Multi-touch attribution (MTA) is a game-changer for marketing – but why is it often ignored?

illustration of figures joined in connective web

MTA (multi-touch attribution) is an approach to measuring marketing effectiveness that works by linking all the known steps in a customer journey prior to a sale being made. It then uses machine learning to allocate weights to the value of each step according to the way they each contribute towards a sale.

MTA works well for all direct channels, both online and offline. However, it does not cover indirect channels such as TV, press, or brand development, all of which are best measured using econometrics (or MMM) alongside MTA.

A marketing campaign will contribute to a number of recognisable steps in successful customer journeys, each of which will be allocated a share of the value of the sale. The campaign’s value is then judged to be equal to the sum of the values of each of the steps it creates.

As well as providing accurate, timely campaign measurement where sales are not double-counted (unlike siloed media reports), MTA provides valuable insight into which individuals have been influenced by which campaigns.

So, what are the barriers to adopting MTA? Here are answers to six key questions.

1. Isn’t introducing MTA very costly because of the investment in technology and time required to make it work?

MTA does need technology and time to build the customer journeys from online data, such as click-throughs to your website, or offline steps like receiving a DM communication, and then reporting the results. But the cloud-based technology to do this exists. Depending on scale, the cost of running an MTA programme should be less than 3% of marketing costs for a marketer spending around £1 million pa on marketing, and less than 1% for a marketer spending upwards of £10 million pa.

2. Does too much data from too many sources make MTA practical?

Well, it is true that multiple data feeds are required to build the journeys, but in a typical retail case, for example, you will not need more than a first-party data feed from your website, where 100% of allowable browsing activity is needed, combined with a selected feed from your e-commerce system, your customer database, and any offline contacts, such as DM.

You may also want to introduce call centre contacts if these play a significant role in the journeys. To join all this data together, and build the journeys, you will need something equivalent to a customer data platform (CDP). However, most CDPs come with built in connectors, and there are cheaper alternatives to Google Big Query for collecting the individual browsing data.

3. Are the customer journeys too complex to weight, and can I trust the scores given to the individual steps?

It is true that some customer journeys can be complex, but others are very simple. At Unifida, we find examples where journeys are just one step, and others when they are longer than ten. The average is often between three and four. However, if the machine learning that drives the weighting is set up correctly, then the number of steps in a customer journey does not influence the accuracy of the result.

These charts give an indication of the number of steps and time length of journeys you can expect to find:

Multi-touch Attribution Graph showing distinct channels

 

Multi-touch Attribution Graph showing customer journey distributions

The best MTA technology makes the scores given to individual steps in each customer journey transparent to the user, so that you can see precisely how each step in each customer journey is weighted. This in itself goes a long way towards developing confidence in the results. The machine learning behind the weighting should be trained on each client’s data to understand the average pattern and shape of the journeys.

At UniFida we allocate the weightings according to the position of each step in the journey (i.e. is it at the start, end or middle of the journey?) and the time intervals before and after each step. So, a step will be given more value the less time it takes to get to the next one, and vice versa. We recognise there is no ultimate right or wrong in the way that journey steps are scored, but we know empirically that this approach provides what appear to be very sensible results.

4. I know that different types of customers respond differently to different types of marketing, but can I understand that from MTA reporting?

This is often a key concern given that different customer groups, such as new or existing customers, respond to marketing communications in varying ways. To be able to look at what different customer groups are responding to, you will need to import data from your customer database so that, at the moment when a sale is made, you can identify what segment the customer is in.

You will know whether it’s a sale made to a new recruit, or a high-value existing customer, for instance. With the customer segment for each sale identified, you can then filter the attribution reports down to the segment you are interested in.

5. How can I allocate the marketing costs to each individual campaign?

Marketers should have the ability to input campaign costs into their MTA platforms. However, we find that marketers often have limited time or lack access to the detail to input costs for every single campaign element. Even so, they will know how much they spent in a month on a particular channel, such as Facebook. This is where a good MTA platform can help because it can take the overall spend in a channel for a month, calculate how many customer journey steps that spend created, and then compute the average cost per successful step.

These steps can then be summed up at an individual campaign level. This will not be perfect because different campaigns will have different costs per step, but it makes for a reasonable approximation. For more accuracy on important campaigns, these campaign costs can be input individually for precise calculations.

6. After all this, when I get the actual MTA reports, will they be of any use to me?

It always possible to totally ignore the MTA reports and carry on spending marketing money based on whim and intuition, but this would be throwing away a great value opportunity.

