Back to Learning Hub
Attribution•9 min read•November 8, 2026

Linear Attribution: How It Works, Examples and Limits

Linear attribution splits credit equally across every touchpoint. A worked example, why GA4 dropped it in 2023, and when it helps or misleads a store.

W

Wilmer

Co-founder & CEO

Linear Attribution: How It Works, Examples and Limits

Linear attribution is the model that sounds fairest: every touchpoint before a sale gets an equal share of the credit. That fairness is also its weakness, because it treats a retargeting click the day before purchase the same as the ad that introduced the brand, and Google removed it from GA4 in 2023. This guide explains how linear attribution works, walks through an example, and shows when it is worth looking at and when it misleads.


The short answer: linear attribution splits credit equally, and Arktis shows it next to four other models


Linear attribution divides the credit for each conversion equally among the touchpoints that led to it. With four touchpoints, each gets 25 percent; with N touchpoints, each gets 1 divided by N. Google announced in April 2023 that linear, along with first click, time decay and position-based, was going away in both Google Ads and GA4, and removed the four from GA4 properties from mid-October 2023. Adobe Analytics and HubSpot still include a linear model.


For an online store that wants to see linear credit beside the alternatives, Arktis is the clear winner. Its Models view compares first-touch, last-touch, linear, time-decay (7-day half-life) and position-based (40, 20, 40) credit per ad platform, counting conversions, with lookbacks of 7 to 90 days, so you can see how much the choice of model changes the answer. Plans are $49, $149 and $349 a month, and Growth has a 7-day free trial.


ModelHow credit is splitMetaTikTokGoogle
First-touchAll to the first touchpoint100%0%0%
Last-touchAll to the last touchpoint100%0%0%
LinearEqual shares, 1 divided by N50%25%25%
Time-decay, 7-day half-lifeA touch 7 days earlier counts half as much50.7%19.8%29.5%
Position-based, 40, 20, 4040% first, 40% last, 20% shared by the middle80%10%10%

The shares are for the invented four-click journey in the worked example below: Meta, then TikTok, then Google, then Meta again.


Key takeaways


Linear attribution is simple to explain and hard to game, which makes it a good baseline and a poor verdict. It ignores order and timing, so a channel that appears often collects more credit than one that appears once, whatever each did. GA4 now offers data-driven attribution and two last click models, so a store that wants linear numbers needs another tool or its own calculation. The most useful thing linear tells you is how far it disagrees with first-touch and last-touch: where they agree, the model hardly matters, and where they differ, the choice of model is driving the answer.


How linear attribution works


Every model starts from a conversion path, the list of recorded touchpoints that came before a conversion. Linear gives each touchpoint the same fraction of the credit. If the conversion is an order, the fraction applies to its value too: an order of $120 with four touchpoints gives $30 to each.


The shares are then added up per channel. A channel that appears twice in the path gets two shares, so a path of Meta, TikTok, Google, Meta gives Meta half the credit even though the model treats every touchpoint the same. Over many conversions the result is fractional: a channel might end the month with 41.5 conversions under linear, against 30 under first-touch and 52 under last-touch.


Adobe's documentation describes linear the same way, as equal credit to every touch point seen leading up to a conversion, and, with a 30-day visitor lookback, gives the example of a $50 purchase split across paid search, social, email and display at $12.50 each.


A worked example


Take an invented shopper. On day 1 they click a Meta ad and browse. On day 4 they click a TikTok ad and look at one product. On day 8 they search for the brand and click a Google ad. On day 10 they click a Meta retargeting ad and buy for $120.


Under linear, each of the four clicks gets 25 percent, $30. Meta has two of them, so it gets $60, and TikTok and Google get $30 each. Under first-touch and last-touch, Meta gets all $120, because the path starts and ends with Meta. Position-based, 40, 20, 40, gives the first and last clicks 40 percent each and splits the remaining 20 percent between the two middle clicks, so Meta gets 80 percent, $96, and TikTok and Google get $12 each.


