A Marketer’s Guide to Major Measurement Methods and Triangulation/ Unified Measurement
A general exploration of key measurement methods, what they’re used for, and how insight and effectiveness teams use various data and techniques to build the effectiveness narrative in the privacy age.
Key Measurement Requirements
Brand Awareness and Perception
Continuous brand tracking
Surveys, panels, focus groups, social listening, network analysis, sentiment analysis, etc.
Brand tracking helps us to understand where our brand sits against KPIs, though it doesn’t influence it.
Sometimes these methods are used as ‘one-off’s, though for larger brands, they are often ‘always on’. Similarly, they can vary between qualitative data (e.g. information gathered in conversational form from focus groups) and quantitative data (e.g. from ongoing survey responses).
Awareness (prompted/unprompted), consideration, ad awareness (prompted/unprompted), penetration, purchase intent, brand desire, top of mind.
Can be continuous or snapshot. Depending on cadence, (e.g. quarterly), may need to build history for two or more years. Insight is then delivered in weeks.
Includes Kantar, YouGov, Ipsos, etc.
- Measures how the brand is perceived, including desirability, and gauges performance against competitors.
- Adds richness and emotional intelligence to your brand story.
- Often limited in relationship to outcomes (e.g. sales) from brand tracking alone. Though can be nested in Marketing Mix Modelling (MMM) – see later – to provide the link.
- Can place too much emphasis on emotions that are not linked to behaviours; what people say they think is not necessarily a driver of what they do.
- Can be limited in statistical validity (sample bias/size for one ‘one off’ or small studies).
Granular, tactical and reactive campaign decisions
Attribution
Last-click, multi-touch (MTA), first-click (rarely)
A prescriptive method that helps with campaign optimizations. Attribution, as we know it today, became popular in the 2010s in digital advertising due to the opportunity to ‘follow’ individuals around the web using cookies. This made it possible to understand the user journey and link paid digital activities, whether search or social (pay per click/PPC/biddable media), to sales. Last-click remains the most popular method, with ‘multi-touch’ (MTA) considered more refined as it assigns a weighting to measurable touch points along the journey.
Conversion metrics, such as a click, website visit or add to cart. Performance KPIs (outcomes that can improve through optimization) include CPC (cost per click), CPV (cost per view), CTR (click-through rate), CVR (conversion rate) , ROAS (Return on Ad Spend), Impressions, Downloads (e.g. for apps), Sales.
While insights are often available within the first day, it typically takes time for campaigns and automated optimizations to stabilize. Ideally 90+ days for history (but no shorter than one week within a campaign) and insights within weeks.
Typically large platform providers such as Google Ads, Google Display Network (GDN), Meta, etc.
- Granular data persists, despite cookie deprecation/ IDFA loss, though not to the extent it once did.
- Rapid feedback allows for intra-campaign optimizations.
- MTA attempts to understand the interaction along the path to purchase in a digital environment.
- Typically limited in channel scope (mainly digital) – i.e. partial channel picture and mis-attribution potential.
- Limited understanding of incremental relationship to outcomes (e.g. sales); some activity may be in the path rather impacting the path.
- Measures of ROI used in attribution may not match broader measures given they relate only to a specific channel/campaign.
Continuous innovation and improvement
Experiments
Consumer- or geo-based tests
Experimentation aims to support taking calculated risks on smaller populations to statistically determine whether a particular lever (e.g. targeting or creative) has a genuine causal impact on an outcome, e.g. sales or enquiry. This is done by comparing response behaviours between a test group, and a control group. Groups could be, for example, customer segments, a set of stores, or regions. The test must be set up properly to ensure the results are statistically valid and learnings can be relied upon when it comes to rolling out ‘winning strategies’.
Sales, CTR (click-through-rate), open-rates, etc.
Insights in weeks, data history required is for the duration of the test (plus previous activity).
Includes sales or other outcome data, the metric of interest (e.g. media spend, store installation etc.). Sometimes combined with broader MMM sources to contextualize and enrich robustness.
- Experiments can be an entry point to building a measurement capability – more focused/less data intensive on a few specific levers.
- Can augment Marketing Mix Modelling (MMM) by filling measurement gaps & pushing the boundaries of MMO optimization potential, resulting in bolder decisions and more agility.
- Helps to focus on learning and innovation and measure new levers; can be a good way to help find new audiences, develop localized tactics or support major shifts in spend with a low-cost approach to a test and learn culture.
- By nature, experiments are hyper-focused and will only provide incrementality for a specific lever/strategy.
- The measured lever needs to be executed such that it is possible to measure incrementality through testing (Note: MMM can measure any past lever).