We find that clients tend to use the MTA reports in two very different ways: first to look at the big picture in terms of the return on marketing investment (ROMI) by channel at different times of year; and secondly, by looking at the individual results for each campaign and testing to see how they are performing, and how they interact with other channels and different customer types.

The benefit from the ‘big picture’ analysis is that you can use it to shift marketing budget to where the best ROMI can be found, thus cutting out dead wood, as well as spending more when the ROMI is highest. A word of caution – although the MTA provides a good retrospective view of how marketing has performed, it does not routinely provide forward predictions. Where these are required, we suggest introducing MMM or econometrics alongside MTA to provide an understanding of the return that could be expected from different levels of future investment in different channels, which can take into account the full marketing mix and external factors.

The benefits from the micro-level campaign and test analysis are that you can change campaigns very quickly, see how they are performing, determine test results, and understand how different campaigns are interacting with each other – without double-counting sales. Your MTA platform should be reporting in near real-time to help with these decisions.

The overall benefits from introducing MTA will be a function of the improved ROMI achieved from the better understanding of marketing performance over the cost of the MTA itself. Every company is different, so generalisations about improvement should be treated with caution, but you should not be surprised to find that, once you are acting on the MTA results, you can get a x10 benefit from your investment in it.

If this encourages you to consider introducing MTA, or undertaking an MTA proof of concept, because you are struggling to get a proper understanding of the returns you are getting from your marketing, then please talk to us about it. We have experience of introducing MTA across multiple companies in industries as different as retail, subscription, insurance, news media, cruising and lotteries. At no cost, we can also give you an assessment of whether your company is a suitable case for introducing MTA, and at what level of investment.

Contact us at [email protected], or call 0203 960 6472, or check out our website www.unifida.co.uk

Read more about Marketing Attribution >

UniFida logo

UniFida is the trading name of Marketing Planning Services Ltd, a London based technology and data science company set up in 2014. Our overall aim is to help organisations build more customer value at less marketing cost.

Our technology focus has been to develop UniFida. Data science business comes both from existing users of UniFida, and from clients looking to us to solve their more complex data related marketing questions.

Marketing is changing at an explosive speed. Our ambition is to help our clients stay empowered and ahead in this challenging environment.

Institute says 80% of CDPs are showing significant benefits

According to the Customer Data Platform Institute’s Industry Update July 2023, around 80% of CDPs are showing significant benefits. Their update also reveals a number of other interesting findings that are worth sharing:

  • The importance of relating data unification projects to business goals – put simply, a single customer view is only worth what you can use it for. The most popular uses mentioned in the report were customer data analysis, orchestration of customer communications and selection of messaging. For our part, we would like to add customer journey-based marketing attribution.
  • The biggest CDP deployment problem is the client’s organisation – in effect users need to get cooperation across an organisation and then invest in the team skills required to use the capability properly.
  • CDP projects that are managed by marketers are more likely to be successful than those managed by IT – the most likely explanation we would suggest is that marketers are going to rely on their CDP to deliver marketing results and this requires having full operational effectiveness. They cannot risk it failing.
  • Companies are showing increasing concerns about privacy compliance – this is most probably the result of increased publicity being given to data security breaches and personal data privacy issues. Indeed, we find it hard to understand how an organisation can manage customer consents without a unified customer view.
  • CDPs are no longer in the category of marketing technology that only very large companies can afford – many of the survey respondents engaged with CDPs were working for companies with sales in the $10m to $100m range, which is small by US standards.

The update is fascinating reading for anyone interested in introducing a CDP, or maximising value from an existing CDP. Membership of the Customer Data Platform Institute is free, visit their website for more information.

Logo and wording about joining the institute

UniFida logo

UniFida is the trading name of Marketing Planning Services Ltd, a London based technology and data science company set up in 2014. Our overall aim is to help organisations build more customer value at less marketing cost.

Our technology focus has been to develop UniFida. Data science business comes both from existing users of UniFida, and from clients looking to us to solve their more complex data related marketing questions.

Marketing is changing at an explosive speed. Our ambition is to help our clients stay empowered and ahead in this challenging environment.

Top 8 ways a CDP can make you a better customer marketer

CDPs – Customer Data Platforms – are often just evaluated from the viewpoint of facilitating
timely, relevant and personalised customer communications. Properly configured, a CDP can
also hold a complete history of every customer interaction, from online to direct mail,
opening up very valuable insights into customer understanding.

Here are the top 8 ways a CDP can help you better manage your customer marketing:

1: Measuring how each and every aspect of your direct marketing is performing.

