Before we show you the delta, here's where every channel actually sits — attribution window and funnel stage, no guessing.
| Channel | Lookback | Funnel |
|---|---|---|
| Google Search Ads | 1-day view, 30-day click | Lower |
| Apple Search Ads | 30-day tap-through | Lower |
| Meta Ads | 1-day view, 7-day click | Mid |
| Snap Ads | 1-day view, 28-day click | Mid |
| 1-day view, 10-day click | Mid | |
| Blog post visit | 1-day view, 30-day click | Top |
| Unattributed (organic, self-driven) visitor | sliding scale | |
| Same session conversion | 1-day view, 1-day click | High-intent, late-stage |
| Conversion within 2 days | 1-day view, 2-day click | High-intent, late-stage |
| Conversion within 3 days | 1-day view, 3-day click | High-intent, late-stage |
| Conversion in 4 to 7 days | 1-day view, 4–7-day click | High-intent, late-stage |
Audit note — New customer acquisition, Remittance USD-INR
Audit note — New customer acquisition, Chat USD-MXN
Let us audit your traffic behavior and map the correct install attribution window, view-through attribution window, and re-engagement attribution window.
Every one of those windows feeds straight into your reported ROAS. Inflate the numerator with overcounted conversions, and your ROAS is wrong by the same margin — not once, but every reporting period.
This is the one that actually produces the overcounting delta. It's the catch-and-release setting: how long a channel gets to hold a lead before releasing it back to the general pool. Hold too long, and a channel keeps claiming credit for someone who already converted through chat-researched, self-driven demand. Release too early, and whoever touches them next takes the win — even if it was never really theirs to claim.
30 days is the common platform default — shown here, not endorsed here. Your own corridor data already told a different story: some of your review-page lags measured in this conversation ran as short as a single day. There's no universal right answer, only the number that matches how your customers actually behave — which is exactly what the audit sets, corridor by corridor.
What platforms claim
Every ad platform runs on a fixed lookback window, set once, for every advertiser, regardless of how their customers actually decide.
Across the platforms and corridors we've measured, the real lag between a click and a conversion sits far closer to zero than any of these defaults — which means a real share of "attributed" credit is being claimed for conversions that would have happened anyway.
What nobody claims
In the businesses we've analyzed, well over half of all downloads and in-app conversions can't be traced to a single ad click at all.
A growing share of that gap starts before anyone opens an app — someone researching a decision in ChatGPT, Claude, or Gemini, then acting on it days later. No ad platform sees that conversation. No MMP tags it. We measure it directly, corridor by corridor, from real conversion data.
Full-service or paid-media-only, we work with agencies the same way we work with advertisers directly — as an independent measurement layer, not competition for the budget. It's the same zero-sum game you already play. You rebalance between Google and Meta every quarter based on what's actually converting. We're not asking you to do something new — we're adding unattributed as one more column in a reallocation you already run.
And it's a capability you can bring to your own client relationships — an independent finding your client's finance team can't dispute, because it didn't come from the agency managing the spend.
Which way the mix should move. What one engagement answers.
How fast you can move it without spooking a board watching this quarter's number.
Whether you're moving fast enough — a competitor already capturing the demand you're still deciding whether to chase makes standing still a decision too.
Direction is a report. Speed and velocity are a relationship — which is why agencies that bring us in once tend to keep us on.
What each approach asks of your team before you get a first answer. No new pixels, no warehouse connection, no waiting for fresh data: we work from the history you already have.
| Task | Typical attribution platform | Mixed Signals Delta Check |
|---|---|---|
| New tracking code (CNAME, pixel) | Required · Dev/IT, Marketing Ops | None |
| Data warehouse connection | Required · Data engineering | None |
| Campaign tracking cleanup | Required · Marketing + agencies | None, we use what you have |
| Ad platform and MMP data | Account authentication | Report exports you already run |
| Conversion data | Ongoing batch file feed | One export of receipts |
| AI-chat signal | Not part of setup | One server log export |
| Waiting for clean data | Yes, while new tracking collects | None, uses your existing history |
| Time to first answer | Setup, then data collection | About 5–6 weeks, end to end |
| Price | Platform subscription, quoted | $2,500, fixed |
Typical setup for pixel-based attribution platforms. Requirements vary by vendor.
