How to Track Your Marketing Performance- From Zero to ROAS to ROMI

Executive Summary (TL; DR)

Following up from our last blog- Mastering Marketing ROI: Tracking Success, this blog looks closely at the difference between two key metrics used in Performance Marketing- ROAS and ROMI, their differences, and when to use either. And then suggests a roadmap for Marketing Teams to follow. Depending on where you are on your journey, your needs and capabilities it suggests a graded approach from basic metrics (like bifurcating Marketing Spends into Branding, Lead Generation and Others) through Conversion Tracking, ROAS and ROMI Models (Last/ First Click through to the holy grail of Multi Touch Attribution).
The blog comes with a downloadable Excel template to guide you on your journey and explores the difficulties in implementing ROI Tracking in non-purely ecommerce environments. It explores how using a Lead Management System/ CRM (like Pronnel) along with an Analytics suite like GA4/ Amplitude/ MixPanel etc can help in building a more robust and accurate framework.
The blog concludes with a short guide on the Golden Rules for Tracking in Performance Marketing which can be a guide to your efforts with whatever tools at your disposal.

What is the difference between Marketing ROI, ROMI and ROAS? When should we use each of them?

ROI (Return on Investment) comes from financial analysis and is defined as:
ROI= [Total Revenue- Total Costs]/ [Total Costs], expressed as a %age.
With rise in Marketing budgets and when firms started looking to quantify the results of the Marketing department’s work, the same concept was applied to Marketing. Marketing ROI, or, as sometimes called ROMI (Return On Marketing Investment) seeks to measure the efficacy of Marketing vs its costs. Use it to determine and modify Marketing Strategy.
ROMI= [Total Revenue-Marketing Costs]/ [Marketing Costs], expressed as a %age
Firms may define Marketing Costs differently. In the strictest interpretation, Marketing Costs should be inclusive of the cost of Marketing hours, agency fees, Ad spend and so on. What is most important is that a company decides on a common understanding from the beginning. As with other metrics, keep in mind the context, that the number is not the end all but trends, and reasons behind are key to actionable changes.
Return On Ad Spend (ROAS) looks only at the efficiency of the Ad Spend in a campaign. Use it to determine which Ad Campaigns need to be boosted and which need to be axed.
ROAS= [ Campaign Revenue]/ [Campaign Costs], expressed as a %age
Both ROAS and ROMI are good metrics and offer different insights.

How Do You Start if you have never tracked the performance of Marketing spends?

Tracking Marketing Performance can give you a significant edge compared to competition. The best-case scenario is when you can track granularly the impact of different touch points in the Sales Funnel and develop an attribution model that is unique to your industry- business-product line.
However, it is easier said than done. Trying to start off from day one with a very complex model can be challenging- Firstly, because of the amount of effort and resources needed to build an effective model, and, secondly, it is very easy to get it wrong and get overwhelmed by a mass of data which is difficult to make sense of.
In cases often across large and small organizations, Marketing Spends tracking just become a couple of lines in the P&L. Marketing spends are just compared to Revenue generated in period, maybe as a %age line item comparing current period to last period- often causing gaps in understanding. Other places, it may be slightly more detailed.
Our Roadmap to truly understanding how effectively your Marketing dollars are working for you is given below. Depending on the stage at which your organization is, you can jump to the appropriate section. Like all things Marketing the following model is generic, and you need to work in changes as you iterate and refine which are applicable to your industry (gestation period, whether it’s a one time/ repeat purchase cycle, sales effort and employee costs needed).

Stage 0: No Tracking Currently- Bifurcate Marketing Spends

If you are not doing any form of tracking, start somewhere. Setup a cross-functional team of Marketing, Sales, and Finance. Classify your Marketing spends between Brand Building, Lead Generation, Administrative/ Statutory Requirements/ Miscellaneous. Try to classify your revenue so that you can have an idea of how much is Base revenue, how much is directly attributable to Lead Gen Campaign. It’s not great, but it’s a starting point and in a few months, you can iterate and refine the model based on patterns you can see. With this base model, you can at least have a clearer picture between the teeth (Lead Generation) and Tail (Brand Building/ Statutory/ Administrative) of your Marketing spend. For example, often a business may just be starting off and inclusion of website/ App build, basic signage, collaterals may bloat the Marketing budget giving the false impression that revenue generation against spend is very inefficient. In such scenarios this analysis may come in handy.