Having every customer touchpoint in place leading up to each of your sales is a critical
requirement for delivering customer journey-based marketing attribution. This works best
when you have a scoring system in place that can share out the value of each sale across the
touchpoints that lead up to it, according to the contribution they make. In this way you can
not only compute the value of each of your online and offline campaigns, but also look at
how different customer groups, for example new vs. existing customers, are impacted by
your marketing activities. And you can start to look at how each channel performs in
different seasons.

2: Knowing the longer-term value of customers recruited through different channels and via different campaigns.

Longer term customer value varies hugely across the different types of customers you
recruit and different channels will attract different types of customers. So, assuming you
know how a customer was recruited, you can start to understand what their longer-term
value is likely to be. This can be problematical where several channels are involved in a
single sale, but you can look at the longer-term value of all the customers gained via a
particular channel or campaign. You can also drill down further and differentiate customers
by other criteria, such as previous relationship with your brand, geography, or even age
band. Understanding customer longer-term value allows you to set maximum costs for
acquisition and get a better understanding of the returns from marketing investments.

3: Predicting lapse levels and lapse timing for subscription products.

Anyone selling a subscription product, whether it be an insurance policy or a magazine, will
know how vital it is to recruit and retain ‘sticky’ customers. Fortunately, the data residing in
your CDP should allow you to model retention with a great deal of accuracy. This is because
you will be holding not only the payment records of each subscribing customer, but also a
great deal of information about them. There are several statistical methods for doing this,
but we tend to use CHAID as it divides customers up into identifiable groups with different
expected levels of longevity. Your historic data can also be used to show what proportion of
your expected lapses are likely to happen each month, which can be of vital importance for
cash flow planning.

4: Building customer segmentations to help you better understand the needs of different customer types.

Marketers need to simplify the problem of dealing with many different types of customer
requirement. The proven way to achieve this is to build a customer segmentation that gives
you a handful of groups for which you can devise different marketing strategies, or even
different products and propositions.
There are many techniques for building customer segmentations, but we like to use one
that allows you to allocate customers with relative simplicity into their correct segment, and
also to find similar types of people outside of your own customer base. For customers within
the customer base, criteria such as value, previous sales or enquiries, and types of
merchandise they buy, often groups customers meaningfully. For customers outside the
customer base, segments are often defined by, for example, affluence or age band. Once
the segmentation structure has been developed your CDP will allow you to allocate each
customer to a segment and plan your communications strategy accordingly.

5: Managing all your GDPR consents in one place.

Customers deposit their consents in many different places. They may unsubscribe from one
newsletter, opt-in to another, decline cookies on a website but approve receipt of customer
marketing when placing an order. A marketer has to establish order in what is often a very
untidy consents landscape and then define clear communications rules about what can be
sent to whom on which pretext.
Your CDP is the one place where, through identity resolution, consents can be bought
together and organised and rules about who can get which communication established. The
CDP can also provide most of the materials for fulfilling Subject Access Requests, as well as
manage anonymisation of data when the right to be forgotten is exercised.

6: Planning business development based on a customer value model.

Businesses need clarity on the growth and quality of their customer base, not least to
understand how to split the marketing budget between acquisition and retention marketing
to meet business objectives. Your CDP will hold a record of the historic value contributed by
each individual customer and you will know what that amounts to in any historic calendar
year. It will also tell you what percentage of customers recruited in previous years typically
order in the current period. Using this information you can, with reasonable accuracy,
predict what value, for example, your customers recruited this year will contribute during
the next year and how this will be distributed month by month.
You will also know how your new customer recruitment usually lands month by month, and
the value new recruits contribute in the period from when they were recruited to the end of
the year. Pulling all this together you can calculate how many customers you will need to
recruit in a future time period to meet a specific overall sales target. We call this the
customer value model, all made possible by data held in your CDP.

7: Providing data for building response and upsell propensity models.

Predicting response by different channels can save a considerable amount of the marketing
budget, enabling marketers to avoid activity that will not produce a strong return on
investment. To build a predictive model you need a target variable, like propensity to make
a second order, as well as predictor variables, which are facts known about the customers –
in this case, both for those who buy the second product and for those who don’t.
The role of the CDP is to provide this data to the data scientist, or the AI tool, which is going
to build the model. But as well as providing the data that allows the predictive model to be
built, it also provides the data that then allows every customer to be scored up with a
probability of doing whatever is being predicted. Indeed, without a CDP, developing and
using propensity models for marketing is made very much more difficult.

8: Recruiting customers to join research panels.

