It depends on the question you're asking. First-touch attribution gives 100% of the credit to the channel that introduced the customer; last-touch gives 100% to the channel that closed the sale. Neither is "correct" — each is a biased lens, and for a short-cycle store the two are often the same click. Use first-touch to judge discovery channels, last-touch to judge closers, and trust blended CAC and MER for the actual budget call.

You run ads. Meta claims the sale, Google claims the same sale, and your Shopify report crowns a third channel. Underneath that mess sit two opposite rules for handing out credit — first touch and last touch — and most "best attribution model" articles never tell you what to do when they disagree. This one does, with the numbers a store owner actually works with.

First touch vs last touch attribution: the core difference

Both models answer one question — "which touchpoint gets credit for this order?" — with opposite rules.

First-touch (first-click) assigns all the credit to the first recorded interaction a shopper had with your brand: the prospecting ad they saw, the organic Reel, the blog post they landed on. It answers "which channels find new customers?"

Last-touch (last-click) assigns all the credit to the final interaction before the purchase: the branded search, the retargeting ad, the email click. It answers "which channels close the sale?"

The trap is that each model credits one moment and erases everything else. A channel that wins on last-touch — retargeting, branded search, email — usually loses on first-touch. A channel that wins on first-touch — prospecting, organic social, a podcast mention — usually loses on last-touch. Pick the wrong lens and you defund the exact channel that feeds the whole machine.

For deeper background on how credit flows across a journey, the attribution and tracking guide walks the full set of models; this page stays on the two that matter most.

A worked example: one order, two completely different stories

Say a shopper meets your store like this:

  1. Sees your Instagram prospecting ad (never clicks, keeps scrolling).
  2. Three days later, a retargeting ad reminds them — they click, browse, don't buy.
  3. A week after that, they Google your brand name, click the branded-search ad, and buy a $48 hoodie.

First-touch credits Instagram prospecting with the whole order. Last-touch credits branded search with the whole order. Same sale, same $48, two channels that look like heroes depending on the lens.

Now add the profit math, because credit without margin is a vanity metric. Say that hoodie costs you $22 from Printify, and Shopify plus payment fees run about $2.30 on a $48 order. Your gross margin before ads is $48 − $22 − $2.30 = $23.70.

Here's the budget mistake. Your last-touch report shows branded search "driving" the sale at a gorgeous return, so you pour budget into branded search. But branded search only harvested demand that the Instagram prospecting ad created — nobody Googles a brand they've never heard of. Cut prospecting and the branded-search "winner" slowly starves, because no new names are entering the funnel. First-touch would have shown you that. Last-touch hid it.

When first touch and last touch are the same click

Here's what the B2B attribution articles miss: for a lot of print-on-demand and impulse products, the whole journey is see ad → click → buy in one session. When that happens, the first touch and the last touch are literally the same event, and model choice changes nothing.

Model choice matters in proportion to journey length. One commonly cited practitioner figure puts the average ecommerce path at eight to ten touchpoints — but that's an average across considered purchases. A $19 novelty mug bought off a single TikTok-driven session has one touchpoint, and first-touch, last-touch, linear, and data-driven all return the identical answer.

So before you agonize over models, ask how long your real buying journey is. Short and impulsive? The model barely matters — spend your energy on clean UTM tracking instead. Long and considered (higher AOV, comparison shopping, repeat visits)? Then the gap between first-touch and last-touch is real money, and you need both lenses.

The budget mistake each model quietly causes

Each model has a known bias. Writers owe you the bias, because you make spending decisions through whichever lens you happen to be looking through.

  • Last-touch over-credits closers. It makes retargeting and branded search look like your best channels because they sit closest to the "buy" button — which is also why it has driven more bad budget decisions than any other model. It's the default in most tools because it's the easiest to compute.
  • First-touch over-credits discovery. It flatters the top of the funnel and ignores whatever actually closed the sale, so lean on it alone and you'll under-fund retargeting and email that genuinely convert.

The honest move is to run both and read them against each other. Where a channel ranks high on both first-touch and last-touch, trust it. Where it's huge on last-touch but invisible on first-touch (classic branded search), treat it as a harvester, not a creator. The marketing attribution strategy guide turns this into a repeatable monthly read.

Why Shopify and GA4 don't use either one by default

You can't just "turn on" first-touch everywhere, because your two main tools default to neither.

Shopify defaults to last non-direct click over a thirty-day window — a last-touch variant that skips "Direct" visits and walks credit back to the most recent non-direct channel (WeltPixel; Order Legend). An order only shows as "Direct" when no real referrer survives the journey.

GA4 removed first-click (along with linear, time-decay, and position-based) in November 2023 and made data-driven attribution the default — you literally cannot select first-click in GA4 anymore (Google Analytics Help). Data-driven attribution splits one order fractionally across channels, but it needs volume to be stable; a commonly cited threshold is roughly three hundred conversions a month before its output steadies.

