Blog: Your Paid Search Drives Leads. Can You Prove It Drives Customers?

Terry Guttman
November 20, 2026
6 MIN READ

Key Takeaways

  • Platform-reported lead volume doesn’t tell you whether paid search leads became customers,
    or whether the ads caused it. Proving that takes three distinct levels of measurement.
  • Level one — optimizing to raw lead volume — is the default and the trap.
    The algorithm scales whatever you reward it for, which usually means cheap form fills, not real buyers.
  • Level two connects CRM data back to Google Ads
    through Enhanced Conversions for Leads, letting Smart Bidding chase revenue instead of volume. But long sales cycles need an earlier, reportable value signal, since conversions uploaded more than 63 days after the associated click won't be imported.
  • Level three — incrementality testing — is the only level that proves causation instead of correlation.
    It matters most where attribution looks strongest: a benchmark study of 225 DTC geo-based incrementality tests found branded search's median incremental ROAS around 0.70x, among the lowest of any channel tested.
  • Incrementality validates budget decisions,
    not day-to-day bids, and clean tests need real volume.

Table of Contents

Every paid search dashboard hands you two numbers: how many leads you got, and what the platform says your ROAS was. Neither one answers the question that actually matters. Did those leads become customers, and did the ads cause it?

Most teams optimize to whichever number is easiest to see, which is usually lead volume. That’s the default, and it’s a trap, because the algorithm will scale exactly what you reward it for. Getting to a real answer means climbing three levels of measurement, and each one requires more data and more volume than the one before it. 

Most teams stop at the first. Some reach the second. Very few make it to the third because the third level has real prerequisites the first two don’t.

This post walks through all three: why lead-volume optimization backfires, how connecting CRM data lets bidding chase revenue instead of form fills, and why even that isn’t proof the ads caused anything, and what actually is.

Level One: Optimizing to Lead Volume Is the Default

Lead volume is the default optimization target because it’s the easiest number to see and the easiest one for the platform to chase. Set “form submission” as your conversion goal, and Smart Bidding will find you more form submissions.

That’s exactly the problem. The algorithm doesn’t know or care whether a lead is a good fit. It optimizes for whatever event you told it to maximize. Reward it for volume, and it will find the cheapest, fastest, easiest-to-get version of that event: people who fill out a form quickly, not necessarily people who are likely to buy.

This tends to look like success from the dashboard. Lead count climbs, cost per lead drops, and the campaign report reads as a win. But lead quality is quietly sliding. The cheaper the leads get, the more of them are unqualified, uninterested, or never going to buy. Sales feels it first, fielding a growing share of leads that go nowhere. The metric improved. The business didn’t.

Level Two: Connecting CRM Data So Bidding Optimizes Toward Revenue

The fix is to give the algorithm something better to chase than form fills. Enhanced Conversions for Leads is Google’s upgraded successor to offline conversion import. It matches hashed CRM data back to the original ad click, so what happens after the lead form (a qualified opportunity, a closed deal) can flow back into Google Ads.

That connection unlocks the potential of value-based bidding: instead of optimizing for lead count, you attach a dollar value to each lead and let Smart Bidding chase value. 

There are two reasonable ways to set that value. Where your sales cycle allows it, use real closed-won revenue. Where it doesn’t (because deals take too long to close within a reportable window) use a defensible proxy, like a lead score or a quote value multiplied by your close rate.

That “reportable window” is the catch for long sales cycles. Google’s import window is time-bound: offline conversions uploaded more than 63 days after the associated click won’t be imported for enhanced conversion leads. A deal that takes four months to close will have aged out of the window long before it ever gets counted. 

Assign value to the earliest reliable signal you can report in time, whether that’s a qualified-opportunity stage or a scored lead, so the algorithm still gets something to learn from before the window shuts.

Level Three: Incrementality Is the Only Level That Proves Causation

Level two proves leads became revenue. But it’s still attribution, not causation. Attribution credits paid search for every customer who touched an ad somewhere along the way, whether or not the ad actually changed their behavior. It can’t tell you which of those customers would have bought anyway.

Incrementality testing answers that question directly. Conversion Lift studies (run by the platforms) and geo holdout tests (run by the advertiser, comparing markets with ads on against matched markets with ads paused) isolate the revenue your ads truly caused, by comparing exposed and unexposed groups rather than crediting every touch along the path.

