Attribution is not only a reporting setting. It is the rule your business uses to decide which channels, campaigns and customer touchpoints deserve credit for revenue or leads. If that rule is too simple, the wrong channel can look more valuable than it is. If it is too complex for your data quality, the model can create confidence without evidence.
The practical decision is not which attribution model sounds most advanced. The better question is: which model matches your conversion type, buying cycle, channel mix, click and view-through windows, and event data volume?
This guide helps you compare attribution methods in a business context, understand click and view-through windows, and choose a data setup that can support the model you want to use.
What Attribution Really Decides
Attribution assigns conversion credit across the path a customer takes before they buy, submit a lead form, start a WhatsApp conversation, complete a secure portal form, or move into a B2B sales process.
In a simple ecommerce journey, the final paid search click before purchase may deserve most of the credit. In a longer B2B journey, the first organic article, a remarketing ad, a sales email and a later direct visit may all influence the decision. Attribution is the method used to turn that messy journey into a reporting rule.
Once you treat attribution as a credit rule, the next question is time. A touchpoint can only receive credit if it falls inside the window your platform or analytics setup is willing to count. That is why click windows and view-through conversions come before model selection.
Click Windows and View-Through Conversions
What is a click window?
A click window is the period after someone clicks an ad during which a conversion can still be credited to that ad. Common platform windows include short windows such as seven days and longer windows such as 30 days, but the right window depends on the buying cycle.
What is a view-through conversion?
A view-through conversion happens when a person sees an ad, does not click it, and later converts inside the selected view-through window. This matters because exposure can influence behavior even when there is no click.
Where click and view-through windows become blind
Ad platforms usually understand their own ad views and ad clicks best. Website analytics tools usually understand website visits, events and conversions best. That creates a measurement gap.
- Ad platforms can overstate their own role because they may not fully credit organic search, email, direct traffic or content touchpoints.
- Website analytics can undercount view-through influence because ad impressions are often not passed into the analytics journey.
- Remarketing campaigns can claim credit from people who were already likely to convert, especially when view-through windows are generous.
- Offline or semi-offline conversions, such as WhatsApp sales and B2B pipeline movement, can break the journey unless CRM or first-party identifiers reconnect it.
These windows define the boundary of what can be credited. But they do not tell you whether the conversion itself is simple, delayed, offline, assisted by sales, or hidden behind a secure portal. Before comparing attribution methods, the business has to define the conversion path it is actually measuring.
Start With the Conversion Type
| Conversion type | Practical attribution approach |
|---|---|
| Online ecommerce | Use last click as a starter for short cycles. Move toward time decay or data-driven attribution when spend, repeat visits and channel mix increase. |
| B2B sales | Use first click, position based or time decay when awareness, nurturing and sales follow-up all matter. Connect leads to CRM stages before trusting revenue attribution. |
| Contact us forms | Use a simple model when volume is low. Add position based analysis when forms are influenced by content, brand search, retargeting and direct visits. |
| WhatsApp conversions | Treat the click-to-chat as a lead event, then reconcile outcomes manually or through CRM. Avoid relying only on platform-reported conversions. |
| Secure portals with long forms | Prioritize first-party IDs, server-side events where appropriate, and clean funnel events. Long forms often need more than last click. |
After the conversion type is clear, the attribution model becomes a business-fit decision instead of a theoretical analytics choice. The same model that works for a short ecommerce purchase can mislead a B2B sales motion or a WhatsApp-led conversion journey.
Compare Attribution Methods
Use each model for three decisions only: what it is, who should adopt it, and who should avoid it.
Last click
What it is: Gives 100% credit to the final touch before conversion.
Adopt for short-cycle ecommerce, simple lead forms, early-stage ad accounts and businesses without strong attribution assumptions.
Avoid for established brands, long journeys, heavy organic traffic, and businesses spending enough on multiple channels that last-touch bias can distort budget.
First click
What it is: Gives 100% credit to the first known touch.
Adopt for top-of-funnel programs, long consideration journeys and businesses that need to value the channel that introduced the customer.
Avoid for activation campaigns, short purchase cycles and teams needing fast daily optimization feedback.
Linear
What it is: Splits credit equally across all known touches.
Adopt for complex journeys with several smaller channels and no clear evidence for weighting one touch above another.
Avoid when one or two channels dominate, the team cannot optimize budget across channels, or journey tracking is incomplete.
Time decay
What it is: Gives more credit to recent touches and less to older touches.
Adopt for longer buying cycles, multi-channel programs and teams moving beyond last click as spend rises.
Avoid for low-spend or low-complexity programs where the output will look similar to last click without adding much decision value.
Position based
What it is: Weights touches by journey position, often emphasizing first and last touch.