Many companies like to maintain continuous panels of customers who have agreed to answer market research questions, usually in exchange for some value given back to them.
The CDP can provide randomly selected customers for recruitment to these panels, as well
as managing the exercise of sending them questionnaires and recording their responses.
Customer panels are very much simpler to manage if you start with a CDP already in place.
Overall, a CDP should be seen as an essential component in the complex process of
maximising your customer revenue through improving your customer marketing.

 

UniFida logo

UniFida is the trading name of Marketing Planning Services Ltd, a London based technology and data science company set up in 2014. Our overall aim is to help organisations build more customer value at less marketing cost.

Our technology focus has been to develop UniFida. Data science business comes both from existing users of UniFida, and from clients looking to us to solve their more complex data related marketing questions.

Marketing is changing at an explosive speed. Our ambition is to help our clients stay empowered and ahead in this challenging environment.

Q & A: Customer Journey-Based Marketing Attribution

Below are some frequently asked questions about customer journey-based marketing attribution – and some insightful answers.

 

What is customer journey-based marketing attribution?

People usually receive or respond to multiple communications on their way to making a purchase. These can be online, such as a Facebook ad, or offline, such as receiving a catalogue. Customer journey-based marketing attribution looks at all the interactions between a customer and an organisation before a sale in order to analyse the role of each interaction, and hence their contribution to the sale.

Which channels are normally included?

We expect to include all ‘direct’ channels, in other words those where there is a one-to-one relationship between the customer and the company, or where there is a direct link through an online user clicking on a digital ad (such as PPC) and visiting the company’s website. An email is direct because it goes to a known recipient, whereas a press ad is not because the recipient is unknown. However indirect channels, such as TV, press, and outdoor, can also be very important. These require a different attribution technique called econometrics.

How do you link the steps in a customer journey?

We start by putting a snippet of code on your website to enable us to download into our attribution platform all of your (first party) website visitors’ browsing activity, including, most importantly, how they arrived at your site. This allows us to distinguish, for instance, between a referral, a branded search or a natural search. We then link the browsing to any offline journey steps by joining them to individuals using identifiers they have provided. To obtain the non-web contact activity we ingest feeds into our Customer Data Platform (CDP) from sources such as your email service provider, direct mail contact history, and your order processing system.

As not all journeys lead to a sale, what do you do with the unsuccessful journeys?

The brutal truth is that for attributing value to marketing activities we ignore them, which is not to say that the unsuccessful journeys are unimportant as they are key when we aredescribing customer journeys overall and understanding the total spend through a particular channel (i.e. we need successful plus unsuccessful spend). But a campaign will only have value attributed to it from the journey steps it created that led to a successful outcome.

How far back in time do you go when looking at customer journeys?

We normally look back 90 days before each sale, although some clients ask us to look at shorter periods, e.g. 30 days. To a large extent it depends on the type of purchase and the channels used. For instance, a catalogue will have a much longer shelf life than an opened email, so we need to give it time to have its effect.

How do you decide on what weight to give to each step in a customer journey?

It is obvious that all journey steps are not of equal importance, so weighting them correctly is crucial to obtaining a successful attribution outcome. There is a great deal of online discussion about this subject, with different approaches being debated, but we have opted for a method which is mainly based on the time intervals before and after each journey step.
If, for example, an event happens just before a sale, we give it a high closing score. In contrast, if there is a long interval after the first event, then we conclude that it could not have had too much of an impact on initiating the sale. We also give credit to events that help keep the customer interested without actually closing the sale. If you would like a more detailed note on how our weightings work, we would be pleased to share this with you.

Are there certain types of event that you ignore?

Yes. Multiple opens of the same email on the same day is one example, as is a visit to PayPal just before closing a sale. We try to eliminate anything that does not contribute to the customer’s decision to purchase.

Can you distinguish the different behaviours of different customer groups when responding
to marketing events?

We can. For example, new and existing customers behave entirely differently in terms of the kinds of journeys they make and what marketing events they respond to. Another way we divide up customers is between those who mainly search and buy online and to those who order through a call centre. But you may also wish to look at the impacts of marketing on different types of customer segment, and our platform can support that.

Do you look at how different channels perform at different times of year?

We do, and we find very significant seasonal differences. To show this, we have a specific report providing month-by-month summaries so that we can, for instance, compare email or any other channel’s performance in one month with another.

Do you always look at sales when calculating marketing attribution, or can you look at other goals, such as lead generation?

We often look at non-sale outcomes, and, for instance, recently we have been working for a charity that they wanted us to look at how they get their users to take up different tools that they provide on their website. When looking at non-sales outcomes we lose the value element that we have in a sale, but otherwise the process works in an identical way.