The key thing is that there is no single figure which can tell you the whole story but looking at numbers in comparison (Current Year vs Last year) or look at outliers and dig deeper. The Excel sheet for this model can be downloaded from here ).

Stage 1: Tracking Leads and Conversions

Suppose we have covered this stage. What next? An interesting staring point would be to look at leakages and throughput of the Sales Funnel. Once you start tracking this data, then the next logical step Conversion Rate Optimization (CRO) to increase your funnel efficiency and start converting more leads becomes easy. Sounds logical right? While the online component, Impression to Click Through to Form Fill/ Call, portion is easy across platforms, unfortunately, even now firms face challenges with this. Any firm which is collecting leads from a larger number of sources, has an offline component to their business, will have difficulties in aggregating all leads into a single window. Even, ecommerce firms may face CRO challenges if their CDP/ Analytics platform are not configured perfectly.
Here, our objective should be to be able to trace each lead from its source medium (unique campaign), through its CTA, to a dashboard where we can see how it is converted. By tracking this granularly, we would find the steps at which there is a leakage. The resulting dashboard should be a data table or a graph which provides the following information.

Start off with an Excel sheet at the least, though it’s recommended that you start using a Lead Management System or a CRM like Pronnel to help you track the conversions much more easily. Besides giving you an idea of what sources contribute to how much Revenue, your dashboard should give you an insight into Conversion Rate Optimization (CRO)- to track the throughput of your Marketing Funnel and identify bottlenecks.
Needless to say, you will need to look at other data sources like your Analytics Tracking tools like Google Analytics/ MixPanel/ Amplitude to keep an eye on your visitor’s behavior online.

Stage 2: ROAS-Last Click Attribution

While we would want to go straight to tracking ROMI, starting off directly maybe a challenge for teams who have never tracked returns on Marketing before. For Teams like this, it would be a practical step to start small. Start with Return On Ad Spend (ROAS). To start with, it’s best to start with something like a Last Click Attribution, which implies that the Ad/ Conversion was totally due to the last interaction of a user with an Ad/ Piece of Content). While this is not the best way, and there is enough academic literature available discussing the drawbacks of using Last Click Attribution (for a discussion on the drawbacks Death of ‘last click wins’: Media attribution and the expanding use of media data. Garry Lee- Aug 2010), using Last Click Attribution can simplify things for teams starting off. Here the key difference between the previous model and this one is that we are not only tracking Marketing Spends, Conversion Rate Optimization (CRO) but also looking at which sources are giving us what kind of ROAS. Another drawback worth mentioning for this ROAS model, particularly for industries with long Sales gestation period, is that it does not track the cost of nurturing a lead during different stages in the conversion funnel. For example, a lead which is lost in the qualification stage has fewer manhours spent on it, while if a firm loses a lead in the final negotiation stage, the number of manhours spent on it can translate into a very tidy amount.
At this stage, you need a proper Lead Management System or a CRM to track and measure performance. There is no way in which Google Sheets or Microsoft Excel can help you track this level of detail. At the least you need a CRM tool like Pronnel to automatically aggregate all leads generated by Marketing and Sales and bring them into a pipeline which can then be acted upon by the Pre-Sales/ Sales teams for conversion.
So, how do we do this? This will involve very close collaboration between the Marketing, Sales and Finance Teams. The first thing to do will be to track the monthly revenue attributable to an unique source ID (you can use a variable like utm_id as an approximate) along with the amount of money spent on the Ad. If an Ad has multiple CTA’s you can use your judgement to appropriately divide the Ad Amount spent (or altogether keep level of analysis at Campaign level). Your dashboard should look something like this.

