How to Tell Which Channel Actually Brought the Sale

You sold 40 desk lamps last month. Four things ran: an Instagram reel, a newsletter, a WhatsApp broadcast and a small ad budget. Your analytics report hands the win to search. Your gut says the reel did it.

Give every channel its own tagged short link and count the clicks separately. Marketing attribution with short links will not give you the whole truth, but it gives you one real number per channel. The single-touch models inside your analytics tool do not.

Those models lie, and they lie in opposite directions.

What one buyer's path really looks like

Call her Ana. She bought one lamp. Here is how she got there.

• Day 1. She sees your reel on Instagram. She taps through, looks, and leaves.
• Day 6. Your newsletter lands. She clicks, puts the lamp in her basket, then gets pulled away.
• Day 12. A friend sends her the lamp in a WhatsApp message. She opens it and reads the reviews.
• Day 14. She types your brand name into Google, clicks the first result, and buys.

Four touches, one sale. Now hand out the credit.

First touch gives everything to Instagram. Last touch gives everything to the brand search.

Both answers are tidy. Both are wrong.

Why does first touch flatter the wrong channel?

First touch is a simple rule. Whatever someone clicked first gets 100% of the sale.

It rewards discovery, which sounds fair. It also rewards noise.

A reel seen by 50,000 people introduces a lot of them. Most were never going to buy. First touch still books the ones who did against the reel.

So the reel looks like your best channel. You pour money in. The newsletter that closed those sales looks weak, so you cut it. Then the reel stops working, and nobody can say why.

First touch answers one question — what started this? It never answers what finished it.

Why does last touch flatter the wrong channel?

Last touch is the opposite rule. Whatever someone clicked last gets 100%.

This is where budgets quietly die.

Think about what sits nearest a purchase. A branded search. A saved tab. A reminder email. Ana searched your brand name, so search takes the sale.

Search persuaded her of nothing. She already knew the name. The reel taught it to her, and the newsletter kept the lamp in her head.

Last touch tips the doorman and ignores the person who booked the trip.

Google Analytics 4 adds a twist here. Every model skips direct visits, unless the whole path is direct. So if Ana had typed your address in by hand, GA4 would look past that visit and credit whatever came before.

Which attribution models can you still pick in GA4?

Three. That is the entire list.

Google's own help page names them: data-driven attribution, paid and organic last click, and Google paid channels last click.

Plenty of articles still tell you to switch GA4 to first click, linear, time decay or position-based. You cannot. Google removed all four in November 2023, and says so on both of its attribution pages.

Data-driven is the only one of the three that is not a last-click rule. Google says it uses your account's own data, and weighs both converting and non-converting paths. It beats a single-touch rule. It is also a closed box, and a small site may not feed it enough data to say much.

One more surprise. Changing your model does not change the Traffic acquisition report. Google says session-scoped fields, such as Session source, are unaffected.

How far back does GA4 look for a touch?

There is a setting for that, called the key event lookback window. It decides how old a touch can be and still earn credit. Google's attribution settings page gives the choices:

Ana's path ran 14 days, so it fits inside either default. A slow business purchase often does not.

Do not mix this up with a newer setting that sounds similar. In August 2026 Google added custom conversion windows under Advertising. Click-through windows now take any whole number from 1 to 90 days, and engaged-view windows from 1 to 30. Those govern conversions shared with Google Ads. They are not the reporting window above.

Where do short links fit into marketing attribution?

Here is the cheap part, and it needs no data team and no budget.

Give each channel its own link. One for the reel, one for the newsletter, one for the WhatsApp broadcast. Tag each one with UTM parameters, the small labels on the end of a web address that tell analytics where a visit came from.

Now every channel has a count of its own. No model decides it. It is not the whole story, but it is real.

Three rules keep it honest.

1. utm_medium picks your GA4 channel. GA4 matches that value against a fixed list of rules. Use email, social, cpc, referral, affiliate, sms, display, audio or organic and the visit lands in a named channel. Invent something like newsletter or qr and it will usually match nothing at all. The visit then falls into Unassigned. Nothing breaks. The number just sits in the wrong row.
2. Case matters. Google's campaign URL guide states that parameter values are case sensitive, so utm_source=google is not the same as utm_source=Google. Two spellings means two rows, with your numbers split across them.
3. Agree the names once, then write them down. A shared sheet beats memory. It is the only thing standing between you and eleven spellings of Instagram.

Two GA4 changes from 2026 alter what your channel report looks like. Google's release notes carry both.

Chatbot visits got their own AI Assistant channel on 13 May. Tidier source names followed on 11 June. Neither one needs anything from you, and neither one rescues a medium you made up.

Open your channel report and look for the AI Assistant row. If your last channel list had no such row, that list was out of date.

clik.so keeps each campaign's links in folders and tags, and its click analytics counts every link on its own.

What a link tool cannot tell you

This is the limit, and most articles skip past it.

A link tool sees the click. It does not see the sale.

It can tell you a link was opened, but not whether that person bought anything. Only your website, shop or checkout knows that.

Joining the two ends is the destination's job. Your UTM tags travel with the visitor, and your shop software files the order under a source.

So the ladder looks like this:

1. Clicks per channel. Free, no code, works anywhere you can paste a link.
2. Sessions per channel. Needs UTM tags and analytics running on the site.
3. Orders per channel. Needs your shop or CRM to hold on to the tags through checkout.

You can stop at step two and still decide better than you did before.

Expect one shock on the way up. Your click count will usually sit above your session count, and the gap has plenty of honest causes.

Questions people ask

Try it on one campaign

Pick your next campaign and give each channel a link of its own. After one run you will hold numbers that owe nothing to a model, and you will know which reports to stop trusting.

clik.so's free plan makes 50 links a month and tracks 2,000 clicks, with one custom domain and no card needed. If the tagging is the part you are unsure about, the plain-English UTM guide has a naming convention you can copy.