AppLovin Attribution Explained: Platform ROAS vs Incrementality

AppLovin attribution counts the purchases AppLovin Ads can tie to someone who clicked an ad, or clicked or recently viewed one, inside the day target you optimize for. That number is the engine’s scorecard. It is not proof of new revenue. To know whether AppLovin is incremental for your business, run a holdout and compare the incremental ROAS to your breakeven ROAS, which is 1 divided by your gross margin.

Every channel’s dashboard is built to grade itself. AppLovin’s is no different, and it’s no worse. The job is to read it for what it is, then check it against a number the platform can’t influence.

The short version

Question	Answer
What does AppLovin attribution count?	Conversions tied to an AppLovin click, or (in the “Clicks and views” mode) a click or a view within one day, inside the reporting window. Check which mode your account uses.
How fast do AppLovin conversions happen?	AppLovin reports that “80% of web conversions happen within 24 hours of click.”
Why doesn’t blended MER move as much as platform ROAS?	Overlap, substitution and scale. Under roughly 10% of total spend, a channel won’t move blended MER detectably in a month (WKND rule of thumb).
What’s a good AppLovin ROAS?	Any incremental ROAS above your breakeven. Breakeven ROAS = 1 ÷ gross margin. There is no universal benchmark.
Is AppLovin incremental?	Published results point both ways. Test your own account with a holdout.
What does NC D28 ROAS mean?	New-customer revenue in a 28-day cohort window ÷ spend, as AppLovin Ads reports it. It is still platform attribution.

What AppLovin attribution actually counts

AppLovin Ads learns from the conversion events your AppLovin Pixel (or the Shopify app) sends back. If those events are missing, duplicated or carry the wrong value, every number below is wrong before attribution even starts. Validate the setup first with our AppLovin Shopify integration checklist.

Once the events are clean, three settings shape what you see.

1. Clicks vs Clicks and views

AppLovin’s own documentation disagrees with itself right now. As of WKND’s last check:

• AppLovin’s Analyzing campaign performance resource says attribution is click-based and that AppLovin does not currently attribute on a view-through basis.
• AppLovin’s Support Center Measurement and attribution page says Ads Manager has two reporting modes, Clicks and Clicks and views, and that the second can count a transaction after a click or within one day of viewing an ad. It also says switching the mode changes reporting, not optimization, targeting or billing.

We covered the conflict in How does AppLovin work for advertisers?. The practical rule hasn’t changed:

1. Check which mode is available and active in your account.
2. Write it down next to every report you share.
3. Never compare a “Clicks” period with a “Clicks and views” period.
4. Don’t switch definitions in the middle of a scale decision.

A view-inclusive number will almost always be higher. That doesn’t make it wrong. It makes it a different number.

2. Day 0 vs Day 7 (the day target)

The day target tells the model which conversion window to optimize toward.

• Day 0 covers conversions within the first 24 hours. AppLovin positions it as the starting point for many brands.
• Day 7 optimizes for a longer purchase cycle (AppLovin’s examples include furniture, mattresses and higher-AOV health products). AppLovin’s support documentation described it as a 192-hour window when we checked.
• D14 and D28 can also appear in reporting for longer-term analysis.

The day target is an optimization choice, not just a reporting filter. A high-AOV product forced onto Day 0 can look bad before its natural consideration period has passed.

3. The speed of AppLovin conversions

AppLovin’s consumer-brands page states: “80% of web conversions happen within 24 hours of click” (AppLovin-reported). If you’ve seen a shorter window quoted elsewhere, go by AppLovin’s current wording.

That speed is useful and a little dangerous. Fast, click-adjacent conversions are exactly the ones most likely to overlap with a customer who was already on their way to buy.

What “NC D28 ROAS” means

NC D28 ROAS is how WKND reports AppLovin case results. It’s an AppLovin Ads new-customer cohort metric: revenue from new customers attributed to AppLovin over a 28-day window, divided by spend.

It’s a stricter view than all-customer Day 0 ROAS, because returning buyers are excluded and the window is long enough to catch slower purchases. It is still the platform’s attribution. WKND’s own caveat applies to every one of our cases: “Platform attribution can differ from blended or incrementality measurement, and results vary by account.”

Why AppLovin ROAS runs ahead of blended MER

Blended MER = total revenue ÷ total ad spend across every channel. It’s the number finance sees. When platform ROAS looks great and MER doesn’t move, WKND checks three causes.

