Channel Attribution for Deep Links: Know What Works

You're running deep link campaigns across email, social media, paid ads, push notifications, and partnerships. Some are working. Some aren't. Channel attribution tells you which ones deserve more investment and which ones are wasting budget.

Without proper attribution, you're guessing. This guide covers how to set up channel attribution for deep links so every conversion is traceable to its source.

The analytics dashboard with date range selector, filters, charts, and breakdowns.

What Channel Attribution Answers

Channel attribution answers specific operational questions:

• Which channels drive the most installs?
• Which channels drive the most revenue (not just clicks)?
• What's the cost per acquisition by channel?
• Which channels bring users who retain longest?
• Where should we increase spend? Where should we cut?

These questions can't be answered with aggregate metrics. You need per-channel data tied to downstream outcomes.

Attribution Models

Last-Click Attribution

The most common model. The last link a user clicked before converting gets full credit.

Pros: Simple to implement, easy to understand, directly actionable. Cons: Ignores earlier touchpoints that may have influenced the conversion.

When to use: For most apps, last-click is a good default. It credits the channel that directly triggered the conversion, which is the most immediately actionable signal.

First-Click Attribution

The first touchpoint gets full credit. Useful for understanding which channels create initial awareness.

Pros: Values discovery channels that introduce users to your app. Cons: Ignores the channel that actually closed the conversion.

When to use: When you want to understand which channels are best at introducing new audiences to your product.

Linear Attribution

Every touchpoint in the path gets equal credit. If a user clicked three different links before converting, each gets 33% credit.

Pros: Acknowledges the full journey. Cons: Treats all touchpoints equally even when some were clearly more influential.

Time-Decay Attribution

More recent touchpoints get more credit. A click 1 day before conversion gets more credit than a click 14 days before.

Pros: Acknowledges both the full journey and recency. Cons: More complex to implement and explain.

Choosing a Model

Start with last-click. It's the industry default and gives you actionable data immediately. Layer in first-click analysis separately to understand discovery channels. Multi-touch models (linear, time-decay) add value at scale when you have clear multi-step user journeys, but they add complexity that's not worth it for most early-stage apps.

For a deeper dive into attribution mechanics, see Mobile Attribution: A Developer's Guide.

Setting Up Channel Attribution

Step 1: Tag Every Link

Every deep link you distribute should carry channel identification. Use UTM parameters consistently:

Channel	utm_source	utm_medium
Email newsletter	email	email
Facebook organic	facebook	social
Facebook ads	facebook_ads	cpc
Instagram organic	instagram	social
Google Ads	google_ads	cpc
Apple Search Ads	apple_search	cpc
Push notification	push	push
SMS	sms	sms
QR code (print)	qr	qr
Referral program	referral	referral
Partner blog	partner_{name}	referral

If a link doesn't have UTM parameters, it can't be attributed to a channel. Make tagging non-negotiable for every link that leaves your organization.

Step 2: Capture Parameters on Click

Your deep linking platform should capture UTM parameters (and any custom parameters) when a user clicks the link. These are stored alongside the click event and associated with the user's session.

With Tolinku, click events automatically capture all URL parameters. You can view channel breakdowns in your analytics dashboard.

Step 3: Preserve Through Install

For new users who need to install the app, UTM parameters must be preserved through the app store redirect. This is handled by deferred deep linking: the parameters are stored server-side on click and retrieved by the SDK after first app open.

Test this regularly. A broken deferred deep linking flow means all new user installs lose their channel attribution.

Step 4: Connect to Downstream Events

Attribution is only useful if it connects to outcomes. Tie the channel information to:

• Registration events
• First purchase events
• Subscription events
• Retention metrics (D7, D30 by channel)
• LTV calculations by channel

This requires passing the attribution context from the initial link click through to all subsequent analytics events. See Custom Event Tracking for Deep Link Campaigns for implementation details.

