One of the most common questions we are asked is “How do I make sure I am maximizing the value I get from online marketing?”.
Given how most companies spend at least 6 figures a year online marketing, this question is fundamental to the overall performance of the whole company, not just to the performance of the marketing team. Without a framework to provide feedback and focus for your online marketing efforts, all those marketing dollars could be going to waste.
In this article, we will discuss two frameworks that can be used to devise your online marketing strategy and to interpret the daily feedback that online marketing data can provide. The frameworks focus on optimising your marketing performance towards return on ad spend or customer lifetime value.
We’ll define the components of each framework and outline how each can be used to help you allocate budget, assess performance and drive online marketing expansion. Finally, we’ll provide some clarity over how to decide which method best suits your business.
In principle, the frameworks are very simple. They both involve generating actions based on how one metric looks compared to a target value for that metric. Where it gets difficult is in calculating the metrics themselves as well as the targets.
There are various pitfalls that can affect their accuracy and interpreting the results and generating actions is not always straightforward. We hope this article will address these difficulties.
Before continuing it is important to note that the frameworks do not attempt to answer all the questions that online marketing will raise. They cannot answer questions like “How can I increase revenue from my paid search activity?” or “How can I improve the conversion rate of my paid social activity?”.
Instead they can help marketers choose what performance to expand, reduce or cut completely. If a given activity meets the criteria of the framework, then you should continue that activity until it does not meet the criteria.
The frameworks are intended to provide a focus for your marketing efforts. The goal is to make it very easy to decide whether a given marketing activity is worth pursuing. In a world with limited resources, focus and prioritisation of the most important tasks is what sets apart the best from the worst performing companies. That is the power of these frameworks – clarity of thought and action.
Before diving into the frameworks, we would like to address what Conjura considers to be non-negotiable traits when approaching online marketing; consistency of application and making data-driven decisions.
Success in online marketing is ultimately a data problem. Marketing online generates a lot of data; in the ad platforms themselves, in tools like Google Analytics and transactional and customer data. Making sense of it is key to making effective decisions.
Once plans have been made, a framework chosen and a strategy devised, it must be consistently applied to all channels and campaigns. If it is not applied consistently, it will not be possible to determine whether it is working. There will be too many caveats and confounding factors that make it difficult to have confidence in any inferences made.
Both methods outlined below focus on defining targets and objectives and ruthlessly cutting any activity that does not meet these targets. Without proper planning, data and consistently sticking to a strategy, these frameworks will not be effective.
Now let’s turn to the frameworks themselves.
Return on Ad Spend, or ROAS, is a calculation of the amount of revenue you receive in return for spending £1 on marketing.ROAS = [Revenue] / [Online Marketing Spend] A high ROAS indicates that a given marketing channel generates a lot of revenue for a relatively low marketing spend. A ROAS of less than 1 indicates that a company spends more than is generated in revenue.
Calculating ROAS appears easy at first glance. But there are several factors that must be considered before it is possible to accurately calculate the metric; the attribution model used, and the quality of tracking implemented on your marketing activity.
Marketing Tracking refers to the practice of adding a snippet of text to the end of a URL to highlight what campaign and channel a click or visitor came from. The most common format used for tracking your marketing activity is that used by Google Analytics – UTM tags [link to UTM tags page].
Adding UTM tags to your URLs enables Google Analytics to categorize traffic into channels and campaigns. This then allows Google Analytics to perform effective attribution and enables users to properly assess the performance of their marketing activity. If you do not effectively track your campaigns, you cannot perform any kind of performance assessment and you are operating blindly.
Having an extensive background in marketing analytics, tracking is very close to Conjura’s heart. There are two characteristics that signal effective marketing tracking; consistency and granularity.
Consistency – all campaigns must be tagged, and all tags must follow the same naming convention and tag format. It is imperative that time is spent devising a comprehensive and future proof naming convention.
Granularity – tracking should include as many details about the campaign as possible. The more information included, the easier it is to determine the objective, target demographic, location etc. of the ad and the easier it is to analyse.
Attribution is a hugely complex topic that has taken up thousands and thousands of words. As such, we are not going to delve deeply into attribution in this article. Instead, we’ll direct you to Neil Patel’s article on attribution. It gives an overview of various models and how to decide which one is best for your business.
Conjura suggests approaching attribution pragmatically – simply choose a model that works for you and stick to it. The simpler the model, the easier it is to understand. Sometimes, this is the best approach.
Only by maintaining a consistent model can you assess performance over time and across channels. Additionally, if everybody in the business can understand the model, and its limitations or idiosyncrasies, then everyone can have an informed and productive discussion about what is and is not working in your marketing activity.
Before using Return on Ad Spend to drive decision making, you must develop a ROAS target. Ideally you will have specific targets for each channel and even at the campaign level.
Defining Target ROAS When defining target ROAS, it is important to account for the attribution model being used, especially if you are setting targets at the channel level.
For example, imagine an Ecommerce company using Facebook campaigns to build awareness and using Google Shopping to convert leads into customers. Brand awareness activity drives leads into the top of the conversion funnel while Google Shopping converts leads at the bottom of the funnel.