Cause	What’s happening	What to check
Overlap	The same customer touched Meta, Google, email and AppLovin. Each platform claims the sale.	Sum of platform-claimed revenue vs actual Shopify revenue
Substitution	AppLovin reached people who would have bought anyway, through another channel. Revenue moved, it didn’t grow.	Did other channels’ conversions drop when AppLovin ramped?
Scale	The channel is too small a share of spend to show up in the blend.	WKND’s rule of thumb: “a channel under roughly 10% of total spend will not move blended MER detectably in a month”

Overlap is sharpest for brands already heavy on Meta. If you’re deciding how to split budget between the two, read AppLovin vs Meta Ads first.

Scale is the one people forget. A channel at 4% of spend could be genuinely incremental and still be invisible in MER. That isn’t a reason to trust the dashboard. It’s a reason to design a test that can detect it.

A worked example (illustrative)

This is WKND’s published illustrative example, not a client result. The numbers are chosen to show the mechanics.

Line	Value
Axon spend	$60,000
Platform ROAS (what Ads Manager shows)	3.2x
Blended MER before AppLovin	2.40x
Blended MER with AppLovin	2.51x
Revenue lift the business actually gained	~$108,000
Incremental ROAS	~1.8x
Breakeven ROAS at 55% gross margin	1.82x

Read it top to bottom:

1. Ads Manager claims about $192,000 in revenue (3.2 × $60,000).
2. The business gained about $108,000 it wouldn’t otherwise have had.
3. So the incremental return is ~1.8x ($108,000 ÷ $60,000).
4. At a 55% gross margin, breakeven is 1 ÷ 0.55 = 1.82x.

The platform said 3.2x. The business made roughly breakeven. Nobody lied. The dashboard answered “which sales did AppLovin touch?” and the business needed “which sales did AppLovin cause?”

In WKND’s decision table (below), this result doesn’t clear the bar to scale. It’s a hold-and-fix case: work on creative and feed, then retest.

What’s a good AppLovin ROAS? Start from breakeven, not a benchmark

Google’s “People also ask” box for applovin roas shows “Is a 2.5 ROAS good?” The honest answer: it depends entirely on your margin, and on whether that 2.5 is platform ROAS or incremental ROAS.

Breakeven ROAS is 1 divided by gross margin.

Gross margin (illustrative)	Breakeven ROAS
30%	3.33x
40%	2.50x
55%	1.82x
70%	1.43x

So a 2.5x ROAS:

• is exactly breakeven at a 40% gross margin;
• is comfortably profitable at 70%;
• loses money at 30%.

And that’s only if the 2.5x is incremental. A 2.5x platform ROAS can be a 1.5x incremental ROAS, as the worked example shows.

Two more cautions:

• Gross margin is the simple version. WKND sets the goal “from contribution margin and CAC,” so if shipping, discounts and returns are heavy, your real breakeven is higher than 1 ÷ gross margin.
• For subscription or repeat-purchase brands, a first-order ROAS below breakeven can still be a good deal if the lifetime value supports it. Decide that before the test, not after.

What about WKND’s case numbers?

WKND publishes three anonymized AppLovin accounts: a jewelry brand at 2.94x NC D28 ROAS (March 4 to 21, 2026), a food and beverage CPG brand at 2.52x NC D28 ROAS (January 24 to February 10, 2026), and a footwear brand at 3.52x NC D28 ROAS in an 18-day validation (August 14 to 31, 2026).

These are specific accounts in specific windows. They are not benchmarks, and they are platform-reported new-customer cohort metrics. A 2.52x can be excellent for one margin structure and a loss for another. The full breakdown is in our AppLovin case studies.

How to run an AppLovin incrementality test

This is WKND’s published method. It’s designed before the test starts, because a test designed after the results come in is just a story.

1. Take a 4-week baseline. Record blended MER, total revenue, new-customer revenue and each channel’s spend before changing anything.
2. Pick the test design by share of spend.

• If AppLovin is 10% or more of total spend: run a geo holdout (AppLovin on in some regions, off in matched regions).
• If it’s under 10%: run an on/off test in 2-week blocks.

1. Run for 2–4 weeks.
2. Hold everything else still. No big Meta budget swings, no sitewide sale, no new launch in the test window. Every other change is noise.
3. Compare incremental ROAS to breakeven. Incremental revenue (test minus control, or on minus off) ÷ AppLovin spend, against 1 ÷ margin.

Then decide with the table:

Incremental ROAS vs breakeven	Decision
≥30% above breakeven	Scale in steps, retest at 2x
Within 30% of breakeven	Hold. Fix creative and feed, then retest
Below breakeven	Cut or restructure

Using the illustrative 55% margin: breakeven is 1.82x, so the scale bar is about 2.37x incremental (1.82 × 1.3).

A few things that break these tests:

• Switching attribution mode mid-test. Lock Clicks or Clicks and views before day one.
• Too little spend in the test cell. If the expected lift is smaller than your normal week-to-week revenue swing, you won’t see it.
• Reading it early. A Day 7 product needs the full window.
• Promotions. A sale during the test will swamp the signal.