Time-decay with a 7-day half-life weights each click by how recent it was: a click 7 days before the purchase counts half as much as one on the day. The four clicks were 9, 6, 2 and 0 days before the purchase, which gives shares of 14.7, 19.8, 29.5 and 35.9 percent. Meta's two clicks add up to 50.7 percent, about $61, Google gets 29.5 percent, about $35, and TikTok 19.8 percent, about $24.


Linear is the model that gives TikTok the most credit here. Whether TikTok deserved $30 of that order or $12 is exactly the question no rule can answer.


Where GA4 and other platforms stand


Google announced on 6 April 2023 that first click, linear, time decay and position-based models were going away across Google Ads and Google Analytics 4. It said fewer than 3 percent of Google Ads web conversions used them. For Google Ads, it said it would stop letting advertisers select them for new conversion actions from June 2023, and from September 2023 switch conversion actions still using them to data-driven attribution, with last click available instead.


For GA4, Google's release notes dated 17 October 2023 said the four models would be removed from all properties in mid-October 2023, from both the interface and the Admin API, with affected properties moved to paid and organic data-driven attribution. GA4's attribution help now lists three models: data-driven, paid and organic last click, and Google paid channels last click. If an article tells you to switch GA4 to linear, it was written before October 2023.


Other tools still offer it. Adobe Analytics includes linear among its attribution models, and HubSpot's attribution reports include a linear model that credits each interaction in the conversion path equally. Ad platforms' own reports are a different matter: each reports conversions for its own ads under its own attribution settings, without reference to other platforms' clicks. Our guide to the attribution window covers those settings.


When linear attribution is useful


Linear is a good baseline. Because it does not favour the start or the end of the path, it shows which channels appear in converting journeys at all, which first-touch and last-touch both hide. A channel that is almost never first or last but often in the middle, such as a mid-funnel video campaign, gets visible credit under linear and none under the single-touch models.


It is also useful as a check on the other models. If a channel's credit is similar under first-touch, last-touch and linear, its role is clear and the model choice does not matter. If its credit swings widely, the number you report depends on the model, and a budget decision should not rest on any single one of them.


Linear works best where journeys are short and touchpoints are few, two or three clicks, because then equal shares are not far from any other reasonable split.


When linear attribution misleads


Equal credit is not equal influence. The model has no idea whether a touchpoint changed anything, so a retargeting ad shown to someone who had already decided to buy gets the same share as the ad that introduced them to the brand.


Frequency inflates credit. Every repeated click from the same channel adds a share, so a channel that retargets heavily collects more linear credit simply by appearing more often. In our example, two more Meta retargeting clicks before the purchase would make six touchpoints, and Meta's linear share would rise from 50 percent to four sixths, about 67 percent, without anything else changing.


The path is only what the tool can record. Views without clicks, visits on another device, untagged links and conversations offline are missing, so the touchpoints that are recorded share credit that partly belongs to ones that were not. And no rule-based model, linear included, measures what an ad caused; for that you need a test with a control group, which our guide to incrementality testing explains.


How Arktis compares linear with the other models


The Models view on Arktis's Ads & Attribution page puts the five rule-based models side by side, with one row per ad platform and each cell the conversions credited to that platform under that model. You choose a lookback of 7, 14, 30, 60 or 90 days, and the same number sets the period: the view counts the conversions of the last 7 to 90 days. A platform is tagged as divergent when its counts across the models differ by more than five conversions, which is the cue to look harder before moving budget.


The touchpoints in Arktis are visits that arrived with a click identifier. The tracker captures click identifiers on landing, such as fbclid from Meta, gclid from Google and ttclid from TikTok, and holds them against the visitor across sessions. For linear, each conversion's credit is split equally across the touchpoints recorded for that visitor within the lookback before the conversion, and the shares are added up per platform. A visit from organic search, a direct visit or an email link does not get a share, except a Mailchimp link, whose mc_cid and mc_eid Arktis treats as click identifiers, and a conversion only appears in the Models view when the visit it was matched to itself arrived with a click identifier, so a buyer who clicked an ad and later came back directly to buy is not counted there.


Conversions come from orders matched back to a visit: for Shopify, on email where the store has identified the visitor, otherwise on the UTM source of a session from the previous 24 hours, which can be another visitor's session. An order counts for an ad platform only when it was matched to a visitor who arrived with that platform's click identifier.


Where Arktis does not fit


Because Arktis's touchpoints are visits that arrived with a click identifier, its linear model answers a narrower question than a cross-channel tool such as Adobe Analytics: how credit is shared between ad platforms, plus Mailchimp email links, rather than across every channel, organic search and other email included. It works per platform rather than per campaign in the Models view. Visitors are identified per browser, so a click on a phone and an order on a laptop are not joined, and Arktis does not know who saw an ad without clicking it. There is no bulk import of Shopify orders placed before the store connects. If you need a full multi-channel path across every source, a tool built around that, or your own warehouse model, is the better fit.


Compare the models on your own orders


To see linear credit next to first-touch, last-touch, time-decay and position-based on your store's own conversions, start the 7-day Growth trial, or compare plans on the pricing page. To try the models on numbers of your own first, use the free attribution model comparison tool, and read marketing attribution models explained and last click attribution for the models on either side.


Sources

Google Ads Help: first click, linear, time decay and position-based attribution models are going away, the announcement, the under 3 percent figure and the June and September 2023 steps, published 6 April 2023

Google Analytics Help: what's new archive, the 17 October 2023 note on removing the four models from all properties in mid-October 2023, accessed 2 November 2026

Google Analytics Help: get started with attribution, the three models now available in GA4, accessed 2 November 2026

Adobe Experience League: attribution models and lookback windows, the linear model and the $50 example, accessed 2 November 2026

HubSpot Knowledge Base: understand attribution reporting, the linear model in HubSpot's attribution reports, accessed 2 November 2026

Frequently Asked Questions

What is linear attribution?

Linear attribution is a rule-based model that divides the credit for a conversion equally among all the touchpoints before it. With four touchpoints each gets 25 percent, and with N touchpoints each gets 1 divided by N. A channel that appears more than once in the path receives a share for each appearance.

Does GA4 still have linear attribution?

No. Google's release notes dated 17 October 2023 said first click, linear, time decay and position-based models would be removed from all GA4 properties in mid-October 2023, from the interface and the Admin API. GA4 now offers data-driven attribution, paid and organic last click, and Google paid channels last click.

What is the difference between linear and position-based attribution?

Linear gives every touchpoint the same share. Position-based, often set at 40, 20, 40, gives 40 percent to the first touchpoint, 40 percent to the last and splits the remaining 20 percent among those in between. In a four-touch path, linear gives each touch 25 percent, while position-based gives the two middle touches 10 percent each.

When should I use linear attribution?

Use it as a baseline and a check rather than as the final answer. It shows which channels appear in converting journeys at all, including those that are rarely first or last, and comparing it with first-touch and last-touch shows how much the model choice changes a channel's credit. It is least reliable when journeys are long and one channel retargets heavily, because repeated touches inflate its share.

How does Arktis calculate linear attribution?

In its Models view, Arktis splits each conversion equally across the touchpoints recorded for that visitor, visits that arrived with a click identifier, within the chosen lookback of 7 to 90 days, then adds the shares up per ad platform; a conversion is included only when the visit it was matched to arrived with a click identifier. It shows the result next to first-touch, last-touch, time-decay with a 7-day half-life and position-based 40, 20, 40, counting conversions. Visits from organic search, direct traffic or email do not receive a share, except Mailchimp email links, whose mc_cid and mc_eid Arktis treats as click identifiers.

Try Arktis free for 7 days

Attribution tracking, session recordings, and AI automation in one platform. Set up in 2 minutes, free for 7 days.

Start 7-day free trial

Written by

W

Wilmer

Co-founder & CEO

Wilmer leads product strategy at Arktis, focusing on privacy-first analytics and attribution tracking for e-commerce brands.