Multi-touch attribution: what it is and how it works

You just watched display ads get cut from next quarter’s budget. They just weren’t converting, according to the company’s most recent report. Meanwhile, the channel that closed the sale took all the credit—as if it did all the work. 

It probably didn’t. It just happened to be the point when the customer finally said yes. 

That’s where most attribution setups quietly fail: the last-touch model gives the win to whoever closes, while first-touch crowns whoever opens. Everything in between—from the social media post that started it all to the marketing email that kept things warm—gets nothing.

Multi-touch attribution models fix that, showing which channels contributed to conversion across the full journey. But there’s one problem: these models only tell you what worked and not why. Experience intelligence shows what visitors from those channels did next—and where their journeys gained or lost momentum.

In this article, you walk away knowing which specific attribution model works for you and how to determine why your channels perform the way they do.

Key insights

• Pick an attribution model—and then stress-test it. Comparing two models side by side often reveals budget decisions a single model would get wrong. 

• Accurately crediting conversions isn’t the finish line. Once you know which channels are contributing, the next question is why they perform that way—and that requires pairing attribution with behavioral data. 

• Look for a platform that closes the loop between the two. Going from ‘which channel contributed’ to ‘why it converted or was abandoned’ in a single workflow saves you from exporting data, switching tools, and losing the thread mid-investigation.

What is multi-touch attribution? 

Multi-touch attribution (MTA) is a model that spreads credit for a conversion across every touchpoint in the customer’s journey—not just the first one, and not just the last.

Here’s what that looks like. Say a customer: 

• Clicks a Google ad on Monday

• Opens a nurture email on Wednesday

• Sees a retargeting ad on Friday

• Converts through direct traffic on Sunday

Under last-touch attribution, direct ads get all the credit. Under multi-touch attribution, all four touchpoints share it, just not always equally—some models weigh certain touchpoints more than others. 

⚠️ What it won’t do for you: MTA solves credit distribution, not cost problems. If you want a true cost per acquisition, you still need to:

• Pull your attributed conversions from your MTA model

• Pair them with the spend data you already track per channel in your ads manager or spreadsheet

• Divide spend by attributed conversions to get the cost per acquisition

Why is multi-touch attribution important?

Multi-touch attribution makes a big difference in accurately deciding which channels to fund. It’s how you ensure you budget more for the channels actually doing the work. Email nurtures, retargeting, social awareness—they toil away in the middle of the journey, but there’s typically nothing to show for it on your dashboard. MTA finally gives them their due.

There’s also a newer reason this matters more than it used to. According to the 2026 Contentsquare Digital Experience Benchmark report, AI-referred traffic is growing +632% year over year. As more discovery happens in AI chat interfaces, customer journeys are getting shorter and blurrier—and touchpoints vanish before they ever hit your marketing analytics. 

You can’t attribute a touchpoint you never saw. And that’s all the more reason to squeeze every ounce of insight out of the ones you can.

How to choose the right attribution model—and calibrate it for your channels

Choosing MTA means selecting a specific model for calculating attribution. The right model depends on your channel mix and how long customers take to convert, and testing more than one is your smartest bet. 

1. Pick a multi-touch attribution model

Each model distributes credit for conversions a little differently. Here’s how three common ones handle it:

• Linear splits credit evenly across every touchpoint. Say a customer clicks a paid social ad, opens two emails, and then converts—each of those touchpoints gets a flat 25% credit. If you’re not sure where to start, this is a simple model that gives you a solid baseline.

• U-shaped, also known as position-based, weights the first and last touchpoints more heavily, with the middle splitting what’s left. Under this model, a marketer might find that display ads—which get just 5% of the budget—earn 18% of attributed conversions, even though it never closes a single deal. Use this model when you value what opens and closes the door—but still want the middle to count for something. 

• Time decay gives more credit the closer a touchpoint sits to conversion. A flash sale is a good example: an early awareness ad might barely register, but the retargeting email you sent an hour before the sale ended gets most of the credit. This works well for short, urgency-driven purchases or for wrapping up a recent campaign.

💡Pro tip: check what models your current customer acquisition analytics tools offer. 

Some platforms have cut back their options—GA4 dropped linear, position-based, and time-decay, leaving just last-click or a black-box model you can’t see into (in other words, it attributes conversion credit to your channels but won’t tell you how they calculated it). 