One month of attributed and unattributed transaction data $2,500 fixed, agreed before we start.
Data gathering — expectation-managed
2–3 weeks, typically. Getting real transaction data out of a corporate entity is never instant — it depends on your team's turnaround, not ours. We send the reminders. You won't have to chase us.
Analysis — firm commitment
15 business days, starting the day we have your data in hand. This part is entirely in our control, so it's the part we're willing to guarantee.
What you get at the end: your real overcounting delta, in writing, with the math shown. No retainer. No year-long commitment. If the honest answer is "there's nothing here," you'll know that for a small, fixed fee — not a guess you paid a year for.
If you manage paid and organic together and have never gotten a straight answer about which is actually working, that's the conversation.
Before we show you the delta, here's where every channel actually sits — attribution window and funnel stage, no guessing.
| Channel | Lookback | Funnel |
|---|---|---|
| Google Search Ads | 1-day view, 30-day click | Lower |
| Apple Search Ads | 30-day tap-through | Lower |
| Meta Ads | 1-day view, 7-day click | Mid |
| Snap Ads | 1-day view, 28-day click | Mid |
| 1-day view, 10-day click | Mid | |
| Blog post visit | 1-day view, 30-day click | Top |
| Unattributed (organic, self-driven) visitor | sliding scale | |
| Same session conversion | 1-day view, 1-day click | High-intent, late-stage |
| Conversion within 2 days | 1-day view, 2-day click | High-intent, late-stage |
| Conversion within 3 days | 1-day view, 3-day click | High-intent, late-stage |
| Conversion in 4 to 7 days | 1-day view, 4–7-day click | High-intent, late-stage |
Audit note — New customer acquisition, Auto Loan
Audit note — New customer acquisition, USD-MXN
Let us audit your traffic behavior and map the correct install attribution window, view-through attribution window, and re-engagement attribution window.
Every one of those windows feeds straight into your reported ROAS. Inflate the numerator with overcounted conversions, and your ROAS is wrong by the same margin — not once, but every reporting period.
This is the one that actually produces the overcounting delta. It's the catch-and-release setting: how long a channel gets to hold a lead before releasing it back to the general pool. Hold too long, and a channel keeps claiming credit for someone who already converted through chat-researched, self-driven demand. Release too early, and whoever touches them next takes the win — even if it was never really theirs to claim.
30 days is the common platform default — shown here, not endorsed here. Your own corridor data already told a different story: some of your review-page lags measured in this conversation ran as short as a single day. There's no universal right answer, only the number that matches how your customers actually behave — which is exactly what the audit sets, corridor by corridor.
What platforms claim
Every ad platform runs on a fixed lookback window, set once, for every advertiser, regardless of how their customers actually decide.
Across the platforms and corridors we've measured, the real lag between a click and a conversion sits far closer to zero than any of these defaults — which means a real share of "attributed" credit is being claimed for conversions that would have happened anyway.
What nobody claims
In the businesses we've analyzed, well over half of all downloads and in-app conversions can't be traced to a single ad click at all.
A growing share of that gap starts before anyone opens an app — someone researching a decision in ChatGPT, Claude, or Gemini, then acting on it days later. No ad platform sees that conversation. No MMP tags it. We measure it directly, corridor by corridor, from real conversion data.
Full-service or paid-media-only, we work with agencies the same way we work with advertisers directly — as an independent measurement layer, not competition for the budget. It's the same zero-sum game you already play. You rebalance between Google and Meta every quarter based on what's actually converting. We're not asking you to do something new — we're adding unattributed as one more column in a reallocation you already run.
And it's a capability you can bring to your own client relationships — an independent finding your client's finance team can't dispute, because it didn't come from the agency managing the spend.
Which way the mix should move. What one engagement answers.
How fast you can move it without spooking a board watching this quarter's number.
Whether you're moving fast enough — a competitor already capturing the demand you're still deciding whether to chase makes standing still a decision too.
Direction is a report. Speed and velocity are a relationship — which is why agencies that bring us in once tend to keep us on.
