Just how effective was your last digital ad campaign? Can you accurately gauge the number of eyeballs on it? Did it lead to more conversions and sales?
Can you confidently say it makes sense to continue pouring money into your digital ad campaigns—and can you back that up with numbers that prove it?
The only way to answer these questions is by tracking media ROI.
In the first of a three-part marketing ROI tracking series, I interview ex–Starcom and Spark media planner and buyer Samood Rehan to paint a multifaceted picture of media ROI: What does media ROI mean? How has it evolved since pre-digital times? How should businesses choose the right metrics for their media campaigns? Where should automation come into play?
And what sort of generalizable media ROI tracking process can marketers modify for their own needs?
Read on to discover all of that, and make it feel concrete with a case study on how ride-sharing app Careem leveraged its ROI insights to improve its media strategy.
“Eyeballs,” answers Rehan.
The most basic definition of media ROI, he says, is how many people have seen or interacted with a campaign.
Each media format measures eyeballs in its own way.
Print uses circulation numbers—multiply this by the average household number for that country—to get readership numbers.
Radio has listenership as a KPI.
Television utilizes metrics like reach, which is the number of ad viewers divided by the total target audience base. Another metric is Gross Rating Points, which calculates duplicated reach: How many times did the same individual watch a particular ad?
At the end of the day, the key ROI of these traditional media avenues is exposure.
That said, it would be a mistake to view exposure as the be-all-end-all of media ROI. Exposure itself is a measure of brand awareness, and top of mind (TOM) recall. The more an audience is exposed to Magnum display ads, the likelier it is they’ll subconsciously choose it from among a dozen ice cream brands at the grocery store, says Rehan.
What’s more, exposure as a key ROI metric is directly tied to engagement as a KPI. In the pre-digital media landscape, when the objective was engagement, marketers might ask viewers to call a number they saw on TV to avail a product or service. They could then track and measure this specific outcome immediately through the number of calls received.
Since the rise of digital media, this approach might be enhanced through QR codes and URLs, which viewers could scan to be taken to a unique landing page. The number of unique visitors can then be tracked through tools like Google Analytics.
And of course, digital marketing adds the element of enhanced traceability to the mix, letting you track which clicks on what ads lead to a website and ultimately to conversions and sales. Marketers can literally assign sales value to their ads and impressions with assurance, meaning the number and volume of sales can become part of media ROI as well.
For Rehan, the question of how a business should determine the metrics they want to track boils down to their unique business and campaign goals.
He points to how metrics like YouTube views and Google ad impressions are usually tied to brand awareness. Brand awareness is the degree to which a target audience (1) recognizes and (2) recalls your brand. How well can they identify your brand when presented with its name, logo or other visual cues—how well can they remember it even when prompted without visual cues?
For campaigns where this is a top priority, metrics should focus on impressions, reach, and even engagement in interactive formats. Rehan notices how ad formats targeting these metrics are highly desired by large multinational corporations like Coca-Cola and Nestle, that are in a constant fight for awareness and TOM recall.
“That’s why most of these multinational brands with millions and sometimes billions of dollars of turnover would adopt each and every media format out there, from print, radio and digital to out-of-home, billboards, you name it,” Rehan says.
But if your business is focused on sales, then conversions and sales value become key metrics. Return on average advertising spend (ROAS), for instance, can be a useful guide for e-commerce sites: Specify the value of a user clicking on an ad, selecting an item into their cart, and then checking out by providing their credit card details. Such a use case more or less gives you an exact value of the return of your advertising spend.
That said, marketer beware.
There is no way to be certain about the source of an individual’s conversion, meaning we can’t pinpoint the exact media that makes the customer decide to engage or convert
Let’s say you develop search-based ads, meaning that if a user enters a search term, they will view a relevant ad. Clicking on that ad brings them to a landing page, with another remarketing campaign that shows display ads. They see those ads throughout the week and eventually convert.
The question remains: Was it first click attribution or last click? Did the individual convert because she saw the display ad or the remarketing ad? Or was it down to the very first search ad?
Some attribution models are pretty direct with these answers. They give all the credit to either the first click or the last, or maybe spread it out linearly so each touchpoint equally contributed to the conversion.
Others are more sophisticated, to better reflect the difficulty of identifying just what made a customer convert.The time-decay approach, for instance, assigns more credit to touchpoints that occurred closer to the conversion event. (Here, the remarketing display ads that the user saw throughout the week leading to conversion would receive more credit than the initial search-based ads.)