If this company is using a last click model, it will assign all transactions to Google Shopping ads, likely leading to a healthy ROAS figure. Meanwhile, the impact of the Facebook’s brand awareness ads is relatively ignored, resulting in a poorer ROAS figure. In contrast, a first click model will assign more transactions to Facebook and less to Google Ads, affecting the ROAS on each channel. When defining a target ROAS, it is very important to adjust it for the attribution model you are using.
Additionally, when defining ROAS, it is important to ensure that the target chosen results in a profitable outcome. This means choosing a ROAS that covers costs of doing business such as product and other operating costs.
Finally, being able to use historical data or, even better, industry benchmarks for ROAS targets is very beneficial. If you know what “good” looks like, you should strive to emulate and outperform it.
Using the ROAS Framework It is understandable to think about ROAS as “the higher, the better”. This is a relatively limited view, however and can result in missed opportunity. To explain this further, we’ve outlined three scenarios below.
Imagine a company sets a target ROAS of 5 for its paid search marketing activity. The company must assess the actual ROAS achieved compared to this target. Consider the following scenarios for the actual ROAS achieved:
This is an easy one – the marketing activity has not been successful. The company is not generating the level of return required given the amount of money they are spending. In this scenario, breaking the paid search channel down into campaigns and reassessing ROAS is a good starting point.
Is there a group of campaigns where ROAS achieved is above target for example? What makes these campaigns better performing than the rest? Is it feasible to reallocate some budget from underperforming campaigns into better performing campaigns?
There are a whole range of causes of underperformance in this scenario such as keywords being very competitive and expensive, the marketing channels in use driving poor quality traffic or poor website conversion rate.
Again, interpreting this is easy – the activity is successful and on target. The challenge when you are meeting targets is scaling your activity. The key question is whether you can increase spend and expand activity while maintaining the same level of ROAS. This is the sign of a great online marketing team.
Of utmost importance in this scenario is experimentation. This entails developing new ideas and allocating a small amount of budget to determining whether these ideas show promise. If they do, allocate more budget. Then iterate repeatedly while always monitoring how actual ROAS compares to your target.
This is probably the most interesting scenario, but also the one that most of Conjura’s customers find difficult to accept. For Conjura, in this scenario the marketing activity has been unsuccessful, despite actual ROAS being greater than target.
While the performance appears good at first glance, the company has left additional revenue on the table – it has not met its full potential. The laws of diminishing returns apply to ROAS; ROAS tends to fall as spend increases. This scenario shows that the company has room to scale spend and expand activity. The company should keep spending until ROAS reaches target.
The above scenarios show how important having a target is when using ROAS. Without that target, the company will not know how to interpret the figures and will struggle to implement a successful strategy.
Customer lifetime value (LTV) is the value of a customer to your business over their “lifetime”. The definition of a “lifetime” can vary but is usually somewhere between 12 – 48 months.
There is no right or wrong answer in what you consider the most appropriate lifetime for your business. It is more important to be consistent with your choice so you can monitor whether LTV is changing over time and make efforts to identify why.
Calculating LTV is straightforward, simply calculate the total value of a customer over their lifetime. Value in this instance could be revenue, but it is very important to account for discounts, returns and shipping. Even better is to define value as the margin generated by each customer.
Often, data availability and difficulty in calculating margin at a customer level makes generating LTV time consuming. What’s most important is that a consistent definition is used, even if it is based simply on revenue.
When using customer lifetime value to assess online marketing performance, Conjura always recommends considering cost per acquisition at the same time. Specifically, the relationship between these metrics will provide the insights and actions required to maximize performance under this framework.
It is necessary to first define what this relationship should look like. This involves developing your investment principles. These are the rules and requirements by which you invest your marketing budget. If your activity meets your investment principles, you are succeeding. If it does not, you are failing. We will discuss investment principles later in this article. First, we will discuss cost per acquisition.
Cost per Acquisition, or CPA, is one of the most common metrics used to judge your marketing performance. Its calculation is straightforward:
CPA = [Online Marketing Spend] / [# Customers Acquired]
Simply put, it tells you how much it costs to acquire one new customer.
Like calculating ROAS, calculating CPA is dependent on the attribution model is use. In this case, the attribution model determines which transactions and new customers were attributed to which channel, so it has a huge impact on CPA per channel. Additionally, how companies identify new customers from existing ones is very important.
The only way to know whether a transaction was made by a new customer is to deduplicate your customer base. This means assigning an ID to each customer and then linking every transaction made by that customer to the ID. This allows you to determine whether a transaction was made by a customer that exists in your database already. If they do not, they are a new customer. If they do, they are an existing customer.
There are many methods for customer deduplication including algorithmic fuzzy matching and points-based systems. If you’re limited for resources, a good starting point is to use the customer’s email address. This won’t be perfect, but Conjura has found it gets customers 90% of the way there.
As explained above, investment principles are the rules by which you allocate budget across channels and campaigns. They define how the relationship between LTV and CPA should look if your activity is successful.
