Interpreting frequency & double counting conversions

 

Frequency.

Frequency is the number of times a particular event occurs.

As a metric, it provides insight into the distribution patterns we can understand about the data set allowing for a better gauge into the proportion of instances that a communication was viewed or interacted with.

There are many instances where understanding frequency can better support marketers to assess, interpret and make decisions from marketing data. All of these instances start with defining the time period in which we are seeking to understand the relative impact to a particular cohort of customers, then understanding the average number of instances in which a unique communication was interacted with by a consumer.

 

Frequency can be assessed in many contexts. 

At its core, assessing frequency allows marketers to better understand and make optimisation decisions on the following three key areas. 

1. The potential limitations on reach, awareness and impact of key campaigns.

2. The potential efficacy of a key message, in terms of its resonance, depth of comprehension, and influence on brand association or recall. This may play a pivotal role in driving effective creative strategies.

3. The phenomenon of audience fatigue, which may arise when a message loses its appeal and becomes perceived as intrusive or excessive. It is at this time the message has the potential to impede effective communication and engagement.

To draw these conclusions however, frequency cannot be interpreted alone. It is important still to maintain a pulse on the outcomes that are being derived from the activity to draw the most accurate hypothesis for further testing. 

In the context of paid media, cost can also be a data point that can indicate the level of efficiency a channel is driving. Cost should therefore be used as a key data point alongside frequency to understand whether the channel is the best choice within the mix of activities to drive the desired outcome. 

The following matrix provides a guide for how to interpret changes in frequency when viewed in the context of the outcomes the activity is responsible for driving impact to.

 

Frequency interpretation matrix

Beyond the specific metric of “frequency” there are similar metrics that provide similar insights in different contexts. Path length provides an indication for the number of website pages that are viewed within the context of a website session prior to engaging in a key converting action. Touchpoints provide a similar metric in the context of customer journey analysis beyond digital metrics and inclusive of a variety of online and offline metrics. 

Regardless of the specific metric and context, monitoring frequency alongside results offers marketers an indicator to better understand how communications are effective, or not effective at influencing the rate of relevance and resonance in their communications plans. 

When used alone, it offers very little insight but when combined with understanding the context of the touchpoints and costs associated to deliver them, frequency can guide marketers to understand how to best structure the most efficient channel and messaging strategy to achieve their goals.

 

Week in review.

 

I was reminded again this week of the art of helping marketers understand the art of double vs. single counting conversions. 

I get it. It feels just a little bit “wrong” to know your walled-garden channel reports are double counting conversions and revenue when you know you’re on the hook to report back a single figure.

For the many marketers who are yet to have their own operating Markov Chain digital attribution model, using the “out of the box” last click reports are the simplest way to de-dupe and offer up a single measure of program effectiveness.

Of course, this leaves all of the top and middle of funnel channel owners seething as the performance channels like search and email take the credit for the conversion “touchdown”.

My advice is to lean into it feeling “just a little bit wrong” but be extremely clear on the narrative. The art is in linking the activity and most importantly, its primary purpose to a visualisation that helps understand alongside it, what it supported as a secondary objective. 

Then, in the same view, regardless of the attribution model you adopt (last touch, DDA, custom MTA, MMM), provide one, de-duped view to allocate channel performance. 

Here’s a quick wireframe I scribbled down this week to illustrate what I mean. In an ideal world, this is combined with a media experimentation program to better understand the relationships between channels in the omni-channel mix.

 

I know what you’re thinking, “But how do I get all of this data in one place to visualise it!” Well that’s where my team come in. Talk to us.

 

As one word, the word “attribution” feels complex but that’s exactly the problem. 

When we break down one question into the following three questions, we can gain the clarity and insight we are looking for:

 

• • What did the channel drive as aligned to its primary purpose?

• • What did the channel just-so-happen to drive as secondary measures of impact?

• • When we measure the effectiveness of the program as a whole, what was the final net number?

That final question can lead down a path of “Which model is the right model for us?” which I look forward to tackling in a future newsletter. Maybe next week!