What each approach asks of your team before you get a first answer. No new pixels, no warehouse connection, no waiting for fresh data: we work from the history you already have.
| Task | Typical attribution platform | Mixed Signals Delta Check |
|---|---|---|
| New tracking code (CNAME, pixel) | Required · Dev/IT, Marketing Ops | None |
| Data warehouse connection | Required · Data engineering | None |
| Campaign tracking cleanup | Required · Marketing + agencies | None, we use what you have |
| Ad platform and MMP data | Account authentication | Report exports you already run |
| Conversion data | Ongoing batch file feed | One export of receipts |
| AI-chat signal | Not part of setup | One server log export |
| Waiting for clean data | Yes, while new tracking collects | None, uses your existing history |
| Time to first answer | Setup, then data collection | About 5–6 weeks, end to end |
| Price | Platform subscription, quoted | $2,500, fixed |
Typical setup for pixel-based attribution platforms. Requirements vary by vendor.
One month of attributed and unattributed transactions data$2,500 fixed, agreed before we start.
Data gathering — expectation-managed
2–3 weeks, typically. Getting real transaction data out of a corporate entity is never instant — it depends on your team's turnaround, not ours. We send the reminders. You won't have to chase us.
Analysis — firm commitment
15 business days, starting the day we have your data in hand. This part is entirely in our control, so it's the part we're willing to guarantee.
What you get at the end: your real overcounting delta, in writing, with the math shown. No retainer. No year-long commitment. If the honest answer is "there's nothing here," you'll know that for a small, fixed fee — not a guess you paid a year for.
If you manage paid and organic together and have never gotten a straight answer about which is actually working, that's the conversation.
Tweak your lookback period & minimize your conversion delta
Before we show you the delta, here's where every channel actually sits — attribution window and funnel stage, no guessing.
| Channel | Lookback | Funnel |
|---|---|---|
| Google Search Ads | 1-day view, 30-day click | Lower |
| Apple Search Ads | 30-day tap-through | Lower |
| Meta Ads | 1-day view, 7-day click | Mid |
| Snap Ads | 1-day view, 28-day click | Mid |
| 1-day view, 10-day click | Mid | |
| Blog post visit | 1-day view, 30-day click | Top |
| Unattributed (organic, self-driven) visitor | sliding scale | |
| Same session conversion | 1-day view, 1-day click | High-intent, late-stage |
| Conversion within 2 days | 1-day view, 2-day click | High-intent, late-stage |
| Conversion within 3 days | 1-day view, 3-day click | High-intent, late-stage |
| Conversion in 4 to 7 days | 1-day view, 4–7-day click | High-intent, late-stage |
Audit note — New customer,USD-INR Remittance
Audit note — New customer,USD-MXN Remittance
Let us audit your traffic behavior and map the correct install attribution window, view-through attribution window, and re-engagement attribution window.
Every one of those windows feeds straight into your reported ROAS. Inflate the numerator with overcounted conversions, and your ROAS is wrong by the same margin — not once, but every reporting period.
This is the one that actually produces the overcounting delta. It's the catch-and-release setting: how long a channel gets to hold a lead before releasing it back to the general pool. Hold too long, and a channel keeps claiming credit for someone who already converted through chat-researched, self-driven demand. Release too early, and whoever touches them next takes the win — even if it was never really theirs to claim.
30 days is the common platform default — shown here, not endorsed here. Your own corridor data already told a different story: some of your review-page lags measured in this conversation ran as short as a single day. There's no universal right answer, only the number that matches how your customers actually behave — which is exactly what the audit sets, corridor by corridor.
What platforms claim
Every ad platform runs on a fixed lookback window, set once, for every advertiser, regardless of how their customers actually decide.
Across the platforms and corridors we've measured, the real lag between a click and a conversion sits far closer to zero than any of these defaults — which means a real share of "attributed" credit is being claimed for conversions that would have happened anyway.
What nobody claims
In the businesses we've analyzed, well over half of all downloads and in-app conversions can't be traced to a single ad click at all.
A growing share of that gap starts before anyone opens an app — someone researching a decision in ChatGPT, Claude, or Gemini, then acting on it days later. No ad platform sees that conversation. No MMP tags it. We measure it directly, corridor by corridor, from real conversion data.
