The history of Return On Investment (ROI) has an unexpected, long and explosive history. Originally known as the ‘DuPont Analysis’, the organisation began using its uniquely created formula in the 1920s. It was invented by Donaldson Brown (a DuPont explosives salesman) in 1912, as an efficiency report with a view to improve performance in the company.The marketing ROI formula we use today hasn’t changed much, still containing three distinct elements as part of its calculation in its purest form:
The theory used comparisons with like-minded industries or organisations as follows:
ROI is very much the word on every marketers lips in current climes and it is set to stay that way. There are pros and cons to this too, with some theories stating that it can over-simplify, create limitations and not give a strong enough insight into real, actual performance.
Throughout the 60s and 70s, the use of ROI models took a strong hold on the job of most marketers, particularly in the competitive world of TV advertising, the undisputed king of marketing during these decades. During this time, ‘share of voice’ was the key measure of choice, with most large brands such as Coca Cola, Unilever and Ford Motor Cars.
Moving into the 80s and 90s large retailers, pharmaceutical companies and motor manufacturers began recruiting marketers with a background in packaged goods, bringing a whole new set of skills to the table as the consumer audience started becoming more sophisticated. Bring on the advent of the ‘always-on’ culture driven by the consumer’s addiction to information and advertising has entered a new dimension of media proliferation, multi-tasking and the ‘brand of me’.
The questions most frequently asked by marketers are “how did this campaign perform, how much did it contribute to revenue and therefore, what was our return on investment?”. But tracking the ROI of any marketing campaign is hard because naturally there are so many variables which lead to a conversion.
Knowing what tactics have been employed on each channel and the variables that lead to the improvement of these tactics is important, but there are a number of other considerations that should be factored in:
Knowing when to measure: Campaigns will have a lasting effect, budgets invested into marketing today may have any number of impacts at various points in the future. Exhibitioning at an event may convert immediate customers, but also may be just one of many brand touchpoints for a customer that converts in 3 years time.
Multiple touchpoints: As alluded to above, the awareness and consideration phases of a customer’s buying journey could be long and winding. You’ve heard the thought process that at least seven touches are required to convert a cold lead into a customer. Forget the number, but hold onto the truth that it takes multiple, memorable touches to convert a customer. However, which touchpoint is most important in creating a conversion?
Multiple influences at each stage of the buying journey: Some purchases are large and complex and require multiple stakeholders, some purchases are influenced by peer recommendations, some are influenced by education from thought-leadership. Each buying journey is different and individuals will react and respond differently dependent on these influences.
Uncontrollable variables: Customers might not want to venture out to a car show in the rain, some might already have plans the weekend you decide to push your ad budget further. Others dislike the colour of the image you have used, some don’t want to be hassled by sales reps… As a marketer, these preferences are all out of your control. Can you claim your marketing strategy was successful and delivered positive returns due to macro-economic trends?
ROI calculation gives you numerous advantages. The first and most obvious? Knowing your investment’s impact on your business. If you determine you’re wasting money on an expense, it’s a no-brainer that something needs to change. Many types of ROI can help you make important business decisions, including but not limited to:
Strategy: Did a specific strategy or tactic help lead to a conversion? Understanding whether a particular strategy drive results will give you a baseline for utilising marketing to boost the revenue and overall profitability of your business.
Purchasing a new tools or software: Adding new tools, equipment and products to your business can be a step in the right direction, but they must be purchased wisely. Calculating the ROI on an equipment and software purchase allows you to gauge how valuable your new tool is and what types of equipment to invest in, in the future.
Hiring new employees: Is your new employee increasing or decreasing your business’ profitability? Tracking the return on investment of your team will help you better understand which kinds of people to hire (or fire).
Adding a new department: Just like hiring a new employee, adding a new department or function to your business can be a smart move if it helps increase profits. You don’t want to play a guessing game here — calculate return on investment to determine the profitability of your departments and identify opportunities for expansion.
ROI stands for Return On Investment. ROI is used to measure the monetary gain or loss generated on an investment relative to the amount of money invested. ROI is usually expressed as a percentage and can be calculated by dividing the NET PROFIT by the COST OF INVESTMENT and multiplying that number by 100.
ROMI stands for Return On Marketing Investment. ROMI is used as a metric to measure the overall effectiveness of a marketing campaign. ROMI then allows marketers to make better decisions in regards to how much and where to invest capital for the best outcome.