How up-to-date are your reports?

They are always available online at any time and the data behind them is processed each night. So on any day you will be looking at results up to midnight the day before.

Do you aim to answer questions other than the value obtained from customer journeys?

We are finding that this is an increasingly important area, and to respond to our clients’ requests we are building a whole suite of customer journey reports. These will answer questions about the lengths of journeys, the mix of channels used and the sequence in which they appear in the journeys.

Is customer journey-based attribution GDPR compliant?

Yes. It uses only your organisation’s first party data and excludes cookie and analysis opt-outs, for example.

Why not just use Google?

Whether you are using Google Ads or the new or old version of Google Analytics, they have well-documented flaws – namely inaccuracy and incompleteness. Google Ads uses the last Google Ads click, but Google Analytics uses the last click across all channels, so over-reports as it does not take into account other channel’s contributions. Google recognises that this is an issue and has developed Google Analytics 4 (GA4) to
replace the previous version called Universal Analytics – but this will not solve the core flaws. GA4 will use black-box algorithms and again will not take into account all marketing activity. Google is sampled and segments cannot be applied retrospectively for analysis.
Google does not use first-party data or individual identifiers, so it cannot be joined to other data sources of marketing activity.

 

UniFida logo

UniFida is the trading name of Marketing Planning Services Ltd, a London based technology and data science company set up in 2014. Our overall aim is to help organisations build more customer value at less marketing cost.

Our technology focus has been to develop UniFida. Data science business comes both from existing users of UniFida, and from clients looking to us to solve their more complex data related marketing questions.

Marketing is changing at an explosive speed. Our ambition is to help our clients stay empowered and ahead in this challenging environment.

Delivering successful marketing attribution projects

You might think that introducing accurate omni-channel marketing attribution is a sensible thing to do and that there would be a limited chance of failure. But think again, as these projects can and do fail.

The reasons are mainly to do with what they are replacing. Prior to the start of such a project there was probably no one person with overall responsibility for marketing attribution, or even someone with it in their job description. Instead, there are likely to have been many local attribution activities, each one designed to prove to the business that team X or Y, or channel Z, was doing a good job. The most typical team silos are digital and direct marketing, yet these teams have the most to benefit from working together on attribution.

Attribution projects

It is common to find attribution projects limited by media channel, such as including selected digital channels with no account for direct channels, or limited by sales channel, such as just including online sales with no account for call centre or store sales – or both. The attribution work may have been done internally, but more often externally by an advertising agency. And the external agency may be driven by a need to keep funds flowing through their channel, rather than necessarily being focussed on the end contribution of their activities.

Some marketing attribution looks at more superficial measures such as opens and clicks, rather than at the longer-term value generated by a campaign. Again, they often ignore the fact that orders are usually achieved through a combination of customer interactions in more than one channel.

So, a new omni-channel marketing attribution project, centrally and professionally managed, is definitely going to be disruptive for certain vested interests.

Avoiding disruption

To avoid internal alienation and disruption, it is essential that the entire company management team, from finance to marketing, buys into the project from the start and has confidence in the methodology that is being used. They also need agree that they will respect the results produced, even when they may upgrade or downgrade the value contributed by certain existing activities.

However, as one US commentator recently put it, ‘algos make a unified approach possible’. The ‘algos’ (or algorithms) are what make omni-channel marketing attribution possible, and if they are well designed, they will deliver trustworthy attribution results that can be used to guide marketing spend, and optimise budget allocation in the future.

If the business is aware from the start of the consequences – as well as the immense benefits – of introducing accurate omni-channel marketing attribution, then the project can succeed, and the business can optimise the marketing budget and reap the sales and growth rewards.

UniFida logo

UniFida is the trading name of Marketing Planning Services Ltd, a London based technology and data science company set up in 2014. Our overall aim is to help organisations build more customer value at less marketing cost.

Our technology focus has been to develop UniFida. Data science business comes both from existing users of UniFida, and from clients looking to us to solve their more complex data related marketing questions.

Marketing is changing at an explosive speed. Our ambition is to help our clients stay empowered and ahead in this challenging environment.

Why marketeers must not lose sight of value-based marketing

Among the objectives marketers often set themselves, ROMI (return on marketing investment) is normally near the top. The problem is that, although it is comparatively easy to measure the overall investment, it is much harder to measure the value. Value and cost per sale have a habit of varying hugely – hence their importance to marketers.