Investment principles determine the desired payback period for your marketing spend. To put it another way, investment principles establish when a company wishes to break even on its marketing spend. For example, if a company sets a 12-month investment principle, the marketing budget should be invested in such a way that if the company spends £10,000 on marketing in January, it expects to have returned £10,000 by the following January. Anything after the following January, will be profit.
What is the correct investment principle? There is no one size fits all answer to this question. Instead it is a decision a company will make based on how aggressive it wishes to be, it’s risk appetite, cash reserves and the industry it operates in. For example, a 36-month investment principle is aggressive. It means that the company is willing to make a loss on every customer for the first 36 months of those customers’ lifetimes. In contrast, a 12-month investment principle means that the company wishes to break even on a customer after 12 months, which is much less aggressive.
In new industries, where there is not a clear market leader, companies may wish to be more aggressive in order to grow marketing share. They are willing to carry the loss for longer. This will likely lead to a longer investment principle. In more established industries, investment principles might be shorter as companies wish to maximise their value. A longer investment principle is also likely to be chosen by a company with high cash reserves as these companies can absorb the initial loss on a new customer for longer.
Whatever investment principle is chosen, it is extremely important that the company then calculates the corresponding LTV e.g. if choosing a 36-month investment principle, the company must calculate what the 36-month LTV of their customers is.
Doing this enables the company to set target CPAs for their marketing activity. In order to determine how well you are performing, you should compare actual CPA with your target CPA. By using LTV to set a target CPA, and then monitoring the actual CPA, it is possible to regularly determine whether your marketing activity is profitable and successful.
Ideally, investment principles and target CPAs will be applied per channel and, if possible, at the campaign level.
This is best explained by way of an example. Imagine a company wishes to break even on a customer 24 months after that customer is acquired. The 24-month LTV is £200. Therefore, the target CPA for this company is £200. The scenarios below outline how to use and interpret the relationship between actual CPA, target CPA and LTV.
In this scenario, the company is spending more to acquire a customer than that customer is worth after 24 months. The company is not successful considering its investment principle. In this scenario, it is advisable to investigate those channels and campaigns at the root cause of this problem. As with the ROAS framework, it is advisable to dive into the campaign level data and understand the relationship between actual and target CPA. Doing this enables marketers to identify and focus on the worst performing campaigns and make relevant changes.
Is the issue caused by poor quality traffic, poor retention, poor conversion rates for example? Regardless of cause, spend should be reduced and reallocated to better performing campaigns, at least temporarily until solutions can be implemented and tested.
In this scenario the company is meeting its investment principles. It is acquiring customers at a cost that will allow it to break even after 24 months. Like the ROAS example, the challenge is in scaling spend and activity while maintaining this performance.
The company has acquired customers for less than the target CPA, meaning that the company will break even on these customers sooner than 24 months. This seems like a positive result; however, this means that the company is not reaching its potential – it could have spent more on marketing, acquired more customers and remained within the parameters set by the investment principle.
CPA tends to have a positive relationship with spend – as spend increases, so does CPA. Therefore, the company should increase spend and activity until the actual CPA rises to meet the target. At this point, the company will have reached its limit on this channel and it may be time to start investigating other promising channels
The great thing about these metrics is that most companies can and should focus on only one. If interpreted and used correctly, they will instill discipline in spending and optimization decisions and provide a clear set of actions to take in each scenario.
So, how do you decide whether ROAS or LTV is the best framework for you?
Lifetime value and ROAS are both measures of profitability. The main difference between them is their time horizon. ROAS covers profitability on a single transaction i.e. the company spends £10 to generate a transaction of £20 resulting in a ROAS of 2. In contrast, LTV covers a longer time horizon i.e. the length of a customer’s “lifetime”. As discussed previously, this time horizon for LTV can vary, but is generally at least 12 months.
The choice of which one is more appropriate is dependent on a key factor – purchase frequency.
Simply put, a company that receives a relatively high purchase frequency e.g. 3 purchases per year over multiple years – will be more interested in assessing marketing performance based on the LTV framework. Customers are more likely to stick around for more than one purchase and this should be considered when assessing marketing performance.
Using ROAS in this case may limit what a company believes they can spend, especially if the item price is relatively low, which it likely is if customers have a higher purchase frequency. This limits their ability to compete for new customers and negatively affects overall performance.
In contrast, companies with a low purchase frequency – mattress companies for example – do not expect many future purchases from customers. Therefore, the lifetime of a customer is usually one purchase. In this instance, ROAS and LTV will be very similar, especially if ROAS accounts for margin. Companies in this position should save themselves the trouble of calculating LTV and use ROAS.
There are many characteristics that can impact purchase frequency such as the opportunity to cross sell and how long the product remains useful to the customer, but these are best covered in a separate article.
The return on ad spend and customer lifetime value frameworks for assessing marketing performance are very straightforward to understand and implement. That’s what makes them very powerful. It is much easier for a marketing team to get behind and implement a simple framework. That said, the frameworks are not without difficulties and pitfalls, especially when combining the data sets required and calculating the metrics involved.
In order to get the most from whichever framework you choose, it is imperative that every decision made is made with your framework in mind. If something does not fit, stop doing it. If something fits, continue or try to scale that activity to maximise performance.