Again, as previously, it will be insightful to look not at numbers individually but to recognize the patterns and zero in on the outliers. The data will give you insights into your best performing campaigns beyond the CPL/ CPM detail. Where using a Lead Management System over Google Analytics is advantageous is that it can handle offline conversions and conversion amounts which are related to repeating revenue streams like subscriptions.
Read along with the Conversion Efficiency metrics, this should give you sufficient actionable insights to streamline your conversion funnels.

Stage 3: ROMI/ Marketing ROI- Last Click Attribution

An improvement on the previous model will be if we start looking at ROMI (Marketing ROI). Just like in typical accounting we use ROI (Return On Investment) to measure efficiency of total spending to generate Revenue, ROMI looks closely at the Marketing and Sales Spends to generate the revenue. Now, we can factor in difficult to track costs like the average spend in nurturing leads between different stages, the Advertising spend, and even allocate the spend on content creation to individual assets.
At this stage, even if you are a full ecommerce business, you will need a full-fledged CRM suite to track and make sense of the attribution of Revenue as indirect costs are not captured in your normal ecommerce analytics suites.
The first step would be to capture Indirect Marketing Costs. This allows us to allocate a cost to the generation of Organic Leads. You will use this data to store annual data on the various indirect costs and record them against Lead Source ID’s. For Organic, you can track it against the page where the conversion happened (data that will come in handy where you needed to defend your position on why the Marketing Team is investing time and money in Organic assets). (Content creation, employee cost attributable to creation of assets can all be accommodated here. You will need to generate a data table as shown below:

Now that we have looked at Marketing costs, let’s look at directly attributable Sales Costs.
Now, in purists’ terms, a ROMI calculation should not include the costs associated with the conversion in the Sales Cycle and you could omit them. But in case you are in an industry with long gestational periods or heavy expenditure in the post-lead acquisition period, you may want to look at this.
Start with getting the Sales and Finance Team together to compute the average cost that is incurred in manpower/ travel/ collateral to advance a lead through each stage. The premise being that a greater amount of investment has been put into a lead at a later stage vs an early stage. This can give an idea of value sinks and together with the CRO Model help you generate efficiencies.

Using the table as a base, you can then allocate a cost to each lead in different stages of the pipeline in a data table as given below:

Now that we have the basic information, it is time to aggregate the data into a model that gives us a ROMI at a channel level.

You should look at the numbers in different ways:

• Overall ROMI for a month vs same number for previous month/ same month last year.
• Look at splits at Organic-Direct/ Paid Search/ Social/ Paid Social/ Paid Influencer/ Events/ Others
• Drill down and look at data at channel level (Facebook, Instagram, LinkedIn, TikTok)
• Drill down and look at large campaigns.

There are a few grey areas here. For example, you may have ads which run across months (split them into monthly spends), ads which generate a lead whose conversion goes to the next month (attribute zero spend in next month vs the revenue). At a macro level, this data will give you a sense of the ROMI and you will have to investigate outliers.
A special case is Organic. Since revenue conversions from Organic is long term, it makes sense to track a lifetime return on Organic if you wish to build a case for why we should invest in Organic Lead Generation. Ideally with time, you should be able to plot snapshots of Organic ROMI on a month-to-month basis with hopefully a slow but steady rise.

Stage 4: Tracking ROMI with Multi Touch Attribution

It takes more than a single Touch Point for a user to convert into B2B sales. Estimates for B2C are higher. Therefore, if we use a Last Click Attribution Model, or, a First Click Attribution Model, we run the error of seriously underestimating the effort needed and ignore channels/ media which are important for us to accurately track the User Journey which resulted in Sales. Tempting as it sounds, it would be amiss of us to jump headfirst into a discussion on implementing a Multi Touch Attribution (MTA) Model for ROMI if we did not consider the challenges:

