the Bento growth platform

Campaign Success: Email Attribution Models and the Value of the Halo Effect for Conversion

Thoughts by Aleksandar Miceski • Writer Bento

Are you tasked with the challenge of proving the effectiveness of an email marketing campaign?

Perhaps you're a seasoned marketer or new to the skill.

At the end of the day, we all want a campaign that pays off. But what’s the best way to relay the benefits of these efforts to the sales teams, to client relations colleagues, or to a client demanding a reliable metric?

Decision makers are always interested in the ROI from actions along the marketing funnel. If there was a way to specifically pinpoint all channels and the points at which conversions happen, this would have been a quite straightforward business. Unfortunately, things are not as simple because we are all dealing with unknowns.

So, we are left with the typical business improvement cycle: gather data and metrics, evaluate their merit, and decide if we should act on it.

Email, still to this day, remains the best way to reach customers in ecommerce, however, more detailed campaign attribution data is needed to inspire greater confidence in the chosen strategy. The quality of the data is used to eventually prove if the strategy works or there are better ways to spend your time.

Now, let’s go over some basics.

Email Attribution Basics

"Did that email actually make the contact convert? How can we be sure?"

Since a lot of a businesses communication takes place over email, logically, marketers would like to know which aspect of their email journey actually made them buy the product in order to attribute the sale to a particular broadcast, automation or sequence.

Of course, if we think holistically across the entire funnel (paid advertising, social media campaigns, affiliates, referrals, influencers, etc), it’s impossible to identify whether an email created the conversion or it was something else that happened before or after the email was sent.

But, as a marketer, we can do our best effort to try and capture as much data as we can to model it out.

Once we capture that data, we need to work out how to model it out.

Email attribution models correlate exposure to content with subsequent action in a bid to establish a relation between the two.

In essence, there are three elements that are taken into account: the type of action (conversion, open, or click), the exposure (or touch – first, last, or in between), and time scale/attribution window (days or weeks from the actual campaign).

First Touch Model

In this model (and in Bento), we attribute a conversion to the first instance at which a customer encountered your product or brand.

This could be an ad (tracked in Bento), a broadcast, a sequence, or a Workflow. In most cases, we would expect to see attribution be tied to an ad here and not so much emails as most new traffic is going to originate here.

This is great for identifying new customers and growth, however, you'll loose some insights about the direct impact of your marketing after the person has learned about you.

If you choose this model in Bento, we recommend you tighten the attribution window to only a day or so.

Last Touch Model

In this model (and in Bento), we attribute a conversion to the last instance at which a customer encountered your product or brand.

This will more accurately capture the impact of your marketing later on down the funnel, say in a Welcome Sequence, Post Purchase Sequence, or Browse Abandonment campaign.

We recommend that people open up their attribution window for these ones to 5 to 7 days.

Last Non-Direct Attribution Model

It’s practically a follow-up on the model above. Last non-direct attribution values the second-to-last channel at which a customer interacted before closing the deal. For example, if your client saw a social media post and then opened their email to convert, this model favors the social media post.

Linear Tracking Model

If the attribution is equally divided between all of the touchpoints, we are talking about a linear tracking model. It’s a very convenient way to calculate ROI for each individual channel; that would be the greatest advantage of going for this email attribution metric. On the flip side, not all contacts contribute to a conversion to the same degree and this model ignores that fact.

Bento currently does not support this model but may in the future.

Time Decay Tracking Model

The closer a contact is to the conversion itself, the hotter it is (and receives high attribution). If the first contact was two months ago on a certain channel, this contact will be attributed, but to a far lesser degree. Time decay tracking overstates recent activity and this can lead you to false conclusions (e.g. that the client converted because of the last email campaign).

Bento currently does not support this model but may in the future.

Positional Tracking Model

If the first and last touch get most of the credit and the rest of it is divided between the channels and contacts in the middle, you are following the positional tracking model. While this can frame the buyer's journey for you, it can also mislead you because all those contacts in between (email, social media, direct visits) are unjustly disregarded.

Bento currently does not support this model but may in the future.

Weighted Attribution Model

To give credit where it's due, you can go for a weighted email attribution model. It values contribution per channel to weigh all of the touchpoints in the process. The chief concern here is – who and how will weigh the value of each contact or channel.

Bento currently does not support this model but may in the future.

Custom Attribution Model

Aside from these standard models, some companies go for custom email attribution models to assess the success of their email marketing campaigns. It’s definitely an option provided you have the need and resources to develop such models. If you can’t analyze your objective with the standard models, you have no choice but to be creative. The disadvantage of custom attribution models is that you are choosing the touchpoints and channels, and this goes against the very purpose of having a model like this. It yields credible results only if you know your industry or customers well enough to not make any mistakes.

