depending on the position within that journey.
Therefore, the closer the channel is to the Last Touch, the higher the allocation. Higher positional
weighting means that all success metrics are hierarchically distributed to each channel accordingly.

In this example, all revenue would be attributed to channels depending on their position:

First Touch Weighted Allocation
First Touch Weighted Allocation measures all the channels in a customer journey and allocates
all purchases and conversions against touchpoints depending on the position within that journey.
Therefore, the closer the channel is to the First Touch, the higher the allocation. Higher positional
weighting means that all success metrics are hierarchically distributed to each channel accordingly.

In this example, all revenue would be attributed to channels depending on their position

Linear Allocation
Linear Allocation measures all the channels in a customer journey and allocates all purchases and
conversions equally against each touchpoint. Therefore, all success metrics are equally distributed to
each channel.

In this example, all revenue would be attributed to each channel equally:

Last Touch Split Allocation
Last Touch Split Allocation measures all the channels in a customer journey and allocates all
purchases and conversions based on location. Therefore, 50% is attributed to the Last Touch and the
remaining 50% is equally distributed to the remaining channels.

In this example, all revenue would be attributed as below:

First Touch Split Allocation
First Touch Split Allocation measures all the channels in a customer journey and allocates all
purchases and conversions based on location. Therefore, 50% is attributed to the First Touch and the
remaining 50% is equally distributed to the remaining channels.

In this example, all revenue would be attributed as below:

Bespoke Allocation
Different suppliers approach the Bespoke Allocation model in different ways, so ensure that you are
considering some of the common mistakes when choosing the right company to build your attribution
model. The trick to a Bespoke model is making sure that the turnaround is efficient enough to
influence change. Your model should adapt around your sales cycles to enable a quick delivery on
actionable insights.

A Bespoke model will attribute to a channel based on a number of different factors but can be
represented in the example below:

Considerations
Each of the above models has pros and cons, so the right model will depend on your business
requirements. Working with a trusted partner will help you define this but, to get you started, here are a
couple of items that you should consider when looking at attribution models:

• Does this model fairly distribute credit to the appropriate channels in a journey?
Am I looking at closing the conversion or starting it? This may vary per department within the business
and is something that you should consider based on the channel usage.

• Will this model give me the speed and flexibility that I need to be effective?
This is dependent on your sales and change cycles, but you need something that will work within your
timeframes. If your marketing plan needs to be updated weekly but you can only turn around a bespoke
model within two weeks, then your data is obsolete. Adapt according to your needs.

• Will this model allow me to predict future outcomes?
If you’re using a one-touch attribution methodology, then the benefit of understanding how channels
work together becomes insignificant. However, understanding permutations and combinations of
channels will develop your ability to spend your budget wisely.

• Will this help me reduce my costs or optimise spend and performance?
This is probably the most important question you will need to ask. Understanding whether the ROI of
implementing a model will outweigh the time and cost spent in creating the model is key in the success
of initiating an attribution or predicting an analytics approach.

COMMON MISTAKES IN CHANNEL ATTRIBUTION
There are a set of common mistakes companies make when implementing channel attribution, which
can hinder the timeliness and effectiveness of the value that can be delivered. The following is a list of
issues we’ve experienced with clients:

Data Sources
This isn’t in regards to getting legacy systems to speak to each other, but more around the fact that
multiple data sources are required to calculate attribution—and they often use different systems
that collect data in different ways and use different metrics. Many models aren’t accounting for this
difference, which can lead to data integrity issues.

For example, if you are using clicks against visits, are you taking into account that a visit can contain
multiple clicks? How are those clicks being calculated? Or, if you are measuring offline activity against
purchases, are you taking into account that offline can drive other channels, not just vanity URLs? What
impact does that have on the journey?

Tag management solutions pull this data together, but you need to check how your solution
amalgamates this data. To get the true impact of attribution, amalgamation alone isn’t enough.
Understanding your data sources and how to accurately collect your data is often overlooked, but is a
key differentiator for a higher level of accuracy.

Tag Management Solutions
Although tag management solutions pull data together, they won’t necessarily push out the models
required to give you the answers you need. Channel attribution requires an investment to be effective
but make sure you fully understand this amount of investment before making your decision; find out
exactly what you’re paying for prior to signing on the dotted line.

Again, finding the right provider will help you make the right decision on tag management solutions to
deliver the right tool for your needs.