Is AppLovin incremental? What the published evidence says

Search applovin incrementality and Google shows evidence on both sides. As observed on September 27, 2026 (titles and claims as shown in Google’s results, not independently verified):

AppLovin-published cases (AppLovin-reported):

• HexClad: AppLovin’s partner page lists $1M+ incremental revenue in 3 weeks.
• IPSY: “IPSY drove a double-digit incremental lift in new subscriptions.”
• MaryRuth’s: an AppLovin case on testing incremental impact across DTC and Amazon.

Agency and measurement-firm views:

• Common Thread Collective (CTC): “AppLovin Discovery Campaigns Deliver 284% Incrementality After Four Months of Testing“.
• Haus: “Is AppLovin More Than a Hype Channel?”
• bluealpha: “Cann: AppLovin’s 2-16x ROAS Tested, Zero Incremental Lift.”

Read together, the point is simple. AppLovin reports meaningful lift for some brands. Another brand’s test, as bluealpha’s title describes it, found none despite strong platform ROAS. AppLovin-published cases are first-party evidence and are selected by AppLovin. Third-party write-ups come with their own methods and incentives.

None of it tells you what AppLovin will do in your account, with your margin, creative and channel mix. That’s why the answer to “is AppLovin incremental?” is a test, not a citation.

The same goes for claims you’ll see on agency sites, such as CPAs 20–40% lower than Meta. They may be true for someone. They’re unverified for you until you run a holdout.

Third-party measurement: Triple Whale, Northbeam, Fairing

Search applovin attribution and several measurement vendors show up next to AppLovin’s own docs. They answer different questions.

• Triple Whale and Northbeam are among the measurement partners AppLovin lists as already integrated for web advertisers, alongside WorkMagic, Measured, Prescient, Converge, Hyros, Daasity, RedTrack, Attribuly, Elevar and Fospha (AppLovin third-party integrations). AppLovin can append a supported partner’s UTMs across Creative Sets. Triple Whale’s article “Does AppLovin Actually Drive Revenue?” ranks for applovin attribution.
• Fairing, an attribution-survey tool (“how did you hear about us?”), also appears in the applovin attribution results with “How to Analyze AppLovin Data in Fairing)”.

What these tools can do: give you one consistent attribution model across every channel, so AppLovin isn’t grading itself on different rules than Meta. What they can’t do: prove causation. A multi-touch model is still a model. Post-purchase surveys are self-reported and under-credit channels people don’t remember.

Different systems use different identity rules, windows, time zones and view-through treatment, so some discrepancy is normal. The management job is to know why the numbers differ and keep one framework. Use a third-party tool for daily steering. Use a holdout for the budget decision.

The SEC and short-seller headlines: what they mean for your numbers

Some of the anxiety around AppLovin measurement comes from the company’s headlines, not from any single account. Reuters reported on February 20, 2026 that an SEC probe involving AppLovin’s data practices remained active, following a whistleblower complaint and short-seller reports. Reuters also said the SEC had not formally accused AppLovin or its leadership of wrongdoing. AppLovin disputes the allegations in pending securities litigation. The fuller due-diligence picture is in Is AppLovin legit and safe?.

For an advertiser, you don’t need to settle the investment debate. The practical answer is the same one you’d use for any channel: a holdout. If AppLovin creates customers who wouldn’t have bought anyway, the test will show it, whatever the headlines say. If it doesn’t, the dashboard won’t save it. As WKND puts it, “no channel earns scaled budget on its own reporting, Axon included.”

How WKND handles this

Tracking, attribution and incrementality is a named WKND service for AppLovin Ads, not an add-on. Across $5M+ in Axon spend on 50+ accounts, two of the eight situations that prompt the call are measurement problems:

• “Platform ROAS looks great. Blended MER didn’t move.” “The first thing we check is whether a holdout has ever been run.”
• “The CFO is asking if any of it is incremental.” The answer we give: “An incrementality read designed before the test starts is.”

What that looks like in practice:

1. Event health first. Purchase value, item data, transaction IDs, AppLovin click parameters and server-event deduplication are validated before anyone reads ROAS.
2. One documented attribution definition. Clicks or Clicks and views, the day target, and the partner tool, written down and held constant.
3. Goals set from economics. The ROAS goal comes from contribution margin and CAC, so breakeven is known before launch.
4. The test designed up front. Baseline, geo holdout or on/off, the window, and the decision thresholds agreed before spend moves.
5. Cases reported with caveats. Our published AppLovin results (2.94x, 2.52x and 3.52x NC D28 ROAS) are anonymized, labeled with their brand type and dates, and carry the caveat that platform attribution can differ from blended or incrementality measurement.