Contentsquare offers multiple attribution models to choose from, and you can filter your attribution table by channel to drill straight into journey maps, funnels, or session replay for that cohort—so you can see why those numbers look the way they do, without switching tools.

Pick your attribution model in Contentsquare's Attribution Analysis

2. Compare models side by side to catch what one model alone would miss

No single model tells the whole story. The real value often shows up in the gap between two of them. 

This doesn’t mean picking a marketing attribution model was pointless, or that you now need to run every decision through five different lenses. Your primary model is still what you report against from day to day. But you should still run comparisons before a high-stakes call or as a gut check on whether that default is steering you in the right direction. 

Say you’re eyeing Email for a budget cut. Under the last-touch attribution model, it barely shows up, because it’s rarely a ‘closer.’ But run that same data through a linear model, and Email’s attributed conversions jump by 1,000. Last touch wasn’t wrong—it just wasn’t looking at Email’s actual job, which is nurturing people who aren’t ready to buy yet. 

👉 You don’t need two separate reports to catch this. Contentsquare’s Acquisition Analysis lets you compare two models side by side in the same table—with a delta column showing you where they disagree—so you can identify over- and under-crediting channels.

Compare your attribution models in Contentsquare

3. Extend your lookback window for long consideration cycles

A lookback window (sometimes called an attribution window) is the length of time that a model examines when assigning conversion credit to touchpoints. A 7-day window, for example, means only touchpoints from the last week count—anything earlier is invisible to the model.

Short windows quietly punish patience. If your attribution window is only seven days, any touchpoint that happened eight days ago (or earlier) gets no credit at all—even if it’s the real reason someone bought. 

This hits top-of-funnel channels the hardest, since they’re the first domino to fall, not the last. 

Say an ecommerce marketer stretches their lookback window from seven to 30 days. Paid Social’s attributed conversions climb—not because Paid Social suddenly started performing better, but because the model can now see the full 30 days it was working in.

Contentsquare’s Acquisition Analysis makes this adjustment easy. You can configure your lookback window—7, 30, 60, or 90 days—right from the same screen where you’re reviewing attributed conversion and revenue by channel.

Extend your lookback window with the click of a button with Contentsquare

How to know why a channel converted the way it did

Here’s the limit of everything we’ve outlined above: even a perfectly calibrated model only tells you what happened. To figure out why channels convert a certain way, you need to pair attribution with experience intelligence and behavioral data. 

Here’s what that workflow looks like in Contentsquare:

1. Spot the anomaly in Acquisition Analysis, which gives you detailed insights into how users arrive at your website. Organic Search ranks first for attributed conversions, but something’s off—drop-off before purchase is unusually high. 

2. Go to Funnels to spot exactly where the drop happens. Then, follow up with Impact Quantification to see the revenue it costs you at that step.

3. Use Journeys to see visitors’ full paths—where they enter and where their momentum breaks compared to visitors who convert

4. Watch sessions from a user’s perspective with Session Replay and find the actual friction. In this case: a page that loads slowly for SEO traffic (and nowhere else).

Four separate questions—what converted, where they dropped, what their path looked like, and what actually went wrong—are all answered without leaving the platform or exporting a spreadsheet. 🎉

Analyze your unique channel mix in Contentsquare’s Acquisition Analysis—and use it as a jumping-off point for further investigation

📣 Pro tip: Sense Analyst, Contentsquare’s AI agent, will serve as a shortcut to getting your attribution answers. Ask something like, ‘Where do users from paid social drop off?’ in plain language, and you’ll get an answer that’s supported by your data. Then, Sense Analyst will walk you through the next steps—with ideas on how to improve drop-off and boost conversions.

Ask Sense Analyst questions about acquisition—and get answers and next steps for how to optimize your marketing funnel and boost conversions

Make budget calls you can actually defend

Attribution used to mean picking a model and crossing your fingers that it was close enough. Now you can choose a model, stress-test it against another, calibrate how long your customers take to make a decision—and when a channel’s numbers still don’t add up, find out why instead of guessing. 

That’s how you move from ‘here’s what worked’ to ‘here’s what worked, and here’s the behavior behind it.’ Plus, you get some major bonuses: fewer spreadsheets, less switching between platforms, and more definitive data points to slide in front of your CFO when it’s time to make tough budget choices.

FAQs about multi-touch attribution