Multi-touch attribution is another popular method that lets you customize credit assignment across touchpoints. The team would first clarify the intended goal of each touchpoint. Is it meant to create brand awareness, generate leads, or drive conversions?
Maybe you determine here that conversions are the primary goal of search-based ads, but a secondary goal for the landing page, which focuses foremost on lead generation. These decisions should be backed by MarTech tools that track user interactions across the touchpoints, and gather data on users' clicks, website visits, ad views, and conversions.
Based on this data, the model may then allocate an attribution weight of 30% towards the search-based ad, the landing page visit 20%, the remarketing display ads 20%, and the repeated display ads 30%. Multi-touch attribution doesn't result in a single "score" like first-touch or last-touch attribution models might, but rather provides a distribution of credit across touchpoints.
What model you end up using for your brand depends on your business goals and data insights. You have to determine whether your customers engage with multiple touchpoints or have a more linear journey. Constraints on resources like time, budget, and technology play a role in this decision, as does the availability and quality of your data. Some attribution models require far more comprehensive and accurate data…select the attribution model that best aligns with all these considerations.
Rehan gives the example of Gallup polls that used to be a prime source of data, where people would physically write down the hours they’d watch TV during, their preferred channels, and even who was watching TV together in a household.
Obviously, this system yields rich data in theory, but it was ultimately flawed. The process depended on individuals manually recording data points about their own viewing patterns and behaviors. Generally speaking, people wouldn’t fill out this information during their actual viewing times but postpone it until when the Gallup employees came to collect it.
The result was data riddled with human errors: After all, could you remember exactly who you watched that episode of CSI: Miami with? The one where Chris Pine was the guest star?
For Rehan, minimizing human input is a crucial part of data accuracy. Automating the data collection process reduces human mistakes.
With this in mind, the Gallup system was updated with remotes and data capture boxes that were programmed to accurately collect data. To track media for ratings, individuals from randomized households would have to enter specific codes through their remotes to access their televisions. This would activate the profile associated with that code on a separate box connected to the television, which would then capture viewing information.
It’s essentially the same thing Netflix does now with your profile – collect automated information on your viewing habits (which it then feeds back into your recommendations. See the end of this blog for a more detailed case study on leveraging media ROI insights into strategy.)
Basically, the more removed humans can be from data, the better.
Data irregularities show up within digital media as well. Rehan recounts a time when he’d see “a heckload” of people clicking on their ads and coming to the website, but no one would convert. (The communication and content was on point.)
So the question arose: Were they bots or fake clicks? Google Analytics indicated that they weren’t either of those.
The turning point came when Rehan and his team grew skeptical of the data. HotJar—a tool that tracks clicks and website actions—gave Rehan the answer to his conversion problem. They had bots.
The conclusion? For accurate and reliable data, it’s imperative to not rely on a single data source. Not even if it’s Google. Always have multiple data sources.
Where Martech can really help with media tracking is in those areas of maximized automation and minimal human interaction, says Rehan. What’s more, it can integrate data from multiple sources so marketers have access to consolidated information. That’s what makes marketing tech such a powerful force for eliminating attribution anxieties – it really helps determine a balanced mix. Keep validating your data though!
Automation, through CRM tools for instance, also makes it phenomenally easier for marketers to set up triggers, rules and workflows for digital media tracking – “if they understand data”, cautions Rehan. With the introduction of AI, this sort of automation can only evolve and become more necessary to yield real-time data insights and optimization.
Take SmartAssets, for instance. This creative optimization tool uses a brand’s performance data to recommend (and execute) tweaks to ads. Marketing technologies like this are at the center of the future of media performance and assessment.
With all of the above in mind, here’s a generalizable step-by-step to planning out your media ROI tracking process:
The Dubai-based ride-sharing app Careem operates in 12 countries across the Middle East, Africa, and South Asia regions. Rehan told me about its importance in Pakistan, where the service helped mobilize working people within metropolitan cities in the absence of well-developed public transportation and mass transit.
Specifically, the service was a turning point for working women, Rehan says, who faced further challenges without safe transport facilities. Men could travel around more easily on bikes—still somewhat outside the norm for women—or avail overcrowded buses without danger of harassment.
So the media tracking team at Careem noticed quite early on that a greater percentage of their rides were for women, and furthermore, that female users usually responded better to the digital ads.
These insights fed into a long-term strategy that incorporated feature planning such as ride tracking and sharing. Women were using Careem to ride with a driver who was a complete stranger, but they could share these rides live with friends and family.
And the ads turned to using more female characters than male ones in affiliation with this newly recognized target audience that was more receptive and responsive to the ads – further pushing up key engagement rates.