The counterintuitive part is where incrementality matters most: exactly the campaigns that look best on attribution. Branded and high-intent search convert at high rates because people searching your brand name are often already planning to buy. The ad captures credit for a sale that was likely happening regardless. 

One benchmark study of 225 DTC geo-based incrementality tests found branded search’s median incremental ROAS around 0.70x, the lowest of any channel in the dataset and well below the platform-reported numbers those campaigns typically show.

A low incremental ROAS doesn’t automatically mean the answer is to turn branded search off, but don’t assume the dashboard ROAS is the true number until you’ve tested it.

What Incrementality Testing Can (and Can’t) Do for Your Bidding

Incrementality is a budget-decision tool, not a bidding input. It tells you whether a channel or campaign is worth the spend at the level you’re running it. It doesn’t feed Smart Bidding the way CRM-connected conversion values do. Treat a lift study result as evidence for a quarterly or annual budget call, not as a signal to plug into daily optimization.

It also has a real prerequisite: a clean test needs enough volume and stable enough markets or segments to detect a lift with statistical confidence. That means the smallest accounts and the longest sales cycles will be the last ones able to run a valid test.

Level three is a maturity milestone you grow into.

How the Three Levels Work Together

  1. Level one (lead volume): the default state. A necessary starting point, but insufficient on its own. If left on its own, it usually ends up optimizing for the wrong thing.
  2. Level two (CRM-connected value): Enhanced Conversions for Leads plus value-based bidding, turning “how many leads” into “how much are they worth.”
  3. Level three (incrementality): Conversion Lift and geo holdout testing, turning “how much revenue is attributed to search” into “how much revenue did search actually cause.”

Each level answers a different question, and none of them replaces the one before it.

Proving Paid Search Earns Its Budget

Lead volume and platform ROAS are the easiest numbers to report and the least reliable ones to act on. Proving paid search drives customers means climbing through all three levels: getting past the trap of optimizing for form fills, connecting CRM data so bidding chases revenue, and eventually testing whether the ads caused that revenue or just claimed credit for it.

If you can’t yet say whether your paid search leads become customers — or whether your ads caused it — Silverback can help you build the CRM connection and, when you’re ready, the incrementality test to prove it. Get in touch to talk through your account.

Terry Guttman

Terry Guttman is the Associate Director of Ad Ops & Data at Silverback Strategies, where he leads the measurement strategy and data science behind the agency's paid media programs. A graduate of James Madison University, Terry has spent more than a decade in digital marketing, growing from hands-on paid media management into measurement leadership.

He's known for helping clients move past unreliable, click-based attribution toward first-party data strategy, incrementality testing, and media mix modeling they can scale revenue upon. Terry brings that measurement clarity to the mid-market lead generation and ecommerce businesses across Silverback's portfolio, and writes and speaks on these topics for the agency.

FAQ

What is Enhanced Conversions for Leads in Google Ads?

It's Google's upgraded successor to offline conversion import. It matches hashed CRM data (like a closed-won opportunity) back to the original ad click, so Smart Bidding can optimize toward revenue instead of lead volume alone.

How do you set a lead value for value-based bidding when your sales cycle is long?

Use real closed-won revenue if your sales cycle is short enough to report it within Google's import window. If it isn't, use a defensible proxy — a lead score or a quote value multiplied by your close rate — tied to the earliest reliable signal you can report in time.

What's the difference between attribution and incrementality?

Attribution credits a channel for every customer who touched it somewhere along the path to purchase, whether or not the channel changed their behavior. Incrementality testing isolates the revenue a channel actually caused by comparing exposed and unexposed groups directly.

Why does branded search often have low incremental ROAS despite high platform-reported ROAS?

Branded search largely captures people who already intend to buy and search your brand name as a final step. The ad gets credit for a sale that was likely going to happen anyway. That's why a benchmark study of DTC geo tests found branded search's incremental ROAS among the lowest of any channel tested, despite typically strong platform-reported numbers.

Do small advertisers need incrementality testing to prove paid search works?

Not right away. Clean incrementality tests require real volume and stable markets to detect a lift with confidence, which makes level three a later milestone for smaller or longer-sales-cycle advertisers. Connecting CRM data and bidding on value (level two) is achievable much sooner and delivers most of the improvement.