ROAS stands for Return On Ad Spend. This metric helps marketers to evaluate campaigns effectively. ROAS can be calculated by ad SPEND DIVIDED by REVENUE – this will tell you how much revenue is earned for every pound that has been spent.
CAC stands for Customer Acquisition Cost. A businesses CAC is the total sales and marketing cost required to earn a new customer over a specific time frame.
CAC is calculated by: SALES AND MARKETING COSTS ÷ NUMBER OF CUSTOMERS ACQUIRED OVER A SPECIFIC TIME FRAME
The LTV:CAC ratio compares the average Cost of Acquiring a Customer to the average Lifetime Value of that customer.
Cost Per Acquisition (CPA) is another word for cost per action and is used interchangeably with this term. CPA measures the advertiser’s per conversion cost from start to finish, from the inclusion to the search engine results to creating interesting landing pages that grab the attention of the visitor. This means cost per acquisition measures how much it costs in advertising to convert one person from a visitor to a client for the company.
CPL stands for Cost Per Lead. This is an online advertising pricing model when marketers can understand the total cost required for each lead required. The metric is : Total cost ÷ Leads = CPL
LVR stands for Lead Velocity Rate. This metric quantifies a companies growth in terms of their qualified leads. The metric to LVR is:
Qualified Leads last month ÷ Qualified Leads last month X 100 = LVR
CPC stands for Cost Per Click. This refers to the amount you pay for each click on a PPC ad. The metric to find CPC is:
Your Total Cost ÷ Your Total Number of Clicks = average CPC
CPO stands for Cost Per Order. This refers to the marketing costs that are required to generate a certain order on your website. CPO can be worked out via the following:
Action Costs ÷ Number Of Reactions
CVR stands for Conversion Rate/Ratio. This refers to the percentage of visitors to your website that complete a desired goal (e.g fill out a form, make contact, reply to an add etc)
CVR can be worked out by:
Number of users converted ÷ number of users who clicked on the ad X 100 = CVR
CTR stands for Click-Through Rate. This metric measures the number of clicks advertisers receive on their ads per number of impressions.
CTR = Total Clicks ÷ Total Impressions X 100
NPS stands for Net Promoter Score. This metric is used for assessing customer loyalty. The NPS is the difference between the percentage of brand promoters and brand detractors. The difference is expressed as a whole number e.g 25% Promoters, 20% Detractors and 55% Passives = +5 NPS.
Certain methods can aid the measurement of ROI in marketing. There are five methods which are most commonly used:
1. Single Touch Attribution
Single Touch attribution allocates all credit to just one click. This can either be the first click on the conversion path, or the last. For example, your business might have several ads on Facebook, however Single Touch attribution will credit the specific ad the user interacted with conversion. Examples of Google Analytics Single Touch Attribution models are the First and Last interaction models.
Single Touch attribution is easy to implement and has a low cost, it is good for lead generation and the investment per lead metric is simple to understand. However, insights generated through Single Touch attributions lack depth; these insights do not account for the impact of lead nurturing.
2. Single Touch Attribution With Revenue Lifecycle Projections
Once you add Revenue Lifecycle Projections to a Single Touch attribution then you can start to understand the long term impacts of your marketing methods. This process involves making assumptions based on historical data, thus you have the ability to begin a campaign with a confident degree of accuracy as to what the ROI might be over a period of time.
Advantages of using this method include that it places emphasis on revenue rather than lead generation. It also makes use of historical data, which strengthens the likelihood that strategies will lead to success. However, this method does not account for a multiple touches and it could be that historical data your company has to hand is not relevant for the campaign.
3. Multi Touch Attribution
Multi Touch attribution refers to crediting more than one click on the conversion path for the conversion. For example, if Linda searches online for “jackets” and is directed to jacket.com, then she clicks on a social media ad which directs her the same site and finally Linda receives a jacket.com marketing email from which she converts, the Multi Touch attribution tool will credit all channels which led to Linda’s conversion. However, this does not guarantee that all channels receive equal credit for a conversion. Different Multi Touch attribution models on Google Analytics weigh the credit per conversion differently, for example the Position Based model which credits the first and last touched channels more in comparison to the channels in the middle of the conversion path. The other Multi Touch attribution models on Google Analytics are the Linear model and the Time Decay model – all explained further below.