- While often mitigated, in some fields, it can be difficult to maintain perfect control that is free from contamination due to various external factors.
Budget planning, performance forecasting, strategic direction
Marketing Mix Modelling
Econometrics, econometric modelling
Marketing Mix Modelling (MMM) captures the full range of business drivers, including external factors, to tell the whole story across all marketing activities and the role each plays in delivering an outcome. It measures interactions between channels and the relationship between an outcome (e.g. sales) and influencing factors (both marketing and external, e.g. economic pressures) to understand ‘incrementality’ (outcomes that would otherwise not have occurred) over the long term. That means it can fill the gaps left by methodologies such as attribution.
Sales, revenue, profit, ROI, brand, traffic.
MMM can take months to deliver insight and require 2-3 years of data history.
Uses a multitude of different data sources throughout the funnel, across channels and outcome responses.
- Impactful insights and an understanding of both different marketing techniques and external influences on outcomes that allow you to determing big budget decisions. Introduces common language and begins to break silos.
- Modern methods – especially facilitated by AI/ML complemented by human domain knowledge/subject matter expertise, integrate with more frequent and more granular data – are overcoming previous perception that MMM is old-fashioned, infrequent, backward-looking and laborious.
- MMM can be expensive to implement and can take longer to produce insights.
- Poor definition of measurement requirements in campaign briefing can result in not capturing the data necessary to the desired analysis.
Strategy linked to tactics through measurement
Triangulation/ unified measurement
MMM, attribution, brand tracking, test & control
A mechanism for integrating all analysis into a singular framework to capture a whole picture. The gold standard has MMM at the heart for integration with different ‘lenses’ from other methods supplementing and enhancing the story, according to the question at hand.
Sales, Revenue, Profit, ROI, AOV, etc – encompasses all previously discussed KPIs
Days to years, depending upon maturity
Data sources for unified measurement are many and varied and across all platforms, e.g. marketing automation systems, media providers, sales/customer systems and more.
- Links strategy with tactics through a unified measurement framework; insights are holistic and have consistency of meaning, e.g. ROI means the same regardless of whether it is a measure of campaign, channel or whole budget ROI.
- Shared language and meaning organization-wide.
- Captures both the granularity and speed of attribution with the holistic, longer-term view of MMM.
- Often described as the holy grail, it can be expensive, complex, resource intensive and lengthy to implement.
- Difficult to achieve in one step, though with the right plan, value can be delivered along the way to unlock the next phase of development.
Brand Awareness and Perception
Continuous brand tracking
Surveys, panels, focus groups, social listening, network analysis, sentiment analysis, etc.
Brand tracking helps us to understand where our brand sits against KPIs, though it doesn’t influence it.
Sometimes these methods are used as ‘one-off’s, though for larger brands, they are often ‘always on’. Similarly, they can vary between qualitative data (e.g. information gathered in conversational form from focus groups) and quantitative data (e.g. from ongoing survey responses).
Awareness (prompted/unprompted), consideration, ad awareness (prompted/unprompted), penetration, purchase intent, brand desire, top of mind.
Can be continuous or snapshot. Depending on cadence, (e.g. quarterly), may need to build history for two or more years. Insight is then delivered in weeks.
Includes Kantar, YouGov, Ipsos, etc.
- Measures how the brand is perceived, including desirability, and gauges performance against competitors.
- Adds richness and emotional intelligence to your brand story.
- Often limited in relationship to outcomes (e.g. sales) from brand tracking alone. Though can be nested in Marketing Mix Modelling (MMM) – see later – to provide the link.
- Can place too much emphasis on emotions that are not linked to behaviours; what people say they think is not necessarily a driver of what they do.
- Can be limited in statistical validity (sample bias/size for one ‘one off’ or small studies).
Granular, tactical and reactive campaign decisions
Attribution
Last-click, multi-touch (MTA), first-click (rarely)
A prescriptive method that helps with campaign optimizations. Attribution, as we know it today, became popular in the 2010s in digital advertising due to the opportunity to ‘follow’ individuals around the web using cookies. This made it possible to understand the user journey and link paid digital activities, whether search or social (pay per click/PPC/biddable media), to sales. Last-click remains the most popular method, with ‘multi-touch’ (MTA) considered more refined as it assigns a weighting to measurable touch points along the journey.
Conversion metrics, such as a click, website visit or add to cart. Performance KPIs (outcomes that can improve through optimization) include CPC (cost per click), CPV (cost per view), CTR (click-through rate), CVR (conversion rate) , ROAS (Return on Ad Spend), Impressions, Downloads (e.g. for apps), Sales.