Adopt for larger programs with long journeys, multiple channels and enough analytics capacity to defend the weighting logic.
Avoid when the team lacks a strong point of view, analysis capacity or validation process.
Data driven
What it is: Uses algorithmic analysis to assign credit based on observed conversion patterns.
Adopt for high-spend, high-volume, complex channel mixes with analyst support for testing and validation.
Avoid for small accounts, one-channel businesses, or teams that cannot audit whether the model is over-crediting smaller channels.
The model descriptions show the trade-offs, but they are easier to apply when they are tied to familiar business situations. The examples below translate the method choice into practical scenarios without naming any company.
Business Examples Without Naming the Company
Short-cycle ecommerce
A direct-to-consumer retailer sells low-consideration products. Most paid search buyers purchase within a few days. Last click is acceptable at the start, but time decay becomes useful once content, email and retargeting all influence repeat sessions.
B2B sales
A software company receives demo requests after buyers read comparison content, attend webinars and respond to sales outreach. First click can protect awareness investment, while position based or time decay can better reflect the full journey once CRM stages are connected.
Contact us forms
A services business receives inquiries from paid search, organic pages and brand visits. Last click may be simple enough at low volume, but it can undervalue content that created demand before the final branded search.
WhatsApp conversion flow
A campaign sends prospects into WhatsApp where sales happen later. The click-to-chat is not the full conversion. The business should reconcile conversation quality and closed outcomes before shifting budget.
Secure portal with long forms
A finance or insurance journey may include login, document upload and later approval. Attribution needs clean event design and first-party identifiers because session breaks can hide the true path.
These examples show why attribution is not only a marketing-reporting question. Every example also creates a data question: can the business store enough events, join them across touchpoints, and process them reliably enough to support the chosen model?
Match Storage and Processing to Event Volume
| Event volume | Recommended processing posture |
|---|---|
| Up to 1 million events/month | A smaller ecommerce or lead-generation site can often rely on platform reports, clean tagging and periodic exports. |
| 1 million to 60 million events/month | A growing multi-channel business should consider warehouse exports, scheduled joins and documented channel rules. |
| 60 million to 600 million events/month | A large portal, marketplace or high-traffic publisher needs governed pipelines, quality monitoring and repeatable transformations. |
| More than 600 million events/month | An enterprise-scale platform needs distributed processing, cost controls, data contracts and ongoing model validation. |
The rule is simple: do not upgrade the attribution model faster than the data system can store, join and audit customer journeys. A sophisticated model built on broken identity, missing events or inconsistent channel rules can mislead the business faster than a simple model.
With the model and data-readiness limits visible, the final decision becomes more practical: choose the simplest model that reflects the journey accurately enough to guide budget, then improve the data system before adding more sophistication.
A Practical Decision Framework
- Use last click when the journey is short, the business is early in paid media, and the team needs a simple default.
- Use first click when awareness channels are under-valued and the purchase cycle is long.
- Use linear when many smaller channels assist conversion and there is no defensible weighting logic yet.
- Use time decay when recent touchpoints matter but earlier nurturing should still receive credit.
- Use position based when first and last touches both matter and the business can justify the weighting.
- Use data driven only when the channel mix, event volume and analytics capability can support testing and validation.
FAQs
What is the best attribution method for ecommerce?
For simple, short-cycle ecommerce, last click can be a practical starting point. As channels expand, time decay or data-driven attribution may give a more balanced view.
Should B2B companies use last-click attribution?
Only with caution. B2B journeys often include organic content, email, retargeting, sales follow-up and direct visits, so last click can hide earlier influence.
Why do ad platform conversions and analytics conversions not match?
They use different visibility, attribution windows and rules. Ad platforms may see impressions and clicks that analytics tools do not, while analytics tools may see site behavior that ad platforms ignore.
Are view-through conversions reliable?
They can be useful, but they need strict windows and validation. Overly generous view-through settings can over-credit ads, especially remarketing campaigns.
When should a business move to data-driven attribution?
Move only when the business has enough channel complexity, event volume and analytics support to validate the model and act on the results.
Attribution should make budget decisions more honest. Before you change the model, ask yourself: which part of your customer journey is currently getting too much credit, and which part is quietly creating demand that your reports cannot see? For discussing this use case in detail, connect with Nitesh Shrivastava on LinkedIn.
About Nitesh Shrivastava
Nitesh Shrivastava is the Head of SEO & Analytics at GrowthOps Asia and a featured speaker at the Google Search Central Deep Dive.
Leveraging his engineering background, 12 years of experience, and a business degree from Nanyang Business School, Nitesh translates complex martech and search engineering into clear business strategy. He is currently experimenting with Agentic AI to automate enterprise SEO and performance marketing workflows to help brands scale organic growth without losing the human touch.