For example:

  • The value of an insurance sale will depend on the annual premium income, the probability of lapse or renewal, and the likelihood of a claim
  • The value of most retail sales depends on what the customer does next. Do they repeat, buy something different, or never use the retailer again?
  • The value of a media subscription relies mainly on their longevity, but can include some cross sales
  • The value from recruiting charity donors is based on the expectation that they will continue to act with generosity.

It’s quite a challenge however to think of a category of sales where there is just a single fixed value. A few come to mind, including:

  • Estate agents selling houses are unlikely to factor in an individual’s next purchase
  • Repeat funeral plan sales are rare, unless the purchaser is buying additional plans for their loved ones
  • One trip on a Virgin Galactic VSS space plane should be sufficient for most people!

Investing where the value is

But understanding the longer-term value of each sale can secure a company’s future because its marketers can place their marketing investments where the value is. Here are a couple of examples:

  • One of our retail clients is selling a product for which there is no necessity for customers to repeat purchase. Nevertheless, it has enormously divergent examples of customer behaviour when it comes to doing so. The bottom 50% of customers recruited have an average life-time value of £50, while the top 5% have an average of £2800.
  • When we analysed the value of policies sold by another client, a life insurance distributor, we found that, depending on where and how the customers were recruited, the contribution per policy sold varied from + £404 down to – £277, after taking into account marketing costs and expected clawbacks from lapsing.

So why do marketeers shy away from looking at the value they are really creating, and yet invest time in looking at explanatory metrics, such as cost per click-through, or the number of impressions? We guess it’s because predicting longer-term value goes into the ‘too difficult’ category and gets conveniently ignored.

But for those prepared to take the plunge and align their spend to longer-term value, there are two necessary parts to the analysis – accurate marketing attribution to determine the cost of making each individual sale, and an approach to predicting the longer-term value of each sale once it has been made.

Determining the cost of each sale

Acknowledging the fact that in today’s world most sales come at the end of a customer journey, this requires joining together the steps in each journey and then understanding what they cost to deliver.

As the charts below show, customer journeys may involve multiple channels and continue for some time.

Distribution of Distinct Channels in Customer Journeys

Distribution of Customer Journeys length

 

The technology required to do this must link both online steps – such as click-throughs to a website from social media – and offline steps, like receiving an item of direct mail. Having joined the steps together, there is the question of how much they each cost. In our opinion, that should be the cost of a campaign, divided by the number of steps where it contributed to journeys that end up with a sale actually being made.

An email campaign may be sent to 10,000 people, but only contribute to 100 sales, so the effective unit step cost of the campaign is just 100th of the overall cost. This calculation can then be further refined by sharing the sale value disproportionately between the steps that led up to it, according to the relative contribution to the sale that they made. Having attributed a cost to each step, these can be summed up to provide a cost per sale.

Predicting longer-term value

Deciding how to do this will be driven by the industry sector and the sale type being made, whether, for example, it is a sale recruiting a new customer, or one to an existing customer. If we take as an example just one type of sale, like an insurance policy or a media subscription, then predicting lapse becomes critical to the value equation. There are many different techniques for doing this, but our preference is to use CHAID* to predict the overall probability of someone lapsing within a given time period.

Chaid Model diagram

This kind of technique will divide policies sold into distinct groups, each with a different expectation of lapse rates, based on the known characteristics of the customer and the policy they have bought.

The next question is: when will they lapse inside that period? This is where we use historical evidence based on different lapse timing for different cohorts of policy purchases. If, however, the sale is a retail one, then we will be looking to forecast for each recruit their expected future value within the next season or year. Every business will have its own unique requirements for predicting longer term sales value.

Providing the best ROMI

All this may, when viewed in the round, look somewhat difficult to achieve and there are many compromises that can be made when aiming to link marketing investments to their future value delivered. Costs per sale may be grouped into costs for a particular product category and value may be averaged over a large cohort of sales.

However, we strongly believe that, given the vastly varying value of individual sales made, and the importance of recruiting customers that provide the best ROMI, it is essential to go down this route.

*CHAID analysis (Chi Squared Automatic Interaction Detection) is a statistical technique is used in market research.

 

UniFida logo

UniFida is the trading name of Marketing Planning Services Ltd, a London based technology and data science company set up in 2014. Our overall aim is to help organisations build more customer value at less marketing cost.

Our technology focus has been to develop UniFida. Data science business comes both from existing users of UniFida, and from clients looking to us to solve their more complex data related marketing questions.

Marketing is changing at an explosive speed. Our ambition is to help our clients stay empowered and ahead in this challenging environment.