• Even if you run a simple ecommerce business with an online acquisition- online conversion business model, your consumers would be exposed to messaging outside of the digital world making it hard to build an accurate MTA. If there is an offline component in conversion, this becomes even more difficult. Without proper care the chances of getting the wrong model and making wrong calls based on it is something to consider.
• Even if your complete business model is online, privacy laws and consumer pushback to tracking has been a spur for browser firms, and mobile OS providers to restrict cookie and tracking devices. Making it harder to get an accurate picture. Errors may creep in unwittingly because of this.
• The rise of #ROPO (Research Online- Purchase Offline) as a trend is making it harder and harder to observe the entire consumer journey, disrupting our ability to understand and track. Even before the pandemic, 87% of consumers preferred researching a brand online before buying offline (87% of shoppers now begin product searches online: Retail Dive, Aug 2018)
• MTA models take time and effort to build, test out and need continual refinement. Consider the time and effort needed vs the incremental value before embarking on the journey.

Now that we have that out of the way, let’s look at what would be a possible strategy for this. If you have a completely online business, it is fairly easier. With limitations, Google Analytics Multi Channel Funnel (at channel level provides First Click, Last Click, First Interaction, Linear, Time Decay and Position based models), Amplitude, Kissmetrics and Adobe Analytics all provide heuristic models (pre-defined rule based models). The problem with all these models is that none of them will exactly match your unique User Behavior and would need to be monitored, revalidated continuously. For high volume businesses, a better way could be to harness Machine Learning to develop a Data Driven Attribution Model (GA4 has a default option for a Data Driven Attribution Model built into the system)
Given the vast set of options available, a full discussion is beyond the scope of this blog. We will take up a simple illustrative example of an online business with a single purchase event and show how we can run an MTA with the help of Google Analytics and a CRM like Pronnel.
Using the All-Channels Report, identify the Revenue generated and ROAS at Channel level. This data can be combined in the previous model for Stage 3: ROMI/ Marketing ROI- Last Click Attribution’s worksheets, replacing the Conversion Amounts.

If you want to go beyond ROAS and arrive at a Marketing ROI for the MTA Model you will need to work a little to factor in indirect Marketing costs. Since this is a completely online model, the tables for Sales Direct Costs will not be needed, but the Indirect Marketing costs can be added on to arrive at a ROMI at Channel Level. If we were to enter some made up numbers which the Marketing and Finance Team have jointly assigned as Indirect Marketing Costs to each of these channels the ROMI numbers would look something like this. You will note that while the ROAS evaluation had shown a 46X return, factor in Indirect Costs using ROMI has brought down returns to 7.9X

Note this will work only for an online-online model. Also, readers are referred to “What is the new Data Driven Attribution (DDA) Model from Google? Is the DDA Model available to everyone?” in our FAQ section to see if they meet the eligibility criteria for using a DDA Model.

Golden Rules for Tracking Performance Marketing

Cost Benefit Analysis.

While having greater insights into the efficacy of your Marketing efforts is great to have you need to weigh the effort and time required to setup detailed tracking. Consider also, the maturity of your team and agency’s ability to put in place the assets for this tracking.

Involve Finance, Sales and Marketing Teams to build your model.

This way you will have alignment and ownership of how you track and the actions thereafter. Not everything is cut and dried and may need cross-functional teams to agree on so that the organization has one consistent and coherent way of classifying spend and revenue attribution. This team will also be able to identify critical parameters to track from an organizational perspective and go after the low hanging fruit.

Start Small with a Simplified Model.

Start measuring. If you have never tracked Marketing spending, move to Stage 0, then Conversion Tracking, Return On Ad Spend (ROAS) before choosing a ROMI Model. For example, if your Ad expenditure is the lion’s share of your Marketing Expenses and Direct Sales Costs are not material, then you could focus on only the ROAS.

Investigate what stages have an outlier and work on them. Act on outliers and grey areas.

Rule of thumb first analyzes outliers to see reasons. Focus on breaking down big blocks of data. Delve deeper. There is so much improvement that a good attribution model can help you achieve.

Iterate and refine your Models-

No tracking model is going to be perfect from the first day. It is key to consider different models, compare numbers and constantly work on identifying areas of grey.