The Multi Touch method is beneficial for long running campaigns as it acknowledges all touch points on the conversion path and allows for you to place credit at different points, rather than associating all credit to one point in the conversion path. Note, you must consider which touchpoints to apply the most or least credit to, for example if all touchpoints are credited equally, then one touchpoint which was clicked by accident receives the same crediting as the touchpoint which made the conversion.
4. Test and Control Groups
Test and control groups allow you to monitor how successful your marketing campaigns are; it is important to test your campaigns before they exhaust a full marketing budget. When implementing tests, you can research tactics, messages, media frequency and spend levels.
There are various tests you can run on your target audience. For example, start with two groups, both made up of your target audience demographics: the control and the variable. Take one group and apply your campaign to them (the variable), but ensure the other group does not witness the campaign (the control). From this point observe the buyer behaviour between groups, if those subjected to the campaign are more likely to buy then your campaign is successful.
Tests can be analytical and cheap to employ, reserving some of your marketing budget for trial campaigns is one of the best ways to utilise results. This also helps plan future campaigns, as you quickly discover what works for your brand and what yields successful results. However, be careful with which type of test you run; they can be expensive and time consuming. Their effectiveness is also dependent on variables like the time of year.
5. Full Market Mix Modelling
Market Mix Modelling reveals how revenue is impacted by multiple touchpoints through the use of statistical analysis. From here it generates an equation as to the overall impact of the marketing on sales and to what extent each point on the conversion path led to the achievement of the goal.
Overall, this is a statistically accurate model; it provides an understanding as to the effectiveness and efficiency of each point along the path to conversion. However, it requires a lot of data, some of which is hard to attain. Advanced statistical skills are required to implement this model and the use of formula to drive short term impacts can interfere with long term brand building.
Why do we use Google Analytics attribution? In simple terms, because it helps us work out which channels are working the hardest to create conversions. The ultimate goal of each touchpoint is to convert a browser into a purchaser, however without the conversion path displayed to us via Google Analytics, we would not be able to determine which channel attributes to the sales made by your company. The attribution model shows us which channels you should value the most in relation to the number of conversions made.
For example, The First Interaction attribution model is able to tell you the click on which channel initiated the consumer’s path to conversion. The attribution model for the First Interaction will credit 100% of the sale to the first touchpoints. On the other hand, the Last Interaction attribution model will highlight the opposite; it will 100% credit the final touchpoints which led to the conversion. Use attribution models which reflect your business goals. Every enterprise will value different points of the conversion path, therefore Google Analytics provides a number of default attribution models, alongside customisable models.
Google Analytics attribution is a great tool for any business for it shows where money is best being spent on digital marketing. The superior platforms thus create the highest ROI for your business, and you know to invest more of your budget into these channels. Google Analytics attribution will also display the channels which are bringing in the least ROI. From here you know to experiment with the type of marketing on these digital platforms or consider removing investment in advertising on these channels altogether.
Default attribution models
The default attribution models are those which Google Analytics automatically supply for you; however, you can customise and add your own. Here is a foundational run-through of how each of the default attribution models work, what they do and how they might be of use to your enterprise:
The Last Interaction Model
As previously discussed, the Last Interaction model attributes 100% of the conversion value to the final channel which the customer interacted with before they completed their transaction. This is useful information if your campaigns are designed to attract a click at the moment of purchase, or if your business is transactional and therefore does not involve a consideration period before the consumer converts. The Last Interaction Model is an example of a Single Source attribution model as 100% of the conversion credit is given to one click on the conversation path.
The First Interaction Model
The First Interaction model attributes a conversion to the first channel the consumer interacts with previous to their conversion. This is a useful tool if you run advertisements to create initial awareness of your enterprise. For example, if your brand is relatively unknown then you may invest in channels which first expose your company to a number of customers. The First Interaction model is a Single Source attribution model.
The Position-Based Model
If your enterprise is looking for a combination of the Last Interaction and the First Interaction models, then you will most likely use the Position Based model. Basically, you split the credit between the first and final interaction channels. A common practise is to assign 40% of credit to each of the first and last channels, then the remaining 20% is given to the channels in the middle. This is most useful model to implement when you value touchpoints which initially interest a customer and the final touchpoint which actually resulted in the conversion. The Position Based attribution model is a Multi Touch attribution for more than one channel is being credited for the conversion.
The Linear Model
The Linear model provides equal attribution to every channel on the conversion path. The Linear model is useful if your campaign is intended to maintain a relationship with the consumer throughout the process leading up to the final conversion. If this is the case for your enterprise, then you value each touchpoint equally as every interaction with your brand led to the final conversion. The Linear attribution model is an example of a Multi Touch attribution, as credit is given to every channel on the conversion path.