Jo Young becomes Managing Director of UniFida

UniFida, a provider of customer data solutions for mid-sized companies, has appointed Jo Young as Managing Director.

She was previously Client Director and succeeds UniFida’s Founder, Julian Berry, who becomes Executive Chairman of the company.

Speaking about her new appointment, Jo Young said: “I am very excited to take on this new role as part of UniFida’s exciting growth plans. I will continue to work with existing and new clients, helping them to access deeper customer insights and deliver customer experiences that drive sales and maximise marketing returns, using our CDP, marketing attribution and data science solutions. I will also be helping to further develop new sustainability innovations, such as our CO2 Counter for marketing.”

Jo is a leading data consultant with extensive marketing analysis experience across multiple disciplines, including digital, database, research, econometrics and pricing. Previously she launched an award-winning econometrics division and has led ground-breaking projects to combine first-party marketing attribution with econometrics for leading-edge media measurement and optimisation.

Jo Young, Managing Director of UniFida
Jo Young, Managing Director of UniFida

Jo has worked at some of the UK’s leading data consultancies, pushing the boundaries of what is possible with data, analysis and technology with companies such as Lakeland, John Lewis, AXA, TUI, Cancer Research, Unibet, P&G, Arriva, American Express and O2.

 

UniFida logo

UniFida is the trading name of Marketing Planning Services Ltd, a London based technology and data science company set up in 2014. Our overall aim is to help organisations build more customer value at less marketing cost.

Our technology focus has been to develop UniFida. Data science business comes both from existing users of UniFida, and from clients looking to us to solve their more complex data related marketing questions.

Marketing is changing at an explosive speed. Our ambition is to help our clients stay empowered and ahead in this challenging environment.

UniFida Customer Data Basics: Session Three – Four Key Customer Data Questions

In this third video in the series, UniFida’s Client Director Jo Young talks about four key customer data questions you need to ask to drive sales, and discover your customers.

We’ll explain the four questions, why are they important, the best way to answer them quickly and what benefits can we expect. 

 

Play Video
UniFida logo
UniFida is the trading name of Marketing Planning Services Ltd, a London based technology and data science company set up in 2014. Our overall aim is to help organisations build more customer value at less marketing cost. Our technology focus has been to develop UniFida. Data science business comes both from existing users of UniFida, and from clients looking to us to solve their more complex data related marketing questions. Marketing is changing at an explosive speed. Our ambition is to help our clients stay empowered and ahead in this challenging environment.

UniFida Customer Data Basics: Session Two – What is a Marketing Attribution?

In the second video of UniFida’s Customer Data Basics series, Jo Young talks about Marketing Attribution.

What is it? Why have it? What is the best way to do it? And what benefits can you
expect to derive.

Play Video
UniFida logo
UniFida is the trading name of Marketing Planning Services Ltd, a London based technology and data science company set up in 2014. Our overall aim is to help organisations build more customer value at less marketing cost. Our technology focus has been to develop UniFida. Data science business comes both from existing users of UniFida, and from clients looking to us to solve their more complex data related marketing questions. Marketing is changing at an explosive speed. Our ambition is to help our clients stay empowered and ahead in this challenging environment.

Marketing Attribution for not-for-profits – it’s not just about sales

Multi-touch Attribution (MTA) for marketing is all about evaluating the effectiveness of marketing investments by examining their role in generating steps in the customer journey. It was initially developed for ecommerce, but not all MTA is about sales.

In the case of not-for-profit organisations (NFPs), it could be about attributing marketing to successful volunteer sign-ups, or donations. Not all customer journey steps are equal – and for accurate results, it is important to ensure that the approach used includes all the relevant channels, and that calculations can include every event in the journeys towards the target outcomes.

The greater the transparency, the better the insight – and the more opportunities there are to optimise marketing to help achieve NFPs’ marketing goals.

UniFida has been working recently for an NFP, an alcohol awareness charity whose objective is to lead people towards areas of their website that offer support. In this case we have been able to plot their customer journeys leading up to the use of one or other of their tools, and from these evaluate the contribution of their different digital marketing campaigns.

This same approach can be used for evaluating marketing in a number of areas, for example, test drives for cars, university courses, or media content. However under the bonnet, the technology and the analytics remain very much the same.

How does MTA work?

1. Collating online and offline customer journeys

This requires a customer data platform (CDP) or equivalent to bring together, for instance, web browsing actions with email opens, and perhaps phone calls and direct mail sends. Each journey will end with the designated result, but the steps to get there need to be assembled in time sequence.