While insights are often available within the first day, it typically takes time for campaigns and automated optimizations to stabilize. Ideally 90+ days for history (but no shorter than one week within a campaign) and insights within weeks.
Typically large platform providers such as Google Ads, Google Display Network (GDN), Meta, etc.
- Granular data persists, despite cookie deprecation/ IDFA loss, though not to the extent it once did.
- Rapid feedback allows for intra-campaign optimizations.
- MTA attempts to understand the interaction along the path to purchase in a digital environment.
- Typically limited in channel scope (mainly digital) – i.e. partial channel picture and mis-attribution potential.
- Limited understanding of incremental relationship to outcomes (e.g. sales); some activity may be in the path rather impacting the path.
- Measures of ROI used in attribution may not match broader measures given they relate only to a specific channel/campaign.
Continuous innovation and improvement
Experiments
Consumer- or geo-based tests
Experimentation aims to support taking calculated risks on smaller populations to statistically determine whether a particular lever (e.g. targeting or creative) has a genuine causal impact on an outcome, e.g. sales or enquiry. This is done by comparing response behaviours between a test group, and a control group. Groups could be, for example, customer segments, a set of stores, or regions. The test must be set up properly to ensure the results are statistically valid and learnings can be relied upon when it comes to rolling out ‘winning strategies’.
Sales, CTR (click-through-rate), open-rates, etc.
Insights in weeks, data history required is for the duration of the test (plus previous activity).
Includes sales or other outcome data, the metric of interest (e.g. media spend, store installation etc.). Sometimes combined with broader MMM sources to contextualize and enrich robustness.
- Experiments can be an entry point to building a measurement capability – more focused/less data intensive on a few specific levers.
- Can augment Marketing Mix Modelling (MMM) by filling measurement gaps & pushing the boundaries of MMO optimization potential, resulting in bolder decisions and more agility.
- Helps to focus on learning and innovation and measure new levers; can be a good way to help find new audiences, develop localized tactics or support major shifts in spend with a low-cost approach to a test and learn culture.
- By nature, experiments are hyper-focused and will only provide incrementality for a specific lever/strategy.
- The measured lever needs to be executed such that it is possible to measure incrementality through testing (Note: MMM can measure any past lever).
- While often mitigated, in some fields, it can be difficult to maintain perfect control that is free from contamination due to various external factors.
Budget planning, performance forecasting, strategic direction
Marketing Mix Modelling
Econometrics, econometric modelling
Marketing Mix Modelling (MMM) captures the full range of business drivers, including external factors, to tell the whole story across all marketing activities and the role each plays in delivering an outcome. It measures interactions between channels and the relationship between an outcome (e.g. sales) and influencing factors (both marketing and external, e.g. economic pressures) to understand ‘incrementality’ (outcomes that would otherwise not have occurred) over the long term. That means it can fill the gaps left by methodologies such as attribution.
Sales, revenue, profit, ROI, brand, traffic.
MMM can take months to deliver insight and require 2-3 years of data history.
Uses a multitude of different data sources throughout the funnel, across channels and outcome responses.
- Impactful insights and an understanding of both different marketing techniques and external influences on outcomes that allow you to determing big budget decisions. Introduces common language and begins to break silos.
- Modern methods – especially facilitated by AI/ML complemented by human domain knowledge/subject matter expertise, integrate with more frequent and more granular data – are overcoming previous perception that MMM is old-fashioned, infrequent, backward-looking and laborious.
- MMM can be expensive to implement and can take longer to produce insights.
- Poor definition of measurement requirements in campaign briefing can result in not capturing the data necessary to the desired analysis.
Strategy linked to tactics through measurement
Triangulation/ unified measurement
MMM, attribution, brand tracking, test & control
A mechanism for integrating all analysis into a singular framework to capture a whole picture. The gold standard has MMM at the heart for integration with different ‘lenses’ from other methods supplementing and enhancing the story, according to the question at hand.
Sales, Revenue, Profit, ROI, AOV, etc – encompasses all previously discussed KPIs
Days to years, depending upon maturity
Data sources for unified measurement are many and varied and across all platforms, e.g. marketing automation systems, media providers, sales/customer systems and more.
- Links strategy with tactics through a unified measurement framework; insights are holistic and have consistency of meaning, e.g. ROI means the same regardless of whether it is a measure of campaign, channel or whole budget ROI.
- Shared language and meaning organization-wide.
- Captures both the granularity and speed of attribution with the holistic, longer-term view of MMM.
- Often described as the holy grail, it can be expensive, complex, resource intensive and lengthy to implement.
- Difficult to achieve in one step, though with the right plan, value can be delivered along the way to unlock the next phase of development.