Data Trust
Is all of your marketing activity tracked? Are you taking data trust into account? If not, how can you trust
the performance of your attribution model? For example, if you don’t correctly tag your paid search
(PPC), it will appear as a natural search in most analytics tools. Your ability to accurately tag your
marketing activity will directly impact your results.

Creating the right governance processes and levels of trust within your business is often overlooked,
but can be a key influencer when looking at channel attribution.

Future Marketing Spend
Next steps are often misunderstood within channel attribution. Many people think that attribution is as
simple as saying, “Optimise your marketing spend by moving your budget accordingly.” In most cases,
it’s not as easy as turning marketing on and off. For example, if you find that third party websites are
delivering a high conversion rate, you can’t simply increase the number of websites that link to you.
This involves research, communication with these third parties, and—potentially—an additional budget.
After this, you will need to measure what impact these new sites have on conversion in comparison
to any budget that has been reallocated. In addition, there is the added complication that elasticity is
rarely taken into account for the longer-term impact.

Here’s the scenario: You look at your attribution model(s) and realise that you need to increase your
paid search budget. Once this happens, you realise that your conversion drops and it’s not as effective
as you initially predicted. Few companies take into account that the increase in traffic may increase
irrelevant traffic or, more importantly, can reduce ROI due to the fact that the marketing may hit the
point of saturation and that each channel could potentially peak, reducing productivity. Any predictive
models have to take into account the elasticity of any marketing predictors.

Also, what is the halo effect on other channels? Is there an impact that actually decreases the
productivity on a wider level? Even though there is a certain level of confidence when making changes,
this is a test and should be monitored accordingly.

Total Attribution Performance
Attribution should be like good analysis and optimisation—ongoing and recurring. Iterative
improvements to attribution help to continually optimise your marketing activity.

Don’t attempt to calculate total attribution all at once. The upfront investment, in time, will not pay
short-term dividends. Graduate your modelling to evolve with your business. By the time you calculate
attribution with a bespoke model right down to the product or activity level, your sales cycle will be
likely to render your data invalid. The product can become out of date, the campaign may finish, or the
impact to change could reduce over time (reduction in competitive advantage).

Think quick wins versus longer-term goals when looking at marketing optimisation.

WHAT ATTRIBUTION TOOLS ARE AVAILABLE?
Although not the primary purpose of this paper, it’s difficult not to mention the tools that can deliver
attribution data. There are a number of tools that have their associated strengths and weaknesses.
Forrester has released their findings and point of view on some of the available attribution tools:
Risky Strong
Bets Contenders Performers Leaders
Strong

Visual IQ

Adobe
ClearSaleing
GroupM
Current
ering Converto

Market presence

Full vendor participation
Weak

Weak Strategy Strong

Source: Forrester Wave: Cross-Channel Attribution Vendors, Forrester
Research, Inc.

Things that you should also consider when researching tools are tag management solutions and data
collation processes. Always ask yourself, “What is the easiest and most effective way to collect data
within my organisation? Can the tools I already have get me started?”

CONCLUSION
There are a few key takeaways that you should always think about when looking at attribution.

• Where am I now?
• Where do I want to go?
• How do I get there?
• What do I need to get there?

You won’t necessarily need a fully bespoke solution. It is dependent on your specific business
requirements, however, what you need is something that you can react to and is clearly actionable.
Continuously evolve your measurement as there isn’t one right answer, and it changes per business.

Remember, you are looking at full customer experiences and you should be analysing their journeys,
not isolated channels or behaviours. Don’t think bigger with attribution and modelling—think smarter.

About the Author
Azlan is an award-winning digital consultant who specialises in digital mar-
keting communications and performance. Working in a range of industries,
Azlan has worked with a number of blue chip organisations to help optimise

their marketing efforts through effective campaign planning, strategy, and
most importantly, accurate measurement.

Shared by: sapientnitro

SapientNitro

sapientnitro.com

About

SapientNitroSM, part of Sapient®, is an integrated marketing and technology services firm. We create and engineer highly relevant experiences that accelerate business growth and fuel brand advocacy for our clients. By combining mu (More...)lti-channel marketing, multi-channel commerce, and the technology that binds them, we influence customer behavior across the spectrum of content, communication and commerce channels, resulting in deeper, more meaningful relationships between customers and brands. SapientNitro services global leaders such as Chrysler, Citi, The Coca-Cola Company, Singapore Airlines, Target and Vodafone through our operations in North America, Europe, and Asia-Pacific. For more information, visit www.sapientnitro.com or follow us on Twitter @sapientnitro.

Related docs