So "first-touch vs last-touch" is partly an academic debate — your platforms have already chosen. The practical skill is reading each tool knowing which lens it's using, and never comparing one platform's self-credited number against another's.

The tracking under both models is leaking

Both first-touch and last-touch depend on the browser remembering the click, and that memory is decaying.

Apple's iOS 17 Link Tracking Protection strips ad-click IDs like fbclid and gclid from links opened in Mail, Messages, and Safari Private Browsing — though your manually tagged UTM parameters survive (Klaviyo). That matters because Safari runs roughly fifty-five percent of US mobile sessions. On top of that, App Tracking Transparency opt-in sat at an industry average of about thirty-five percent as of the second quarter of last year, which gutted click-based visibility into iOS buyers.

Two takeaways. First, disciplined manual UTMs are now your durable attribution layer — build them consistently with a UTM link generator and carry them into email with Klaviyo UTM tracking. Second, a post-purchase "how did you hear about us?" survey catches the zero-click discovery (podcasts, word of mouth, organic social) that no click model can see — single-question confirmation-page surveys pull forty to sixty percent response rates, making them a real dataset, not anecdote.

The number neither model can fake

Here's the guardrail that sits above the whole first-touch-vs-last-touch argument: blended numbers, computed from your real sales and your real invoices, cannot be gamed by any attribution model.

  • Blended CAC = total marketing spend ÷ total new customers.
  • MER (blended ROAS) = total revenue ÷ total ad spend.

Say you spend $2,800/month on Meta and $1,200 on Google, and land 340 new-customer orders. Blended CAC = $4,000 ÷ 340 = $11.76 to acquire a customer. If your per-order margin before ads is $23.70, you're clearing about $11.94 per new order after acquisition — model debates aside.

The power move: if your Meta and Google dashboards both "improve" while that blended CAC and MER stay flat, no new profit was created — credit just got reshuffled between models. Blended reality is the anti-self-deception layer.

How to actually use first touch and last touch

  1. Fix UTMs first. Most "mystery Direct" traffic is an untagged-link problem, not a measurement mystery. It's free to fix.
  2. Give each tool one job. Shopify tells you how many orders and how much revenue. First-touch tells you which channels find customers. Last-touch tells you which close them. The survey tells you what clicks can't see.
  3. Triangulate, don't model-shop. Where first-touch, last-touch, and the survey agree, trust the ranking. Where they diverge hard, you've found a zero-click channel last-touch is starving.
  4. Let blended CAC and MER settle the budget. If a model change "improves" a channel but blended numbers don't move, nothing real happened.

The slow part is doing this across six platforms every week. That's the job PodVector AI's Victor does. Victor is an AI employee who connects your Shopify store, Meta Ads, Google Ads, and Klaviyo, pulls your true product costs from Printify, Printful, and Gelato, and computes the real per-order profit on every SKU — so he sees every order and every ad dollar together, not as two disconnected reports. When the numbers say a campaign is losing real margin, Victor pauses the ad set, shifts the budget, and delivers the profit report to your Google Drive. He's an AI employee, not a dashboard you have to read.

Put Victor to work on your store's attribution and profit

FAQs

Which is better, first touch or last touch attribution?

Neither — the question is malformed. Each is a biased single-touch lens: first-touch over-credits discovery channels, last-touch over-credits closers. Use first-touch to judge which channels find customers and last-touch to judge which close them, then settle budget with blended CAC and MER, which no model can distort.

Does Shopify use first touch or last touch attribution?

Shopify defaults to last non-direct click over a thirty-day window — a last-touch variant that skips Direct visits and credits the most recent real referrer (WeltPixel). You can switch the model per report, but Shopify does not ingest ad spend, so it can't compute ROAS natively.

Can I still use first-touch attribution in GA4?

No. GA4 removed first-click, linear, time-decay, and position-based models in November 2023 and made data-driven attribution the default (Google Analytics Help). Its remaining options are data-driven, paid-and-organic last click, and Google-paid-channels last click.

Why do first touch and last touch sometimes show the same channel?

Because for short-cycle purchases the journey is one session — see ad, click, buy — so the first touch and the last touch are the same event. Model choice only changes the answer when journeys span multiple touchpoints over days or weeks, which is where the gap between the two becomes real money.

Does iOS 17 break first-touch and last-touch tracking?

It weakens the click-based layer: Link Tracking Protection strips fbclid and gclid IDs in Apple Mail, Messages, and Safari Private Browsing, but your manual UTM parameters survive (Klaviyo). That's exactly why disciplined UTM tagging matters more now, not less — it's the durable layer under both models.

What's the one number I should trust over any attribution model?

Blended CAC (total spend ÷ new customers) and MER (total revenue ÷ total ad spend), both computed from your actual sales and ad invoices. They can't be reshuffled by a model switch, so when platform dashboards improve but blended numbers stay flat, you know credit moved but profit didn't.