First touch vs last touch attribution, for founders who just want to know what works

A customer pays you $49 a month. Where did they come from? Your answer depends on which visit you look at, and most tools quietly pick one for you. That choice is your attribution model, and for a small company it decides which channels look like they work.

One customer, four visits

Here is a common path to a paying customer:

Ask four models who gets the $49 and you get four answers.

The models, in plain English

First touch: what found the customer

First touch answers "where do new customers discover us?". That is the question an early product needs answered, because discovery is the hard part. Nobody can search your name or click your retargeting ad until something introduced you.

• Good for: choosing which communities, content and launches to invest in.
• Weak spot: it ignores everything that happened afterwards, and if the first visit was dark social it gets credited to Direct.

Last touch: what closed the customer

Last touch answers "what pushed them to buy?". It is the default in a lot of tools because it is the easiest to compute: look at the session where the conversion happened.

• Good for: judging pricing pages, onboarding emails and retargeting, which work on people who already know you.
• Weak spot: it systematically over-credits branded search, Direct and retargeting. Those are where people end up once they have decided, not where they came from.

What Google Analytics 4 does

GA4 attributes key events with a data-driven model by default, and offers last click as the alternative. In 2023 Google removed the first-click, linear, time-decay and position-based models from GA4. Data-driven attribution is reasonable at large volume, but with a few dozen conversions a month it has little to learn from, and you can't easily see why it credited what it did.

GA4 still records each user's first source as First user source and First user default channel group. Use those dimensions in an exploration next to your sign-up or purchase events for a first-touch view.

What a small SaaS should use

1. Lead with first touch for decisions about growth: which channels bring people who become customers.
2. Check last non-direct touch for decisions about conversion: what brings people back to buy.
3. Skip the fancy models until you have hundreds of conversions a month. With small numbers, two simple views you understand beat one opaque one.
4. Store the first source yourself. When someone signs up, save the source of their first visit on the user record. Then revenue from your billing system can always be grouped by it, whatever your analytics tool does.

// on the first visit, in the browser
if (!localStorage.first_touch)
  localStorage.first_touch = JSON.stringify({ ref: document.referrer, utm: location.search, at: Date.now() })

// on signup, send it with the form and store it on the user
user.first_touch = JSON.parse(localStorage.first_touch ?? "null")

The takeaway

Attribution models don't find the truth; they answer different questions. Early on the question that matters most is how customers find you, so look at first touch first. Then remember that Direct in either model mostly means "someone shared it privately", not "they already knew us" (more on that here).