The Time Decay Model
The Time Decay model attributes a channel depending on where they stand in relation to the timing of the conversion. For example, a channel towards the end of the conversion path will be credited more heavily in comparison to a channel which was interacted with only initially. The Time Decay model has a default “half-life” of 7 days, which means that Google Analytics attributes half the amount of credit to a touchpoint which is clicked a week prior to the conversion when in comparison to the amount of credit given to a touchpoint what is interacted with on the day of the conversion.
For example, “Credit 10” is given to a channel which is interacted with on the 14th of September when the conversion takes place on the 21st of the same month. Channels interacted with on 21st of September receive “Credit 20”. It follows to say that interactions which occurred 2 weeks prior to the conversion receive one quarter credit of the conversion, so, to continue the example, interactions on the 7th of September would be worth “Credit 5”.
The use of the Time Delay model is useful if you run one-day or two-day campaigns as you may wish to attribute higher credit to the interaction occurring on those days. It is likely that if this is the case then clicks which occurred one or two weeks previous to the conversion are of less value to your enterprise as they contributed less to the influence of the consumer.
The Time Decay model is a Multi Touch attribution model for the last channel is given the most credit, but the penultimate interaction is also credited, and the amount of credit decreases further down the conversion path. Each click is credited unequally, but they remain credited for they led to the conversion.
The Last Non Direct Click Model
The Last Non Direct Click model simply credits the channel which was last accessed before the conversion was made. Google Analytics uses this model by default when attributing conversion value in non-Multi-Channel Funnel reports. Due to this, the Last Non Direct Click model is useful when providing comparison to results from other attribution models. Furthermore, you may wish to filter out direct traffic and focus on the marketing which led to the click through to the direct channel leading to conversion.
The Last Google Ads Click Model
It is possible that your digital marketing strategy relies on Google Ads. If this is the case, then through the Last Google Ads Click model Google Analytics will attribute 100% of the conversion value to the ad which was last clicked before the conversion was made. This allows your enterprise to place value on Google Ads and track where they are best placed or which ad type is working the best. In regards to this information, your enterprise may chose to remove YouTube pre-roll ads and focus on SEO and keywords.
The Data-Driven Model
The Data-Driven attribution model credits conversions based on the contribution of each keyword across the conversion path, therefore it is different for every enterprise. By comparing the click paths of those who converted and those who did not, the Data Driven attribution model discovers patterns in the behaviour of the clicks which led to the conversion. These valuable clicks are then given greater credit by this model. This is useful for your enterprise if you wish to know which Google Ads are working the hardest. The Data-Driven model is also useful if you use an automated bid strategy to drive conversions for the whole conversion path is of some value to your enterprise.
Custom attribution models
As previously mentioned, there are customisable attribution models in addition to those default models discussed above. Customisable models allow for greater precision over how you distribute credit for conversions. Once created and saved, they will remain on your Google Analytics page.
Custom models can be manipulated to a precise set of assumptions you wish for your company to evaluate. For example, if your company values how much time a potential customer spends on a certain channel pre-conversion, then you may wish to adjust the credit accordingly. Other common customisable features relate to page depth, bounces and how much credit is given to which channel dependent on where they stand in the conversion path.
It is worth noting that you can only add 10 custom attribution models to your Google Analytics page, and the Data-Driven attribution model – discussed in the previous chapter – counts towards that limit.
Each customisable attribution model must be built from a baseline default attribution model. Each baseline model allows for access to different customisable options when applying attribution:
The Lookback Window adjusts how far back in time the attribution model will look to give credit to interactions with your ads. This option only narrows the existing Floodlight Activity Lookback Window, it does not expand the view.
Adjust Credit By Interaction Type
This tool enables you to set how much credit you wish to attribute to impressions on your digital content over interactions in the conversion path; for example, you can set a decimal value of less than 1 if you wish to give impressions less credit than interactions. You can also add timings to this attribution model, so impressions made 2 weeks before the conversion will be of less value than impressions made on the day of the conversion. The type of impression is also important to bare in mind when adjusting credit, if the impression drives clicks then it is of a greater value to the company.
For beta testers of Twitter engagement, this custom model enables you to set weightings for high value social engagements and low social engagements. High social engagements are worthy of greater credit for they reach other users and therefore expand your audience, for example retweets and shares. Low value social engagements, such as expansions and profile views, can be credited less as they usually involve a single user.