This requires a 100% data feed from the website (from which we collect first-party data), a feed from an email service provider and a contact history file for the direct mail if needed. The average number of steps in a journey are around three, but the average is deceptive as the steps can range from one up to ten or more. Existing customers also usually take more steps than new ones.

 

MTA diagram

In this example, the journey started on the 12 March and ended with a sale on 1 June. The recipient opened two emails and a catalogue, and undertook two entries to the website (one via PPC and the other going directly).

 

2. Weighting each step in the customer journey

All steps are not equal and can play different roles. They can help initiate a journey, maintain a customer’s interest and help reach the end result. We use advanced mathematics to undertake the weighting of the steps and this also looks at the role they play. We split the roles into Initialiser, Holder and Closer.

We train the mathematics algorithm to take account of the particular characteristics of the journeys for each individual client, as they can vary considerably. Higher priced items, or choices with more consequences like university courses, will tend to have longer journeys as more consideration is usually required. The algorithms respond to the time periods before and after each step as well as, for a browsing step, the level of engagement with the website.

MTA diagram

To continue with the same purchase example above, we have now added in the scores for each step in the journey. The IHC score is the combination of the Initialiser, Holder and Closer scores. Each column adds up to a total of 2, so a 0.4978 IHC score will give that step a quarter of the sale vale. The two emails received some credit for Initialiser, but not as much as they would have done if the sale date had been earlier, whereas the Closer rewards went to the catalogue and PPC.

 

3. Aggregating the results

Some marketers are interested in puzzling over individual journeys to understand how channels work together for different customers. Others want accurate answers to questions like ‘How did the email test campaign do?’, or ‘Do I get better returns from using Facebook at certain times of the year?’.

To answer these kinds of questions we have to aggregate up the results. If we take an email campaign, for example, it will have contributed to steps in many different customer journeys, some of which will have led to a successful outcome, and usually a much larger number that have not.

We only look at the opened emails within successful journeys and give a value to each of these steps, depending on the value given to the individual outcome, multiplied by the fraction of the overall journey process that the step contributed. So, mathematically, if the outcome is worth £50, and the step has contributed 30% of the journey, then that step is judged to be worth £15.

In this way every step in a successful journey gets a value and these can be summed up to give a value to the overall campaign. This example will help explain how fundamental the weighting approach is to the evaluation of the campaign results – give the step contribution just 10% of the journey and then it’s only worth £5.

Aggregation can provide answers to a number of different questions. It can sum up to an individual campaign, or to a channel, and that is usually for a time period, such as PPC in July. By looking at different time periods for the same channel we can understand the impact of seasonality on response. But we can also turn the data around and look at how different customer groups respond to different types of marketing.

We have already mentioned the differences in behaviour between new and existing customers, but what about, from the charity example above, whether different campaigns lead people to visit different tools in different parts of the website. A car vendor might, for example, be interested to know which media are better at driving test drives for different models.

channel and share of value diagram

Here is an example of an aggregated report, taking the channel view for a particular time period. Each channel has a share of the overall value based on the steps it contributed that led up to completed sales. It is interesting to see how the sales impacted are for every channel greater than the share of sales. So, if in this case the marketer decided to stop sending out catalogues, then 26,478 sales would have been impacted and many may not have happened.

 

4. Avoiding the ‘black box’ classification

There is a natural fear of basing decision making on results created by unknown algorithms that live inside impenetrable black boxes. Happily, this does not need to apply to MTA, as long as the technology can provide a table that shows the values given to every step in every customer journey. With the individual scores visible, the results can be challenged.

We sometimes set up manual scoring systems that give declining weightings to events that are further away in time to the ‘result’. This is a common-sense approach and useful for comparing the outcome with what the algorithms have come up with. For the more mathematically inclined, it is also important to be able to challenge how the algorithms themselves are programmed and to receive back a full explanation.

 

5. When does MTA not apply?

MTA works when there is a direct link between the person making the customer journey and the steps that take place. So, it is not applicable for TV, or outdoor advertising for instance. For these indirect channels econometrics is the right approach, and the good news is that we can now merge the outputs from the econometrics with the MTA results to provide a truly 360-degree view.

 

UniFida logo

UniFida is the trading name of Marketing Planning Services Ltd, a London based technology and data science company set up in 2014. Our overall aim is to help organisations build more customer value at less marketing cost.

Our technology focus has been to develop UniFida. Data science business comes both from existing users of UniFida, and from clients looking to us to solve their more complex data related marketing questions.

Marketing is changing at an explosive speed. Our ambition is to help our clients stay empowered and ahead in this challenging environment.