Marketing attribution models masterclass: what they are, how to use them, and the models available to you

What is marketing attribution?

Marketing attribution is the analytical practice of understanding which tactics are contributing to conversions.

Tactics wise, marketers have a vast and fragmented array of platforms and media types at their disposal when looking to reach consumers and as a result, the consumer journey prior to conversion is becoming more complex and individual. This is in turn makes it even more vital to understand which channels and messages are contributing towards conversions and thus how credit should be apportioned.

Why does marketing attribution matter?

Marketing attribution brings into play different scenarios on how targeting and messaging influences a conversion. Armed with this data, marketers can make more informed decisions on how to assign marketing budgets; how to calculate ROI and how to measure effectiveness.

How to use attribution in your marketing strategy

To highlight the benefits of incorporating marketing attribution into your strategy, imagine the following scenario:

• Your company offers a promotion on a product
• To help drive sales, you advertise the promotion of that product via the brand's social channels
• After seeing the product or service on social channels, the target audience clicks through to the brands website
• The consumer adds the item to their shopping cart, unfortunately, they later abandon the purchase
• After cart abandonment, the company sends a follow up email reminding them to purchase that product before the promotion ends
• This prompts to user to eventually complete the purchase.
• Differing attribution models would tell you what point of contact is responsible for the eventuasl conversion.
• In this scenario, a first-touch attribution model would highlight the paid ad on social channels as being solely responsible for the conversion. Whereas a last-touch attribution model would pinpoint the cart abandonment email as being wholly responsible for the conversion.

Attribution in marketing - the various models

There are several standard marketing attribution models applicable to measuring performance.

In this section, we'll share which attribution models are typically considered. We'll also detail their benefits, drawbacks, and when you should consider using them.

Single-touch attribution models

Arguably, single-touch attribution models are overly simplistic, aiming to credit 100% of a conversion to a single visit. There are two primary variations of single-touch attribution models; the first-touch attribution model and the last-touch attribution model.

First-touch attribution

The first-touch attribution model credits 100% to the first engagement regardless of further interactions with your campaign.

Pros of the first-touch attribution model

• It's incredibly easy to set this model up.
• The amount of analysis and calculation involved in its application in minimal.
• Incredibly simple to track.

Cons of the first-touch attribution model

• Reductive and prone to errors.
• Emphasis on just the first component of the customer journey only details a small segment along the road to conversion.

Why/when should this model be used?

While reductive by nature, this model is primed for measuring which engagements at the top of the funnel are the most effective in driving conversions. It is useful for assessing how influential generic search is in a conversion journey as often generic search suffers at the hand of brand search in direct response campaigns.

Last-touch attribution (default)

Whilst the first-touch model credits the initial touchpoint along the road to making a sale, by comparison, the last-touch attribution model recognises the final touchpoint. Last-touch / last-click attribution has been the industry norm for decades. In software such as Google Analytics, you'll find the last (non-direct) touch attribution model also happens to be the default attribution model. Google Analytics favours last (non-direct) click because:

Google acknowledges that it’s hard for any person to arrive at a site directly without prior knowledge of its existence, if any other source has been involved in the journey, then Google Analytics automatically helps you work towards a truer picture of the most applicable last action touchpoint. In this rule, all direct traffic is ignored, and 100% of the credit for the sale goes to the last channel that the customer clicked through from before converting.

Pros of the last-touch attribution model

• Last-touch attribution models detail what prompted the consumer to convert, closest to the conversion action.
• Incredibly user-friendly with a minimal learning curve.
• The easiest to analyse and align with your web traffic.

Cons of the last-touch attribution model

• Incredibly reductive, every touchpoint in the lead up to conversion is ignored.
• This attribution model makes it impossible to analyse which aspects within your sales funnel have been successful.

Why/when should this model be used?

If you're new to online advertising or getting to grips with conversion rates, this is the model for you. It's the most commonly used, and gives a clear indication as to which channels / messages persuaded the customer to convert. Beware it can over-credit brand search and direct channels which carry lower costs and thus skew decision making, but if you account for this by setting different targets for different channels then you can't go far wrong.

Multi-touch attribution models

Unlike single-touch attribution models, multi-touch attribution models take an altogether more holistic approach to the various touchpoints leading to an eventual conversion. They generally tend to be more accurate than single-touch attribution models ever could be, however they are not as easy to comprehend or set-up. There are a few multi-touch attribution models you should be aware of.

Linear attribution

Also referred to as the equal attribution model. The linear attribution model apportions credit to each touchpoint consumers engage with when attributing how the conversion has been reached.

One complication within multi-touch attribution models is that pinpointing which touchpoints deserve the most credit can be challenging. Linear attribution provides a simple answer to this; all messages involved in the sales funnel have the same level of importance.