The Complexity of Current Digital Marketing Techniques

Back in the day, the conversion gospel used to be, “do what it takes to obtain the first name, the last name, and an email account of your potential customer”. Then, as subscribers, they are within your reach and you can cash in on that. Social media has changed the landscape and traditional conversion has become more complex. Although email is the prevalent way to convert, now the client base is exposed to multiple channels. It’s increasingly hard to qualify conversions or to determine the exact ROI on your costs associated with marketing across channels.

How do you take into account more than one channel?

Tools like Bento serve to provide data, but the decisions can’t be based solely on these results.

Customers can come in as a result of a referral, then visited the website, subscribed to your list, read your blog, saw a social media post, and only then converted. But, any of the touchpoints can change their order and who can quantify their unique significance? Some clients encounter a product through organic search based on search engine ranking because of properly optimized content. So, social proof in the form of a testimonial or referral from a friend or an expert is not needed. You get the point.

And whilst Bento will capture a lot of that information if you are properly tracking your links, there's a lot that isn't seen in the raw data or logs.

In addition to this lack of data, there is also misuse of data to make companies seem like they are generating way more revenue than they actually should be attributed too.

At Bento, we've migrated some clients over who, when properly tracked, saw a 50% DECREASE in tracked revenue from email - yet their delivery increased and total sales increased for their store. This only happens because companies are greedy at claiming they generated the sale.

The Case for Email Marketing

As much as it’s hard to exactly pinpoint the value of email marketing campaigns, no one denies they do have an effect. Subscribers engage with the messages and we have metrics like open rates and click-through rates to gauge the interaction. While we can’t establish a direct link between email open rates and conversions, some type of underlying relation definitely exists.

Email marketers use the term “halo effect” to refer to the effect of an email, even if there is no open or click. It’s sort of a residual effect that extends beyond immediate exposure to your email campaign. In psychology, the halo effect is used to explore the reason people ascribe favorable values to a positive experience (e.g. a beautiful face with a moral virtue). You can easily translate this to your product and contact with a client base.

We've seen at Bento that the most successful customers just always have the channel on. They're always sending emails, have their basic automations up, and just have good email hygeine. That's what you're shooting for.

Halo Effect: The Proof

Every business, be they small or big, can put this concept to the test. There is more than one way to approach the issue. If you want to find out the effect of your email campaign compared to contact on all of the other channels, you can stop sending emails. Of course, no one would like to miss out on potential revenue, so this ought to be done strategically. For example, you can analyze daily revenue across all channels and periodically withhold emails. This is not applicable to every industry, but if you find that daily revenue across all channels is higher on the days when you also send emails – there’s your proof.

If you want to put this to the test, in Bento you can create an A/B test which tags some users as qualified for marketing and others that aren't. If you divide your client base into two groups – a regular group (exposed to a campaign) and a control group (no longer receives emails) – you will have a better understanding of the halo effect as you'll see if your emails are having an impact.

Only bother with this if your list is above 100k.

Obviously, you can trim the control group down to 10% of your overall clients to minimize the loss of revenue during the test. Another important aspect of holdout testing is to make sure your sample is random. The actual method to achieve this depends on the specifics of your business model.

By withholding your email campaign from a group of subscribers, you can take note of their (buying) behavior. If the regular group brought more revenue (per customer), then the halo effect works.

These types of experiments are very similar to A/B testing of an on-boarding email sequence or cart abandonment emails. The difficulty lies in giving quantifiable credit to a specific element, channel, or touchpoint. Another example is extending limited time offers like discounts. How would you separate the customer that converted because of the discount from the customer that would convert anyway, but the discount offer happened to coincide with the moment of purchase?

The pitfall, in this sense, lies in falsely ascribing credit to an email campaign for transactions that would have happened regardless. Will customers convert if they are not exposed to a campaign?

We will not get a definitive answer on this question nor will we pin down the reason why all of them converted.

Maintain a Presence

Email is one of the cheapest and most effective ways to remain in your customers field of view. Consistent exposure will inevitably lead to some level of a halo effect, though you can’t put your finger on it. This is practically the most important point in email attribution. Particularly as it relates to go/no-go decisions on further investment in email marketing.

It’s a balancing act. If a marketer defines the halo effect in generous terms and puts too much money, time, and effort into a campaign, they might not get the expected ROI. On the other hand, if a marketer considers the halo effect as less relevant and refrains from investing in their email campaign, they might end up missing out on revenue as a result.

This is the conundrum that has to be put across to stakeholders. Investments in marketing are important to maintain presence, but it's almost impossible to perform an exact cost/benefit analysis.

Final Thoughts

At one point or another, a marketer will need to have a clear understanding about the outcome of an email marketing campaign. They either need it to improve their own performance or share metrics with relevant decision makers.

The ability to digitally track customer behavior is crucial for this type of insight. Such data can be crunched in standardized email attribution models, however, their shortcomings should also be taken into consideration. Going for the halo effect is typical and proven successful, but just because you can’t quantify it, it doesn’t mean you can overestimate it. So, proceed with caution when measuring the success of email marketing campaigns because underestimating its importance is also not as desirable.