This custom attribution model is unavailable if you select the Final and Last Interaction baselines for they attribute 100% of credit to a single click rather than the whole conversion path.
Apply Custom Credit Rules
Your company is able to set up their own custom credit rules using specific criteria regarding which interactions deserve more or less credit than others. Custom credit rules are very flexible, providing your enterprise with full control over how to distribute credit. For example, you add credit to interactions which exactly match a paid search keyword.
If you create credit rules which apply to the same touchpoint, then the credit weighting will overlap and be multiplied according to your custom rating. For example, if you give credit of 0.2 to a video ad and 3 to interactions which occur on Facebook, then an interaction from this video ad on Facebook will be credited with 6.
When using First Interaction and Final Interaction models as baselines, Google Analytics will only allow you to use one custom credit rule as these baseline models 100% of credit to a single interaction.
Set Half-Life Of Decay
When using the baseline model of Time Decay, it is possible to amend the half-life period from 7 days. For example, if you wish the half-life to be 3 days then interactions on the third day will be credited with half of what interactions on the sixth day will be credited with, as the conversion occurred on the day 6. This is useful when your enterprise is running campaigns for a set amount of time.
Specify The Amount of Conversion Credit Based On The Position
If your baseline model is Position-Based, you can customise the amount of credit given to the initial interaction and final interaction, along with the amount of credit attributed to the interactions in-between on the conversion path. The overall total of credit must still be 100%, but this is spread over clicks.
You can conduct ROI analysis in Google Analytics by using the ROI Analysis or Cost Analysis reports.
ROI Analysis reports explain the ROI implications of your attribution models and helps to identify where your business could be optimising certain channels to increase your company’s ROI.
The Cost Analysis report shows you the revenue, session and cost data through non-Google marketing channels, for example Facebook ad campaigns and email marketing campaigns. To see Google Ads cost metrics, use the Google Ads reports tool.
Marketing ROI Analysis Report
Through an ROI Analysis report, you can track Cost Per Acquisition (CPA) and Return on Advertising Spend (ROAS) for each channel under different attribution models. The higher the ROAS, the more money your enterprise should be investing in these channels. Therefore, rather than just the regular CPA, you can also determine:
Similarly, alternatively to the standard ROAS, you can determine:
However, the default attribution model for an ROI Analysis Report is the Data-Driven attribution model. It is worth noting that the ROI Analysis report tool is only available on Google Analytics 360 Suite, the premium Google Marketing Platform Tool.
There are four main requirements for correctly completing ROI analysis in Google Analytics:
Without the above being set up as fully accurate, the ROI Analysis will not be accurate.
Cost Analysis Reports
A Cost Analysis Report calculates the cost of the campaign with the revenue made, helping you calculate ROAS and RPC (Revenue per Click). To access the Cost Analysis section of Google Analytics, look under the Campaigns subheading of Acquisition. You will find a line graph showing both ROAS and Cost Analysis data. Check how these results correlate with one another, the less you spend for greater ROAS the better for your company.
Note that without cost data (as explained above) Google Analytics cannot accurately provide a Cost Analysis Report. If this information is missing, the line graph presented to you will be flat as there is no data to compare. A reminder that cost data involves:
It is worth noting that in a Cost Analysis Report all ROAS data is determined using only the Last Non-Direct Click attribution model. For results using any other attribution models, you will need to use the ROI Analysis Report as explained above.
As marketers measure and attribute the impact of touchpoints and channels, using outdated attribution models can lead to misattribution, which can skew the accuracy of ROI measurements. Leveraging aggregate measurements like media mix models will not provide the granular insights marketers need. On the flipside, granular measurements like multi-touch attribution models will not indicate the impact offline channels and external factors have on marketing ROI.
With this in mind, it’s crucial for marketers to establish clear goals that indicate what internal and external factors make up your ROMOs, as well as how these unique factors can be measured (and subsequently applied to marketing ROI calculation).
Utilising the right attribution models and marketing measurement strategies is incredibly important to track consumers across the omnichannel landscape, leading to clearer holistic and granular results. Focus on understanding how platforms such as Google Analytics provide the capability to unify disparate attributions alongside your online and offline measurements. Armed with the knowledge of these analytics platforms, marketers will have clearer insights to use in their formulas—leading to more efficient and accurate ROI measurement.