Marketing Attribution

Marketing attribution is the discipline that splits the credit for a conversion among the different touchpoints a customer went through before buying. Its central question is simple but elusive: what really made that sale happen? Since a typical buyer sees an ad, opens an email, searches for the brand on Google and comes back through WhatsApp before closing, the challenge is deciding how much credit each step deserves instead of rewarding only the last click.

In practice, attribution is a unified data job: it requires connecting ad spend, web behavior, the CRM and revenue in one place for the calculation to make sense. That is why it is part of what an analytics layer like Metrix models and reports, where marketing touches are cross-referenced with closed opportunities to show which channel moves the needle and which one only consumes budget.

Unlike a standalone metric such as CTR, attribution does not measure an isolated channel: it measures the full chain that ends in revenue, which is why it is the basis for calculating ROAS and CAC honestly.

Why it matters

Without attribution, budget decisions are made blindly or, worse, on last-click intuition. The problem is that the last touchpoint is almost never the one that created the demand: often it is simply the one that was there when the customer had already decided. If a company in Buenos Aires allocates its spend by looking only at the final conversion, it usually overfunds branded search (people who already knew the brand) and underfunds the content or campaign that sparked the interest weeks earlier. Attribution exists to correct that bias and show each channel's real contribution along the journey.

How it works

The process has three pieces. First, the touchpoints are tracked (ad clicks, email opens, visits to the landing page, WhatsApp conversations) and tied to the same person or account using identifiers such as cookies, UTM parameters or the email address. Second, those touches are linked to a conversion recorded in the CRM (a qualified lead, an opportunity or a sale). Third, an attribution model is applied that splits the credit according to a defined rule.

The most common models are:

• Last click: all the credit goes to the final touchpoint. Simple, but it ignores the discovery phase.
• First click: all the credit goes to the touch that opened the journey. It rewards demand generation, but ignores what closed the deal.
• Linear: splits the credit equally among all the touches.
• Time decay: gives more weight to the contacts closest to the conversion.
• U-shaped (position-based): concentrates the credit on the first and last touch and spreads the rest across the middle.
• Data-driven: uses statistical or machine learning models to infer the real weight of each touch from historical data, instead of a fixed rule.

Single-touch vs multi-touch

The big divide is between single-touch and multi-touch attribution. It is worth being clear on it before choosing a model:

In an Argentine B2B business with sales cycles of several months (consulting, software, industrial equipment), multi-touch is almost mandatory: a single sale can involve ten or fifteen interactions, and attributing it to a single click completely distorts where the value is being created.

A concrete example

A management software company in Córdoba invests in LinkedIn Ads, blog content and email campaigns. With last-click attribution, 70% of sales showed up as coming "from" branded searches on Google, so the team almost cut the content. When they switched to a multi-touch model, they discovered that the blog was the first touch for 60% of won opportunities: people read an article, searched for the brand weeks later and only then converted. Cutting the blog would have dried up the top of the conversion funnel. Attribution did not change sales; it changed the understanding of what produced them.

Common mistakes

• Blindly trusting the last click because it is the figure the ad platform provides by default.
• Not unifying the data: measuring each channel in its own tool leads to double counting (Google, Meta and email each claim the same sale).
• Ignoring channels with no trackable click (WhatsApp, phone calls, word of mouth), which carry a lot of weight in Latin America.
• Changing models without warning: a migration from last-click to multi-touch reorders the entire report and, if nobody explains it, it looks like a channel "dropped".

Done well, attribution stops being a debate of opinions and becomes the objective basis for deciding where to put your next dollar of investment.