Full-service or paid-media-only, we work with agencies the same way we work with advertisers directly — as an independent measurement layer, not competition for the budget. It's the same zero-sum game you already play. You rebalance between Google and Meta every quarter based on what's actually converting. We're not asking you to do something new — we're adding unattributed as one more column in a reallocation you already run.
And it's a capability you can bring to your own client relationships — an independent finding your client's finance team can't dispute, because it didn't come from the agency managing the spend.
Which way the mix should move. What one engagement answers.
How fast you can move it without spooking a board watching this quarter's number.
Whether you're moving fast enough — a competitor already capturing the demand you're still deciding whether to chase makes standing still a decision too.
Direction is a report. Speed and velocity are a relationship — which is why agencies that bring us in once tend to keep us on.
What each approach asks of your team before you get a first answer. No new pixels, no warehouse connection, no waiting for fresh data: we work from the history you already have.
| Task | Typical attribution platform | Mixed Signals POC |
|---|---|---|
| New tracking code (CNAME, pixel) | Required · Dev/IT, Marketing Ops | None |
| Data warehouse connection | Required · Data engineering | None |
| Campaign tracking cleanup | Required · Marketing + agencies | None, we use what you have |
| Ad platform and MMP data | Account authentication | Report exports you already run |
| Conversion data | Ongoing batch file feed | One export of receipts |
| AI-chat signal | Not part of setup | One server log export |
| Waiting for clean data | Yes, while new tracking collects | None, uses your existing history |
| Time to first answer | Setup, then data collection | About 5–6 weeks, end to end |
| Price | Platform subscription, quoted | $2,500, fixed |
Typical setup for pixel-based attribution platforms. Requirements vary by vendor.
One month data (attribution & conversion data) $2,500, fixed, agreed before we start.
Data gathering — expectation-managed
2–3 weeks, typically. Getting real transaction data out of a corporate entity is never instant — it depends on your team's turnaround, not ours. We send the reminders. You won't have to chase us.
Analysis — firm commitment
15 business days, starting the day we have your data in hand. This part is entirely in our control, so it's the part we're willing to guarantee.
What you get at the end: your real overcounting delta, in writing, with the math shown. No retainer. No year-long commitment. If the honest answer is "there's nothing here," you'll know that for a small, fixed fee — not a guess you paid a year for.
If you manage paid and organic together and have never gotten a straight answer about which is actually working, that's the conversation.
Tweak your lookback period & minimise conversion delta.
Before we show you the delta, here's where every channel actually sits — attribution window and funnel stage, no guessing.
| Channel | Lookback | Funnel |
|---|---|---|
| Google Search Ads | 1-day view, 30-day click | Lower |
| Apple Search Ads | 30-day tap-through | Lower |
| Meta Ads | 1-day view, 7-day click | Mid |
| Snap Ads | 1-day view, 28-day click | Mid |
| 1-day view, 10-day click | Mid | |
| Blog post visit | 1-day view, 30-day click | Top |
| Unattributed (organic, self-driven) visitor | sliding scale | |
| Same session conversion | 1-day view, 1-day click | High-intent, late-stage |
| Conversion within 2 days | 1-day view, 2-day click | High-intent, late-stage |
| Conversion within 3 days | 1-day view, 3-day click | High-intent, late-stage |
| Conversion in 4 to 7 days | 1-day view, 4–7-day click | High-intent, late-stage |
Audit note — New customer, USD-MXN
Audit note — New customer, USD-INR
Let us audit your traffic behavior and map the correct install attribution window, view-through attribution window, and re-engagement attribution window.
Every one of those windows feeds straight into your reported ROAS. Inflate the numerator with overcounted conversions, and your ROAS is wrong by the same margin — not once, but every reporting period.
This is the one that actually produces the overcounting delta. It's the catch-and-release setting: how long a channel gets to hold a lead before releasing it back to the general pool. Hold too long, and a channel keeps claiming credit for someone who already converted through chat-researched, self-driven demand. Release too early, and whoever touches them next takes the win — even if it was never really theirs to claim.