Key measurement methods & triangulation/unification
Key measurement methods & triangulation/unification
Main Challenges of Measurement
Missing or poor-quality data can lead to flawed analysis. Attribution (in particular) only sees a partial channel picture, made more opaque by privacy.
Proxies may be possible but require skillful handling and interpretation. MMM is becoming favoured once more.
Methods that satisfy frequency often miss the whole picture, but lengthy timescales from execution to insight frustrate commercial implementation. The faster you can react to new insight, the faster you realise the benefit.
New approaches to MMM deliver faster, more granular insights to overcome drawbacks of attribution.
For example, ROI can mean many things. Campaign ROI, Channel ROI, Budget ROI are all different and different methods can produce different results, especially if not able to expose incrementality and impact on profit.
Choose measures that can be understood business-wide.
Mathematical models have degrees of performance. Good models can predict well, but only to a certain tolerance, meaning care is needed in real-world application.
The skill is in stakeholder engagement to ensure assumptions and techniques produce the best possible results.
More data, more products, more territories means more complexity both in data engineering and modelling.
Start with key questions or use cases and build from there.
More data, more products, more territories means more complexity both in data engineering and modelling.
Start with key questions or use cases and build from there.
Analytics and business teams often communicate poorly. Commercially- focused measurement experts can understand the business, commission and translate analysis.
Ensuring insight is useful, usable and used depends on the culture of data in the org.
Competitors, consumer behaviour, politics, pandemics. All can have a significant impact, despite best endeavours. Models no longer work and need to be rebuilt.
I&E teams are becoming increasingly agile with more in their toolkit to help overcome.
Measurement Maturity Curve
Measurement Maturity Curve
Basic analytics and data exploration applications.
Not yet scalable/ compliant with IT.
Proof of Value local initiatives; low appropriation overall.
Not yet covering the entire business.
Pionneer department building first platforms.
C-suite supportive. More request from the business.
Strategized data science, with pipeline of use cases and roadmap to serve.
Centralized data/IT investments in data platform.
Centre of excellence with talent strategy, resources and commitment.
Data-driven decision making is the norm at C-suite and all functions.
Data is served in appropriate formats, for all needs, from exploration to operations.
Global governance dissolved silos, common performance monitoring.
Basic analytics and data exploration applications.
Not yet scalable/ compliant with IT.
Proof of Value local initiatives; low appropriation overall.
Not yet covering the entire business.
Pionneer department building first platforms.
C-suite supportive. More request from the business.
Strategized data science, with pipeline of use cases and roadmap to serve.
Centralized data/IT investments in data platform.
Centre of excellence with talent strategy, resources and commitment.
Data-driven decision making is the norm at C-suite and all functions.
Data is served in appropriate formats, for all needs, from exploration to operations.
Global governance dissolved silos, common performance monitoring.
Marketing Data Ecosystem
Turn decisions on and off to reveal data sources
NB this is illustrative/simplified in terms of data sources/complexity of acquiring data. Every organisation and every request of the analysis requires different data items.
Measurement Triangulation Maturity
Used by individual teams with different approaches and different requirements
Working together to try to make sense of different techniques, perhaps adding MMM.
Often consolidated to a single measurement team for a given brand/territory. Attempting to create a holistic view. MMM still separate, if used.
With MMM used as an integration point for all analyses, with different lenses.
There are many paths to each state that will differ from one organisation to another.
Measurement Triangulation Maturity
Used by individual teams with different approaches and different requirements
Working together to try to make sense of different techniques, perhaps adding MMM.
Often consolidated to a single measurement team for a given brand/territory. Attempting to create a holistic view. MMM still separate, if used.
With MMM used as an integration point for all analyses, with different lenses.
There are many paths to each state that will differ from one organisation to another.
Glossary of Major Terms
About Ekimetrics
Ekimetrics is a leader in data science and AI-powered solutions. Since 2006, we’ve pioneered the use of AI and advanced data science applied to unified marketing measurement, holistic business optimization and broad-ranging sustainability goals.
Our goal
From data engineering to analytics, data culture to transformation, our holistic approach to marketing effectiveness has earned us significant recognition.Including Analytics Company of the Year 2023 (British Data Awards) and being named a ‘Leader’ in the Forrester WaveTM: Marketing and Optimization Q3 2023, where we scored top for talent, with 5/5 on 16 different criteria, including Marketing Strategy Consulting and Global Client Management.
Ekimetrics combines cutting-edge tools with bold vision and innovation… providing marketers with next-generation analytics… [and] strong engineering.
Want to know more about how to achieve measurement triangulation?
We’d love to hear from you.