In the following articles of this series, we will be investigating the media landscape, different avenues for lead generation and conversion, paying close attention to scenarios where consumers have a hybrid online/offline acquisition/conversion path. We will briefly look at tools like Google Analytics et al before answering questions on how to track conversions offline and tracking their efficacy. Finally, we will look at how privacy regulations around the world have affected lead tracking methods (cookie phase out, restrictions on PII) etc.

FAQ’s

Google Analytics provides me with Cost Per Lead. So does Facebook provide me with the cost per Lead. Why do I need to invest in a CRM to track Marketing ROI?

Firstly, except in the simplest cases like ecommerce with no after Sales component you need a CRM, or at least a Lead Management System. While Google Analytics can trace and track the online portion of a buyer’s journey, anytime an offline component, or, post Sales, or a long purchase cycle comes in, these tools fall short. Software like Pronnel can act as your Customer Data Platform (CDP) and give you insights into customer behavior over time.
Secondly, as platforms and lead sources multiply, you end up with multiple logins and dashboards. Making it harder for teams to manage them. An integrated omnichannel Lead Management System or CRM like Pronnel can cut down this effort greatly.
Finally, granularity of data and the ability to slide and dice data for analysis as you need. By bringing all leads and their conversion path into one platform you will be able to reduce the effort needed to analyze.

What is the best attribution model to use?

The best attribution model for conversion tracking is one that your team can setup easily and interpret. As we have discussed, setting up Marketing Tracking is time consuming, and one can easily get overwhelmed with data. Unless you are sure that your model is accurate and you understand the data, the best is to take baby steps, identify gaps, iterate. Also keep in mind that there are no universally applicable Attribution Models. While Data Driven MTA Models come close, remember that such Machine Learning based models require a lot of data to learn and it is always a good idea to keep cross checking this model with other models. See also, our question on the different attribution models in vogue.

How do you calculate Marketing ROIs for non-revenue generating objectives? Like signups for newsletters or accessing a Store Locator?

Calculating ROI for non-direct revenue generating events can be tricky. But then again it is definitely true that these events like newsletter signup, Store Locator, edge the client to a purchase. Here is an interesting blog- What is a conversion worth?- Roboboogie to help you get started.

What is the Ideal Marketing ROI number?

The answer to that is it depends. The stage of growth (typically high growth firms have lower Marketing ROI in initial periods), your type of business, and ultimately up on your strategic objectives. Typically, traditional firms with well-established business models and stable growth will gun for a Marketing ROI of 10:1. Newer firms in the growth stage will try to generate a Marketing ROI of 3:1, a number popularized by VC’s and investors. Early-stage firms with a ROI> 3, would typically need to reassess their business models and ask if they are underinvesting in Marketing and losing opportunities to grow further. Whatever the scenario, a Marketing ROI below 3, say around 2.5 is a sign of danger. Most likely, this firm is spending too much to acquire customers, or is not able to retain customers. This firm needs to analyze what’s wrong and set to work addressing that issue before making any further efforts on acquiring new customers.
As the above formulae indicates there are multiple ways in which you can impact the Marketing ROI number. For example, building customer loyalty through Customer Centric programs can help you retain the customer over a longer period resulting in additional cash inflows from the client. In such cases, you should evaluate how much extra you could invest in Customer Support and Retention to lead to a better ratio.
A great starting point when embarking on this journey is first tracking and recording Leads from your different Ads and Campaigns.

What are the different Marketing Attribution Models? Which are more popular? When should we use them?

Attribution models can be single point (First two in list below) or multi-point (last seven examples in list below). The commonest Attribution Models in vogue are:

• Last Click Attribution-

  All credit for a final conversion is assigned to the last interaction a User had with your brand. While simple to implement the disadvantage is that it really doesn’t do justice to measuring how a User converted. User journeys are becoming more and more complicated and using Last Click does not show you the different assets through which your User was moved along in the purchase journey.

