How to find why direct gets last-click credit, step by step

To find why direct gets last-click credit, check GA4's lookback window first. Then read Direct's row in the Attribution models report, which only counts paths that were direct from start to finish. If you sell on Shopify, put Last click beside Last non-direct click: the gap is typed visits that closed another channel's journey.

Step by step

Use one full month in GA4 and Shopify, so both tools describe the same buyers. Changing GA4's attribution settings needs the Marketer role or above on the property.

1. Read how long GA4 remembers a source. Open the attribution settings and read Key event lookback window. Purchases default to 90 days, with 30 or 60 as the other choices. Google says the same window applies to session attribution, so a short one sends more returning visits to Direct in every report. Menu path: Admin > Data display > Events > Attribution settings. Google's newer help page calls the link Key events attribution.
2. Note the reporting model, then stop worrying about it. On the same page, read Reporting attribution model. Every GA4 model skips direct visits on mixed paths, so none of them can hand Direct a sale another channel touched. A change here also rewrites past reports, so make it rarely. Menu path: Admin > Data display > Events > Attribution settings > Reporting attribution model.
3. Read Direct's row in Attribution models. Pick purchase in the key event menu at the top left. Set one Attribution model (non-direct) column to Paid and organic last click and the other to Data-driven. Direct should read nearly the same in both, because GA4 credits it only on all-direct paths and on key events with no path data. Menu path: Advertising > Attribution > Attribution models.
4. See how fast those buyers bought. Open Attribution paths, pick purchase, and in Path length choose equal to, enter 1 next to touchpoint(s) and click Apply. Read Purchase revenue and Days to key event on the rows that show only Direct. A low day count means GA4 first met these buyers on the day they bought. Menu path: Advertising > Attribution > Attribution paths > Path length. Google's report page calls it Key event attribution paths, under a Key events drop-down.
5. Check who GA4 can follow across devices. Read your reporting identity: Blended and Observed use a user ID when you send one, while Device based ignores it. Then open Traffic acquisition, click + Create new in the top right and compare on Signed in with user ID. No signed-in users means each new device starts its own all-direct journey. Menu path: Admin > Data display > Reporting identity, then Reports > Acquisition > Traffic acquisition.
6. Put Shopify's two last clicks side by side. Filter Reports to the Marketing category, open Performance by referring channel and select Last click and Last non-direct click in the Attribution menu. Direct's drop between the two columns is typed final visits that followed another channel. That part of Direct is a setting, not a mystery. Menu path: Shopify admin > Analytics > Reports > Category: Marketing > Performance by referring channel > Attribution.
7. Read the orders behind Direct. Open recent orders and read each Conversion summary: order count, visits in the last 30 days and the first visit's origin. Click View full sessions to see the referral of every visit. Sort the buyers into regulars and strangers. Menu path: Shopify admin > Orders > order number > Conversion summary > View conversion details.

A worked example

Start with one store's export, which shows the pattern step 3 looks for. On its Channels sheet, Direct holds 57.7% of revenue under last click, first click and the touched view alike. When all three views agree, the journeys that touched Direct started and ended there, so no model can move that share.

Step 4 would find the same shape there. On its Journeys sheet, journeys with 1 touch hold 79.5% of revenue, with 0.5 days to buy. Most of that store's revenue reached GA4 as a single visit, about half a day before the purchase.

Now your own month, for illustration. Say your Performance by referring channel report shows Direct at €18,000 under Last click and €11,000 under Last non-direct click. In this worked example, €7,000 of Direct was typed final visits after another channel, and step 6 hands it back.

For illustration, say GA4's Attribution models report shows Direct at €9,000 under both models. In this worked example, that €9,000 is revenue from buyers GA4 only ever saw arrive without a source.

Step 7 sorts out who those buyers were. Say you read twenty recent orders that ended in a direct visit. Fourteen are repeat orders whose first visit, months earlier, came from search or email. Six are first orders with a single visit.

The fourteen regulars are Direct by habit, and no tag would change that. The six strangers are the open question. Check whether they came through a link you sent without tags, and whether GA4 had a user ID for any of them. Whatever stays unexplained after that is honest Direct: real demand with no visible trail.

What should you check when the report looks wrong?

• Direct-only paths run many touches over many days. GA4 knew these visitors all along and never once saw a source. Links you send without tags are the first suspect, before loyalty.
• Direct wobbles a little between the two model columns. GA4 can reattribute conversions for up to 7 days after they happen. Compare full months that ended more than a week ago.
• Direct includes sales with no visit at all. If an app or a data import sends purchases to GA4, they carry no path data. The Attribution models report files those under Direct, whatever the buyer did.
• Shopify's Direct is far bigger than GA4's. That is expected. Shopify's Last click counts direct visits, GA4 skips them on mixed paths, and each tool keeps its own cookies and its own memory.
• Direct moved the week you changed the window. Lookback changes apply only going forward. Compare months on the same setting, or note the change date on your report.

What to do this week

1. Split Direct into setting and memory. Run step 6 for last month and write down Direct under both Shopify models. Pass: Direct shrinks a lot under Last non-direct click, so much of it is a model choice; report weekly on Last non-direct click. Fail: it barely moves, so Shopify saw no earlier click either, and steps 4 and 7 matter more.
2. Sort twenty Direct orders. Use step 7 on recent orders whose last visit had no referrer. Pass: most are repeat orders with an older first visit, so Direct is mostly your regulars. Fail: most are first orders with one visit, so new buyers arrive without a trail; fix tags and user IDs before you judge any channel.
3. Lengthen GA4's memory if it is short. If step 1 showed 30 or 60 days, set purchases to 90 and note the date. Pass: the setting reads 90 days and the date sits on your report, so next month compares cleanly. Fail: you leave it short, and returning visitors keep falling back to Direct.

Check the homework. Your GA4 Attribution paths export already holds the evidence. Causality Engine reads that one file and shows what each channel caused next to what last-click gave it, in 1 to 2 minutes, for €99 once (excluding VAT), refundable within 30 days. Check the homework

Sources, 1 October 2026: Select attribution settings (Google); Change the reporting attribution model for key events (Google); Get started with attribution (Google); Key event attribution models report (Google); Key event attribution paths report (Google); Reporting identity (Google); Measure activity across platforms with User-ID (Google); Apply comparisons to reports (Google); User acquisition report vs. Traffic acquisition report (Google); Traffic acquisition report (Google); Marketing reports (Shopify); Viewing order conversion summary (Shopify); Changes to sessions and conversion rate in Shopify Analytics (Shopify); About attribution reports (Google).

Related answers

• All Direct traffic questions
• Why does direct get last-click credit?
• Why does direct get last-click credit in a pet store?
• The quarterly attribution checkup