Channel Comparison Framework

Once attribution is set up, compare channels across multiple dimensions:

Volume Metrics

Channel	Clicks	Installs	Install Rate
Email	8,000	2,400	30%
Facebook Ads	25,000	3,750	15%
Instagram Organic	12,000	960	8%
Push Notifications	15,000	N/A (existing users)	N/A
Referral	3,000	1,800	60%

Install rate tells you about user intent. Referral links have high install rates because the recommendation carries trust. Social media has lower install rates because users are browsing casually.

Quality Metrics

Channel	D7 Retention	D30 Retention	Avg Revenue (30d)
Email	28%	14%	$4.20
Facebook Ads	18%	8%	$2.10
Instagram Organic	22%	11%	$3.00
Push Notifications	35%	20%	$5.50
Referral	38%	22%	$6.80

Quality metrics reveal which channels bring users who stick around and generate revenue. Referral users almost always retain best because they come with a personal endorsement.

Efficiency Metrics

Channel	Spend	Installs	CPI	Revenue (30d)	ROAS
Email	$500 (tooling)	2,400	$0.21	$10,080	20.2x
Facebook Ads	$11,250	3,750	$3.00	$7,875	0.7x
Google Ads	$8,000	2,000	$4.00	$6,000	0.75x
Referral	$1,800 (rewards)	1,800	$1.00	$12,240	6.8x

ROAS (return on ad spend) is the ultimate efficiency metric. In this example, email and referral dramatically outperform paid ads on efficiency, though paid ads deliver more volume.

Common Attribution Challenges

Cross-Device Attribution

A user might click a link on desktop and install the app on mobile. Without cross-device matching, this install appears organic (no attribution).

Solutions:

• User-level matching: If the user logs in on both devices, you can connect the journeys
• Probabilistic matching: Match based on IP address, device fingerprint, and timing (less accurate)
• Accept the gap: For most apps, cross-device installs are a small percentage

Organic Cannibalization

Some users who click a paid ad or email link would have found your app organically anyway. This means your paid CPI is effectively higher than reported because some of those "attributed" installs weren't truly incremental.

Measuring incrementality properly requires controlled experiments (holdout groups where a subset of users don't see the ad), which is complex. For practical purposes, be aware that paid channels may overstate their contribution.

Dark Social

Links shared privately (WhatsApp, DMs, SMS forwards) often lose their UTM parameters. The recipient sees a bare URL without tracking. This traffic appears as "direct" or "organic" in your analytics.

Mitigation:

• Use short branded links that have tracking built into the redirect, not just URL parameters
• Encourage sharing through your app's built-in share mechanism (which can generate tracked links)
• Accept that some word-of-mouth will always be unmeasurable

Last-Click Bias

Last-click attribution systematically undervalues channels that create awareness (social media, content marketing, PR) and overvalues channels that capture intent (search, email, retargeting).

Mitigation:

• Run separate first-click reports to see which channels start the journey
• Use assisted conversion analysis to credit channels that appear in the path but aren't the last touch
• Consider incrementality testing for large-spend channels

Making Decisions from Attribution Data

Attribution data is only useful if it drives action. Here's how to use it:

Budget Allocation

Shift budget toward channels with the best LTV:CPI ratio. If referral users have 3x the LTV of Facebook ads users at half the CPI, investing more in your referral program is a better use of budget.

Channel Optimization

Within each channel, use attribution to optimize:

• Which creatives drive the highest-quality users (not just the most clicks)?
• Which audience segments convert best?
• What time of day or day of week produces the best results?

Kill Underperformers

If a channel consistently delivers users with poor retention and negative unit economics (LTV < CPI) after reasonable optimization, stop spending there. Not every channel works for every app.

Double Down on Winners

When a channel shows strong efficiency, test scaling it. Increase budget by 20-30% and monitor whether efficiency holds. Some channels scale linearly; others hit diminishing returns quickly.

For a comprehensive view of analytics for deep links, see Deep Link Analytics: Measuring What Matters.