• First Click Attribution-

  All credit is assigned to the first instance a User had with your brand. An advantage is that it can help you uncover the time a User takes between the first interaction and the time taken to convert finally, whereas its disadvantage is that it is difficult to track and can throw up erroneous data.

• Linear Distribution Model-

  A linear distribution model tries to consider all the different interactions and give them an equal weightage. While it considers different interactions, the disadvantage is that it will give inaccurate data as all interactions are not equal.

• Time-Decay Attribution Model-

  An improvement on the Linear Distribution model, the Time Decay model is based on the principle that the closer interaction between a User and the brand has a higher weightage than an interaction in the distant pass. It assigns weights to different interactions by the time distance from the final conversion, usually considering a half-life of 7 days. The disadvantage is that it is difficult to set up and still has inaccuracies. Participation Model- This gives 100% weightage to each interaction point, inflating the number of conversions.

• U-Shaped Attribution Model-

  It gives 40% weightage to first point and last point each and divides the remaining 20% equally between all the other points. This recognizes that the first asset that brought a customer to your site, and the last asset before conversion should have higher credit. For example, if a User had a total of 7 interactions, then the first point and last point would get 40% weightage each, and the remaining points would have 4% each.

• J-Shaped Attribution Model-

  Based on the principle that the first interaction is important but more important is the last interaction, giving them 20% and 60% respectively. If there were a total of 8 interactions, then the conversion weightages assigned would be 20%, 2.857%, 2.857%, 2.857%, 2.857%, 2.857%, 2.857% and 60% respectively.

• Inverse J-Shaped Attribution Model-

  Similar to the J-shaped Attribution Model, this one is based on the idea that the first interaction is most important and last also important, and gives them 60% and 20% weightage respectively, and divides the remaining credit equally between all other points.

• Data Driven Attribution Modeling (Google Analytics)-

  As we can see, there are myriad ways in setting up rules for Attribution Models. Which one is best is left up to interpretation. A step ahead, takes advantage of Machine Learning and uses actual account to build Attribution Models. While discussing the various Attribution Models it is worthwhile to mention in passing the new Data Driven Attribution model launched by Google with GA4 because of the popularity of GA4 and the fact that it is likely to become a de facto industry standard. All the above attribution models are rule based. Unlike the above 4 models, this model is not based on pre-decided rules. Data driven attribution models use actual data to allocate weightages to different interactions. We will discuss the new data driven attribution model from Google in detail in the next section.

What is the new Data Driven Attribution (DDA) Model from Google? How Does It Work? Is the DDA Model available to everyone?

GA4 has come up with a new Machine Learning based Attribution Model which uses an account’s own data to devise an optimal Attribution Model. It does it by analyzing the available path data to develop conversion rate models for each of your conversion events, and then using the conversion rate model predictions as input to an algorithm that attributes conversion credit to ad interactions.
One drawback is that for advertisers with low conversion volumes, the DDA model will not work. As per Google’s documentation, need to have 3,000 ad interactions and 300 conversions over 30 days to be eligible for data-driven attribution. To remain eligible, these conversion actions must continue to generate 2,000 ad interactions and 200 conversions every 30 days, making it out of reach for most MSME businesses.

What Attribution Models are being sunset by Google? Will we have to rely only upon Google Data Driven Attribution Model?

On 7th April 2023, Google announced, that they will be sunsetting all Rule Based Attribution Models, except Last Click Attribution. That means that going forward, First Click, Time Decay, Positional Attribution models are not going to be available on GA4 and Google Ads platforms.

Today we announced we’ll be deprecating non-last-click rules-based attribution models, which includes first-click, linear, positioned based & time decay, in Google Ads & GA4.

Data-driven attribution (DDA) is recommended & last-click will remain available.

More on why & timing:

— AdsLiaison (@adsliaison) April 6, 2023

While the data driven attribution models represent black boxes, for Marketing Teams the old faithful of Last Click Attribution will continue to be there. Of course, if you are using other suites like MixPanel, Amplitude, Kissmetrics, Oracle Net Suite, you will still have access to the different models for now.

Read Also: