Marketing Analytics
& Attribution
Measurement plans, attribution after GA4, marketing mix modelling, incrementality tests and the KPIs finance actually trusts. This hub explains how to prove what marketing is worth in a privacy-first world, then routes you into guides and analytics software comparisons.
Explore marketing analytics by focus area
Good measurement answers three questions: what happened, why it happened and what would have happened without the marketing. Each focus area below serves one of them.
How a trustworthy measurement system is built
Most analytics problems are definition problems. Teams argue about numbers because nobody agreed what a lead, an active customer or a conversion is. These five steps prevent that.
- Start from business goalsWrite down the two or three outcomes the business cares about this year — revenue, qualified pipeline, retention — and work backwards to marketing KPIs.
- Build a KPI treeBreak each outcome into drivers: traffic, conversion rates, deal size, repeat rate. Every campaign metric should connect to a branch.
- Define and implement eventsDocument each tracked event, its trigger and owner. Test tracking after every site release; broken tags are the most common source of bad data.
- Triangulate methodsUse attribution for daily optimisation, MMM for budget allocation and experiments to settle disputed questions.
- Report decisions, not dataEvery report should end with what will change as a result. A dashboard nobody acts on is a cost, not an asset.
Marketing analytics guides
What marketing analytics is and how to use it, attribution models explained, and how voice and intonation analysis reveals customer emotion.
Marketing Attribution Models Explained: How to Measure What Really Drives Revenue
β‘ TL;DRAttribution models determine which marketing touchpoints get credit for conversions. No single model is correct β each...
What Is Marketing Analytics and How Do You Use It?
A clear guide to marketing analytics β what it covers, the key metrics across channels, how attribution works, and how to turn marketing data into better, more profitable decisions.
Intonation analysis: a powerful tool for understanding customer emotions
Uncover the benefits of intonation analysis in customer relations. Learn how variations in tone, rhythm, and emphasis can reveal customer emotions and enhance personalized responses.
Channel measurement guides elsewhere on Kurums
Attribution, MMM and experiments compared
No single method answers every measurement question. Mature teams run all three and use each for what it does best, a combination often called unified or triangulated measurement. When methods disagree, experiments usually settle the argument because they are the only approach that directly measures cause and effect.
| Method | Question it answers | Data needed | Strength | Weakness |
|---|---|---|---|---|
| Multi-touch attributionUser-level | Which touchpoints preceded conversions? | Tracked user journeys and conversion events | Granular and fast; useful for daily optimisation | Blind to untracked users and offline media; correlation, not causation |
| Marketing mix modellingAggregate | How much did each channel contribute to sales? | Two or more years of weekly spend, sales and outside factors | Privacy-safe; covers TV, out-of-home and offline sales | Slow to update; needs spend variation and statistical skill |
| Incrementality experimentsHoldout and geo tests | What would have happened without this marketing? | A randomised or matched control group | Measures true causal lift | Costs some sales during the test; one question at a time |
Attribution models and what changed after GA4
Universal Analytics stopped processing standard property data in July 2023, and GA4 made data-driven attribution the default. In the same year Google retired the first-click, linear, time-decay and position-based models in GA4 and Google Ads, leaving data-driven and last-click. The older rule-based models still appear in other tools and are worth understanding, because each one encodes an assumption about how buying works.
| Model | How credit is assigned | Bias |
|---|---|---|
| Last click | 100% to the final touchpoint before conversion | Overvalues branded search, retargeting and email |
| First click | 100% to the first recorded touchpoint | Overvalues discovery channels; ignores closing |
| Linear | Equal share to every touchpoint | Treats a passing impression like a demo request |
| Time decay | More credit to touchpoints closer to conversion | Undervalues early brand and content work |
| Position-basedU-shaped | Typically 40% first, 40% last, 20% split across the middle | Arbitrary weights chosen by convention |
| Data-driven | Algorithm compares converting and non-converting paths | Better, but still limited to tracked, consented journeys |
Marketing mix modelling is back — and cheaper
MMM was long the preserve of large consumer-goods advertisers with econometrics consultancies. Two forces brought it back into the mainstream: privacy restrictions that weakened user-level tracking, and open-source tools. Meta released Robyn, Google released Meridian, and libraries such as PyMC-Marketing give in-house analysts a Bayesian modelling framework without licence fees.
A useful model needs about two years of weekly data, meaningful variation in spend across channels, and the outside factors that also move sales: price changes, promotions, seasonality, distribution and competitor activity. The output is an estimate of each channel’s contribution and diminishing returns curves that show where the next dollar works hardest. Calibrate the model with experiment results wherever you have them; an MMM that disagrees with a well-run geo test is usually the one that is wrong.
The KPIs that survive a finance review
Channel metrics such as click-through rate help practitioners optimise, but leadership decisions are made on a short list of commercial measures. Report these consistently, with definitions agreed with finance, and show the channel metrics as drivers underneath them.
| KPI | Formula | What finance reads into it |
|---|---|---|
| Customer acquisition cost | Sales and marketing spend ÷ new customers | Whether growth is getting more or less expensive |
| CAC payback | CAC ÷ monthly gross margin per customer | How long cash is tied up in each new customer |
| LTV : CAC | Lifetime gross margin ÷ CAC | Whether acquisition creates value; around 3:1 is a common benchmark |
| Marketing efficiency ratio | Total revenue ÷ total marketing spend | Program-level return without attribution disputes |
| Marketing-sourced pipeline | Value of opportunities first created by marketing | Marketing’s contribution to future revenue in B2B |
Software for a measurement stack
A measurement stack combines analytics, experimentation and qualitative insight. Our independently researched comparisons cover each category with pricing and fit.
Related Marketing pillars
Marketing analytics questions teams ask
What is marketing analytics?
Marketing analytics is the practice of measuring, connecting and interpreting marketing data so that budget and effort go to what actually drives revenue. It combines web and product analytics, advertising platform data, CRM and sales data and methods such as attribution, marketing mix modelling and controlled experiments. Its purpose is better decisions, not bigger dashboards.
Which attribution model should I use?
For day-to-day optimisation inside digital channels, data-driven attribution is usually the best available option, and it is the default in Google Analytics 4 and Google Ads, which retired the first-click, linear, time-decay and position-based models in 2023. Keep last-click as a simple comparison. No attribution model proves causation, so validate major budget decisions with incrementality tests or marketing mix modelling.
What is marketing mix modelling?
Marketing mix modelling, or MMM, is a statistical method that estimates how much each channel contributed to sales using aggregated weekly data on spend, sales, pricing, seasonality and outside factors. It does not rely on tracking individual users, so it works despite privacy restrictions and covers offline channels such as TV. Open-source tools such as Meta’s Robyn and Google’s Meridian have made it accessible to mid-sized advertisers.
What is an incrementality test?
An incrementality test measures the extra sales caused by marketing by comparing a group exposed to it with a similar group that was not. Common designs are audience holdouts, where a random share of users is excluded from ads or emails, and geo tests, where campaigns are switched off in matched regions. The difference in outcomes is the true incremental effect.
Why do Google Analytics and ad platforms report different conversions?
They count differently. Ad platforms credit conversions to their own ads using their own attribution windows, often including view-through conversions, while analytics tools apply one model across all channels and can only see users who accepted tracking. Consent choices, ad blockers, cross-device journeys and modelled conversions widen the gap. Reconcile both against actual orders or CRM revenue.
Which marketing KPIs matter most to a CFO?
Finance teams focus on customer acquisition cost, CAC payback period, customer lifetime value relative to CAC, the marketing efficiency ratio of total revenue to total marketing spend, and marketing-sourced pipeline and revenue. Present them with consistent definitions over time and show how each campaign metric connects to them.