Pros of the linear attribution model

• The linear attribution model represents a significant improvement on single-touch attribution models, every message within the sales funnel is considered.
• The model is relatively easy to understand and can be used effectively as a means of comparison with other models.

Cons of the linear attribution model

• Linear attribution is not the best model to use if you're looking to optimise.
• The reality is, not all messages within the sales funnel are equally responsible for your conversions.

Why/when should this model be used?

The linear attribution model is generally best utilised when working within a long sales cycle. It gives a holistic representation of all touchpoints that lead to conversion whilst also papering over cracks left by the single-touch attribution models.

Time-decay attribution

The time-decay model maintains the more holistic approach typical of multi-touch attribution models. In this model, all touchpoints are once again credited but the difference is that those appearing closer to the point of conversion are giving more credit to closing the conversion.

Pros of the time decay attribution model

• This attribution model is ideal for those looking for ways to think differently about the touchpoints that led directly to a conversion, optimising based on recency.

Cons of the time decay attribution model

• Devaluing earlier touchpoints, particularly the first interaction, is not always wise.
• Circumstance depending, that first interaction may have been crucial in the consumer entering the sales funnel in the first place.

Why/when should this model be used?

This model of attribution could be beneficial if your business has a long sales funnel with many touchpoints. Using the time-decay model, you'll have a greater focus on the stages that actually led to conversion and appreciate the lag time from first interaction to conversion.

U-shaped / position based attribution

Distinctly different from linear attribution, the U-shaped attribution model acknowledges that some touchpoints will be more impactful than others within the sales funnel.

The u-shaped model credits the first and final touchpoints with 40% of the responsibility for conversion. The remaining 20% of the credit is split between all the touchpoints in between these two key messages.

Pros of the u-shaped attribution model

• Unlike single-touch attribution models, the U-shaped model prioritises the role that both the first and final engagements play within the sales funnel.
• Crediting the most significant points in the customer journey also gives you more scope to focus on areas for improvement in-between.

Cons of the u-shaped attribution model

• It can be too simplistic to highlight the first and final engagements as being the most important within the overall customer journey.
• There is an entire customer journey between these two engagements that is potentially de-valued.

Why/when should this model be used?

This model doesn't attribute much credit to engagements within either end of the customer journey. This means it's not well suited for businesses operating with a longer sales cycle or lead nurturing strategies.

This attribution model is ideally suited to when the first point of contact with prospective consumers is swiftly followed by the end conversion with few interactions in between.

Data-driven / algorithmic attribution

This is the most accurate, data-driven method of crediting touchpoints in the user journey.

The model creates an attribution model following a process of machine learning of every touchpoint. The data derived from this analysis is what is used to compose the resultant model. The outcome of this technique will likely be unique for each business that adopts it.

Pros of the algorithmic attribution model

• This attribution model is useful if you happen to have a very large or complex customer journey.
• Credit is given to the touchpoints that actually matter to the business using this model.
• Provides the most accurate data derived from your consumer journey.

Cons of the algorithmic attribution model

• A complex process, calculations will require a data analyst or the use of advanced and powerful software that may have high price points.

Why/when should this model be used?

If you operate with a short, simple customer journey, there are more efficient attribution models you could use. However, if you have a much longer and more complex customer journey, the algorithmic attribution model is ideal. This is particularly true if your reporting includes marketing qualified leads (MQL) and sales qualified leads (SQL).

Top 3 attribution modelling mistakes

Digital bias

This mistake is particularly applicable if your business is in a position where it has both digital and offline marketing activity and a bricks-and-mortar presence.

Digital bias occurs as attribution models don't factor in that there can often be a relationship between digital and offline advertising touchpoints as well as sales at brick-and-mortar locations.

If your business is in this position, you need to be able to make optimisation decisions with an appreciation of this more complex ecosystem.

Missing message signal

It's easy to say that a touchpoint is ineffective because it's not moving users further along the sales funnel.

Marketers may misattribute this as a result of the wrong creative copy. This may not be the case.

It may be that the audience targeting related to the touchpoint is all wrong. The copy could prove more effective in front of a smaller, more targeted audience.

Don't make the mistake of dismissing great creative and copy because your targeting is wrong.

Market bias

Attribution modelling could never account for the off-chance that the converting consumer was already in the market for a product/service of your description at the time of seeing your touchpoint.

This is simply a case of right-place-right-time, though the touchpoint will get the attribution for converting the customer.

Some allowance for these sort of incidents should be considered in your reporting.

If you're keen to learn more, further reading is available courtesy of our Strategy Director, Kherrin, whom discusses her views on marketing attribution topics.

Want to learn more?

Read: Marketing attribution, what is it and what model should i use?

Including:

The importance of weighting your data

Deciding on a look-back window

Why you should do lots of testing before deciding on a model for your business