30 days is the common platform default — shown here, not endorsed here. Your own corridor data already told a different story: some of your review-page lags measured in this conversation ran as short as a single day. There's no universal right answer, only the number that matches how your customers actually behave — which is exactly what the audit sets, corridor by corridor.
What platforms claim
Every ad platform runs on a fixed lookback window, set once, for every advertiser, regardless of how their customers actually decide.
Across the platforms and corridors we've measured, the real lag between a click and a conversion sits far closer to zero than any of these defaults — which means a real share of "attributed" credit is being claimed for conversions that would have happened anyway.
What nobody claims
In the businesses we've analyzed, well over half of all downloads and in-app conversions can't be traced to a single ad click at all.
A growing share of that gap starts before anyone opens an app — someone researching a decision in ChatGPT, Claude, or Gemini, then acting on it days later. No ad platform sees that conversation. No MMP tags it. We measure it directly, corridor by corridor, from real conversion data.
Full-service or paid-media-only, we work with agencies the same way we work with advertisers directly — as an independent measurement layer, not competition for the budget. It's the same zero-sum game you already play. You rebalance between Google and Meta every quarter based on what's actually converting. We're not asking you to do something new — we're adding unattributed as one more column in a reallocation you already run.
And it's a capability you can bring to your own client relationships — an independent finding your client's finance team can't dispute, because it didn't come from the agency managing the spend.
Which way the mix should move. What one engagement answers.
How fast you can move while watching this quarter's number.
Your rate of change v/s that of competitor's.
Direction is a report. Speed and velocity are a relationship — which is why agencies that bring us in once tend to keep us on.
Currently running a live measurement pilot with a global remittance provider, testing this against a real budget decision — not a backtest.
Every recommendation is validated with a scoped, reversible test before it touches a live budget. We'd rather be slow and right than fast and wrong about someone else's spend.
What each approach asks of your team before you get a first answer. No new pixels, no warehouse connection, no waiting for fresh data: we work from the history you already have.
| Task | Typical attribution platform | Mixed Signals POC |
|---|---|---|
| New tracking code (CNAME, pixel) | Required · Dev/IT, Marketing Ops | None |
| Data warehouse connection | Required · Data engineering | None |
| Campaign tracking cleanup | Required · Marketing + agencies | None, we use what you have |
| Ad platform and MMP data | Account authentication | Report exports you already run |
| Conversion data | Ongoing batch file feed | One export of receipts |
| AI-chat signal | Not part of setup | One server log export |
| Waiting for clean data | Yes, while new tracking collects | None, uses your existing history |
| Time to first answer | Setup, then data collection | About 5–6 weeks, end to end |
| Price | Platform subscription, quoted | $2,500, fixed |
Typical setup for pixel-based attribution platforms. Requirements vary by vendor.
One month of attributed & unattributed transaction data $2,500, fixed, agreed before we start.
Data gathering — expectation-managed
2–3 weeks, typically. Getting real transaction data out of a corporate entity is never instant — it depends on your team's turnaround, not ours. We send the reminders. You won't have to chase us.
Analysis — firm commitment
15 business days, starting the day we have your data in hand. This part is entirely in our control, so it's the part we're willing to guarantee.
What you get at the end: your real overcounting delta, in writing, with the math shown. No retainer. No year-long commitment. If the honest answer is "there's nothing here," you'll know that for a small, fixed fee — not a guess you paid a year for.
If you manage paid and organic together and have never gotten a straight answer about which is actually working, that's the conversation.
Tweak lookback window & minimize conversion delta
Before we show you the delta, here's where every channel actually sits — attribution window and funnel stage, no guessing.
| Channel | Lookback | Funnel |
|---|---|---|
| Google Search Ads | 1-day view, 30-day click | Lower |
| Apple Search Ads | 30-day tap-through | Lower |
| Meta Ads | 1-day view, 7-day click | Mid |
| Snap Ads | 1-day view, 28-day click | Mid |
| 1-day view, 10-day click | Mid | |
| Blog post visit | 1-day view, 30-day click | Top |
| Unattributed (organic, self-driven) visitor | sliding scale | |
| Same session conversion | 1-day view, 1-day click | High-intent, late-stage |
| Conversion within 2 days | 1-day view, 2-day click | High-intent, late-stage |
| Conversion within 3 days | 1-day view, 3-day click | High-intent, late-stage |
| Conversion in 4 to 7 days | 1-day view, 4–7-day click | High-intent, late-stage |
Audit note — New customer acquisition, Fintech
Let us audit your traffic behavior and map the correct install attribution window, view-through attribution window, and re-engagement attribution window.
