Metric Definition
Crediting the channels that convert
Track from
Attribution modeling
Attribution modeling is the method used to assign credit for a conversion across the marketing touchpoints a customer interacted with before they bought. It turns a list of clicks, opens, and visits into a defensible answer to the question of which channels actually drove revenue. Different models split that credit in different ways, so the model you choose changes which channels look effective.
8 min read
What is attribution modeling?
Attribution modeling is the method used to assign credit for a conversion across the marketing touchpoints a customer interacted with before they bought. A buyer rarely converts on a single click. They might see a paid social ad, read a blog post weeks later, click a retargeting ad, and finally convert from a branded search. Attribution decides how much of that sale each step earns.
The reason it matters is budget. If you credit the whole sale to the last click, branded search and retargeting look like your best channels and the awareness channels that started the journey look worthless. Cut those, and the pipeline dries up a quarter later. Attribution modeling exists to give you a consistent, comparable way to value every channel so you can move spend with evidence rather than instinct.
Definition note
Attribution is a model, not a measurement. No model observes true causality directly, it applies a rule for splitting credit. Treat the output as a decision aid you can defend, not a fact, and be explicit about which model produced any number you report.
How to calculate attribution modeling
There is no single attribution formula, because the formula is the model. Every model takes the same raw input, the ordered list of touchpoints per converting customer, and applies a weighting rule. The credit a channel receives is the sum of the weights it earns across all the journeys it appeared in, multiplied through to the conversions or revenue you are crediting.
The practical work is less about arithmetic and more about getting clean inputs. You need a stitched view of each customer journey, a defined conversion window, and a chosen model. Once those three are fixed, the credit split falls out mechanically.
- 1
Stitch the customer journey
Join every touchpoint to a single identity across sessions and devices. Without identity stitching, each touchpoint looks like a separate person and the model breaks.
- 2
Set the conversion window
Decide how far back a touchpoint can sit and still earn credit, for example 30 or 90 days. Touchpoints outside the window are excluded.
- 3
Choose the model
Pick the weighting rule: first touch, last touch, linear, time decay, position based, or data driven. Each splits the credit differently.
- 4
Apply weights and sum
Assign each touchpoint its weight, multiply by the conversion value, then sum per channel to get total credited conversions or revenue.
Attribution modeling in a metric tree
Attribution is most useful when you treat the credited revenue from each channel as a node you can decompose, not a single headline number. The credit a channel earns is a product of how many journeys it touched, where it sat in those journeys, and the value of the conversions it helped close. Breaking it down this way shows you whether a channel is winning because it reaches volume, because it sits at high-value moments, or simply because your model favours its position.
A metric tree makes the chain of cause and effect visible. When attributed revenue drops, you can walk down the branches to see whether touchpoint volume fell, the conversion window changed, or a model assumption shifted, rather than guessing.
Metric tree insight
A channel can show rising attributed revenue purely because you switched from last touch to linear, not because it performed better. KPI Tree decomposes attributed revenue into volume, weighting, and value so you can tell a real performance shift from a model artefact, and it assigns RACI ownership so the marketer who owns the channel is the one informed when its credit moves.
Attribution modeling benchmarks
There is no universal benchmark for an attribution number, because the right model depends on your sales cycle and channel mix. What you can benchmark is how much the credit shifts between models. A useful rule of thumb is that moving from last touch to a multi-touch model typically reallocates between 20 and 40 percent of credit away from bottom-of-funnel channels and towards awareness channels.
The table below sets out where common models fit so you can pick a starting point rather than chase a single target figure.
| Model | Credit logic | Best suited to |
|---|---|---|
| Last touch | 100 percent to the final touchpoint | Short cycles, direct-response, low-touch buying |
| First touch | 100 percent to the opening touchpoint | Demand generation and awareness measurement |
| Linear | Equal credit across all touchpoints | Longer cycles where every step matters roughly equally |
| Time decay | More credit to touchpoints nearer the conversion | Considered B2B purchases with a clear closing phase |
How to improve attribution modeling
Improving attribution is rarely about finding a cleverer model. It is about trusting the inputs and the decision the model feeds. The gains come from cleaner identity data, an honest conversion window, and validating the model against real spend changes rather than accepting it on faith.
Fix identity stitching first
Most attribution error is data error. Invest in joining touchpoints to one identity before debating models. A clean journey beats a sophisticated model on broken data.
Compare models side by side
Run the same period through two or three models and look at where credit moves. The disagreement between models tells you which channels are most sensitive to your assumptions.
Validate against spend tests
Run a holdout or a geo test, cut a channel, and watch conversions. If real demand barely moves, the credit the model gave that channel was overstated.
Account for tracking loss
Consent rejection and cross-device gaps remove touchpoints from view. Estimate the blind spot so you do not credit visible channels for journeys you simply could not see.
Common mistakes when tracking attribution modeling
- 1
Treating one model as the truth
Reporting a single model as fact hides how fragile the number is. Always state the model and ideally show how credit shifts under an alternative.
- 2
Ignoring the conversion window
A window that is too short drops early touchpoints and over-credits the close. Set the window to match your real sales cycle, not a default.
- 3
Double counting across tools
Two platforms each claiming the same conversion inflates results. Reconcile to a single source of truth before you allocate budget.
- 4
Optimising to the model, not the business
Chasing the number a model rewards can starve channels that do not get credit but still drive demand. Sense-check with incrementality tests.
Related metrics
Conversion rate
CVR
Marketing Metrics
Metric Definition
Conversion Rate = (Number of Conversions / Total Visitors or Leads) × 100
Conversion rate measures the percentage of visitors, users, or leads who take a desired action, such as making a purchase, signing up for a trial, or submitting a form. It is the fundamental metric for evaluating the effectiveness of any acquisition funnel, landing page, or marketing campaign.
Return on ad spend
ROAS
Marketing MetricsMetric Definition
ROAS = Revenue from Ads / Ad Spend
Return on ad spend measures the revenue generated for every pound spent on advertising. It is the primary profitability metric for paid media, telling you whether your ad campaigns are generating more revenue than they cost and by how much.
Cost per acquisition
CPA
Marketing MetricsMetric Definition
CPA = Total Campaign Cost / Number of Acquisitions
Cost per acquisition measures the total cost to acquire a single converting user, whether that conversion is a purchase, sign-up, or lead. CPA is the bottom-line efficiency metric for paid marketing, connecting ad spend to actual business outcomes rather than intermediate metrics like clicks or impressions.
Customer acquisition cost
CAC
SaaS Metrics
Metric Definition
CAC = Total Sales & Marketing Spend / Number of New Customers Acquired
Customer acquisition cost (CAC) is the total cost of acquiring a new customer, including all sales and marketing expenses divided by the number of new customers gained in a given period. It is one of the most important unit economics metrics for any growth-stage business.
Customer acquisition cost: a metric tree approach
Metric Definition
Attribution modeling decides which channels get credit for conversions, so it feeds directly into how you decompose and lower acquisition cost across those channels.
Metric trees for marketing teams
Metric Definition
This guide shows marketing teams how attribution modeling sits within a wider tree of channel and conversion metrics they own.
Build attribution as a tree, not a single number
Model attributed revenue as a decomposition of touchpoint volume, credit weighting, and conversion value, with a named owner on every channel branch. KPI Tree pushes the change to the accountable marketer when a channel credit moves and verifies whether the budget shift actually moved demand.