If you want a partner that ties AppLovin’s numbers back to your P&L, see our AppLovin tracking, attribution and incrementality work.

To run the holdout with us, see AppLovin incrementality testing.

FAQ

How does AppLovin attribution work?

AppLovin Ads attributes a purchase when its Pixel or Shopify integration records a conversion from someone who clicked an AppLovin ad or, in the “Clicks and views” reporting mode, clicked or viewed one within one day. Reporting is shown by day target (Day 0, Day 7) and longer windows such as D28. AppLovin’s documentation currently conflicts on view-through, so confirm the active mode in your account.

Is AppLovin incremental?

Sometimes. AppLovin-published cases such as HexClad and IPSY report incremental lift, while bluealpha’s result “Cann: AppLovin’s 2-16x ROAS Tested, Zero Incremental Lift” describes a test that found none. Results vary by account, margin, creative and channel mix. A geo holdout or on/off test in your own account is the only reliable answer.

Is a 2.5 ROAS good on AppLovin?

Only if it’s above your breakeven. Breakeven ROAS is 1 divided by gross margin, so 2.5x is breakeven at 40% margin, profitable at 55%, and a loss at 30%. It also has to be incremental ROAS. A 2.5x platform ROAS can be much lower once overlap and substitution are removed.

Why is my AppLovin ROAS higher than my blended MER?

Usually overlap (several platforms claiming the same sale), substitution (AppLovin reaching buyers who would have purchased anyway) or scale (AppLovin is too small a share of spend to move the blend). WKND’s rule of thumb is that a channel under roughly 10% of total spend won’t move blended MER detectably in a month.

How do I run an AppLovin incrementality test?

Take a 4-week baseline. If AppLovin is 10% or more of spend, run a geo holdout; if less, run on/off tests in 2-week blocks. Run for 2–4 weeks, hold every other channel and promotion steady, then compare incremental ROAS to breakeven. Scale in steps if it’s at least 30% above breakeven.

What does NC D28 ROAS mean in AppLovin?

It’s an AppLovin Ads new-customer cohort metric: revenue from new customers attributed to AppLovin over 28 days, divided by spend. It excludes returning buyers, so it’s stricter than all-customer ROAS, but it’s still platform attribution and can differ from blended or incrementality results.

Does AppLovin work with Northbeam or Triple Whale?

Yes. AppLovin lists Northbeam and Triple Whale among its third-party measurement partners for web advertisers and can append a partner’s UTMs across Creative Sets. These tools give you one consistent attribution model across channels. They don’t prove causation, so pair them with a holdout for budget decisions.

Get a straight answer on your AppLovin numbers

If your AppLovin ROAS looks better than your P&L feels, the gap has a cause, and it can be measured. WKND designs the test before the spend moves, reads it against your breakeven, and tells you whether to scale, hold or cut.

HAVE US RUN AN INCREMENTALITY READ →

Or start broader: GET AN AXON ACCOUNT REVIEW → GOAL HISTORY · FEED AND EVENT HEALTH · CREATIVE FATIGUE · INCREMENTALITY READINESS. WRITTEN FINDINGS INSIDE FIVE BUSINESS DAYS. See /applovin-axon-services/.

Reporting shown is anonymized. ROAS figures are AppLovin Ads new-customer cohort metrics from the stated reporting windows. Platform attribution can differ from blended or incrementality measurement, and results vary by account.

Sources

1. AppLovin Ads for consumer brands, AppLovin (80% within 24 hours of click, AppLovin-reported). Read 2026-09-27.
2. Measurement and attribution, AppLovin Support (Clicks / Clicks and views; D0, D7, D14, D28). Via WKND’s verified article, 2026-09-06.
3. Analyzing campaign performance, AppLovin (click-based attribution statement). Via WKND’s verified article, 2026-09-06.
4. Third-party integrations, AppLovin Support (Northbeam, Triple Whale and other partners).
5. AppLovin Partner program, AppLovin (HexClad result, AppLovin-reported). Read 2026-09-27.
6. IPSY drove a double-digit incremental lift in new subscriptions, AppLovin.
7. SEC probe involving AppLovin still active, Bloomberg News reports, Reuters, February 20, 2026.
8. AppLovin Axon Ads Manager Agency, WKND (measurement framework, worked example, decision table, case results). Read 2026-09-27.
9. Google results for applovin attribution, applovin incrementality and applovin roas, observed signed out, hl=en gl=us, 2026-09-27 and rechecked 2026-09-28: Triple Whale, Fairing), Haus, CTC, HexClad, MaryRuth’s.