Every one of those windows feeds straight into your reported ROAS. Inflate the numerator with overcounted conversions, and your ROAS is wrong by the same margin — not once, but every reporting period.
This is the one that actually produces the overcounting delta. It's the catch-and-release setting: how long a channel gets to hold a lead before releasing it back to the general pool. Hold too long, and a channel keeps claiming credit for someone who already converted through chat-researched, self-driven demand. Release too early, and whoever touches them next takes the win — even if it was never really theirs to claim.
30 days is the common platform default — shown here, not endorsed here. Your own corridor data already told a different story: some of your review-page lags measured in this conversation ran as short as a single day. There's no universal right answer, only the number that matches how your customers actually behave — which is exactly what the audit sets, corridor by corridor.
What platforms claim
Every ad platform runs on a fixed lookback window, set once, for every advertiser, regardless of how their customers actually decide.
Across the platforms and corridors we've measured, the real lag between a click and a conversion sits far closer to zero than any of these defaults — which means a real share of "attributed" credit is being claimed for conversions that would have happened anyway.
What nobody claims
In the businesses we've analyzed, well over half of all downloads and in-app conversions can't be traced to a single ad click at all.
A growing share of that gap starts before anyone opens an app — someone researching a decision in ChatGPT, Claude, or Gemini, then acting on it days later. No ad platform sees that conversation. No MMP tags it. We measure it directly, corridor by corridor, from real conversion data.
Full-service or paid-media-only, we work with agencies the same way we work with advertisers directly — as an independent measurement layer, not competition for the budget. It's the same zero-sum game you already play. You rebalance between Google and Meta every quarter based on what's actually converting. We're not asking you to do something new — we're adding unattributed as one more column in a reallocation you already run.
And it's a capability you can bring to your own client relationships — an independent finding your client's finance team can't dispute, because it didn't come from the agency managing the spend.
Which way the mix should move. What one engagement answers.
How fast you can move it without spooking a board watching this quarter's number.
Whether you're moving fast enough — a competitor already capturing the demand you're still deciding whether to chase makes standing still a decision too.
Direction is a report. Speed and velocity are a relationship — which is why agencies that bring us in once tend to keep us on.
What each approach asks of your team before you get a first answer. No new pixels, no warehouse connection, no waiting for fresh data: we work from the history you already have.
| Task | Typical attribution platform | Mixed Signals POC |
|---|---|---|
| New tracking code (CNAME, pixel) | Required · Dev/IT, Marketing Ops | None |
| Data warehouse connection | Required · Data engineering | None |
| Campaign tracking cleanup | Required · Marketing + agencies | None, we use what you have |
| Ad platform and MMP data | Account authentication | Report exports you already run |
| Conversion data | Ongoing batch file feed | One export of receipts |
| AI-chat signal | Not part of setup | One server log export |
| Waiting for clean data | Yes, while new tracking collects | None, uses your existing history |
| Time to first answer | Setup, then data collection | About 5–6 weeks, end to end |
| Price | Platform subscription, quoted | $2,500, fixed |
Typical setup for pixel-based attribution platforms. Requirements vary by vendor.
One month data ( attributed + unattributed) $2,500, fixed, agreed before we start.
Data gathering — expectation-managed
2–3 weeks, typically. Getting real transaction data out of a corporate entity is never instant — it depends on your team's turnaround, not ours. We send the reminders. You won't have to chase us.
Analysis — firm commitment
15 business days, starting the day we have your data in hand. This part is entirely in our control, so it's the part we're willing to guarantee.
What you get at the end: your real overcounting delta, in writing, with the math shown. No retainer. No year-long commitment. If the honest answer is "there's nothing here," you'll know that for a small, fixed fee — not a guess you paid a year for.
If you manage paid and organic together and have never gotten a straight answer about which is actually working, that's the conversation.