Where it stood in May
Every stage was leaking, and the top of the funnel was leaking hardest.
35%
Valid lead
rate
rate
3
Demos booked
2
Meetings
17 leads in. 6 worth calling. 3 demos. 2 meetings. One sign-up out the other end.
The Funnel
May against September, stage by stage.
Three Moves
Fix the signal. Fix the destination. Close the loop.
01 · Stop rewarding the wrong thing
Stripped out every secondary conversion action and redefined the conversion as a qualified lead only. This isn't a settings change, it's telling the platform who the product is for. The algorithm learns from whatever you reward it with, and it had been rewarded for junk since launch.
Valid lead rate 35% → 52%
02 · Fix the destination before scaling traffic
Mobile ads were running to a page never built for mobile. Paused spend rather than keep buying clicks into a leak. Read the call data first to see who had converted on mobile before and what had worked for them, then rebuilt around that: mobile-first layout, value proposition in the first screen, one CTA, social proof above the fold.
Demos 3 → 11
03 · Close the loop in the CRM
A qualified lead only counts as a conversion if the CRM can say what qualified means, so the criteria went in as a rule rather than a judgement call, and every lead was traced through to its outcome. Lead routing was rebuilt to cut the handoff delay — a lead that sits before anyone calls it goes cold, and the sales cycle was carrying that lag at every stage. Then CRM outcomes were synced back to Google Ads, so the bidding learned from what sales actually accepted instead of what the form recorded.
Routing · Qualification criteria · CRM → Google Ads sync
The Build, month by month
Five months. Each fix unlocked the next one.
May
Read the CRM, not the ads dashboard
Traced all 17 leads through to their outcome. 11 were never sales conversations. The ads reporting had no idea, because every one of them counted as a conversion.
Diagnosis · Lead-level audit
June
Redefine what counts as a conversion
Removed secondary conversion actions, set qualified lead as the only conversion, and let the campaign re-enter learning on the corrected signal. Quality moved before volume did.
Conversion objective
July
Pause and rebuild the mobile page
Rather than keep spending into a page that couldn't convert, paused mobile spend, reviewed call recordings from previous mobile converters, and rebuilt the page around what they had responded to.
Landing page
August
Shorten the handoff, sync the outcomes
Rebuilt lead routing so qualified leads reached a rep quickly instead of ageing in a queue, which took time out of the cycle at every stage. Synced CRM outcomes back into Google Ads so the campaign optimised against what sales accepted, not what the form captured.
CRM routing · Outcome sync
August
Turn traffic back on, cut the budget
Relaunched against the rebuilt page. With the campaign no longer buying junk, the same pipeline came from less money — so the budget came down 18% rather than up.
Efficiency
September
Every stage compounding
23 leads, 12 qualified, 11 demos, 9 meetings, 5 sign-ups. Each improved rate multiplied the one below it, which is why the bottom of the funnel moved further than the top.
Compounding
Results
May → September.
Valid lead rate
35→52%
share of leads worth calling
Leads
+36%
17 → 23
Qualified opportunities
2x
6 → 12 leads worth a sales conversation
Demos booked
3.7×
3 → 11
Meetings completed
4.5×
2 → 9
Ad spend
−18%
while all of the above went up
Cost per acquisition
−84%
per sign-up
How the cost figures are derived. Client spend is indexed, with May set to 100 and September at 82 after the 18% cut.
Cost per qualified opportunity: 100 ÷ 6 = 16.7 in May, 82 ÷ 12 = 6.8 in September — −59%.
Cost per demo: 100 ÷ 3 = 33.3 → 82 ÷ 11 = 7.5 — −78%.
Cost per sign-up: 100 ÷ 1 = 100 → 82 ÷ 5 = 16.4 — −84%.
The funnel chain holds at every step: 23 leads × 52% valid = 12 qualified, × 92% = 11 demos, × 82% = 9 meetings, × 56% = 5 sign-ups.
Cost per qualified opportunity: 100 ÷ 6 = 16.7 in May, 82 ÷ 12 = 6.8 in September — −59%.
Cost per demo: 100 ÷ 3 = 33.3 → 82 ÷ 11 = 7.5 — −78%.
Cost per sign-up: 100 ÷ 1 = 100 → 82 ÷ 5 = 16.4 — −84%.
The funnel chain holds at every step: 23 leads × 52% valid = 12 qualified, × 92% = 11 demos, × 82% = 9 meetings, × 56% = 5 sign-ups.
How I tell it in the interview
The 60-second version.
Situation
ITILITE, a B2B travel and expense SaaS, run through Revv Growth in 2025. The account looked healthy in the ads dashboard. In the CRM, 17 leads in May produced 6 real conversations, 3 demos and 1 sign-up.
Insight
The campaign was set to max conversions with every secondary action switched on, so the platform was buying the cheapest form fill it could find. It was doing its job. We'd given it the wrong job.
Action
Redefined the conversion as a qualified lead only and stripped the secondary actions. Then paused mobile spend, read the call data from previous mobile converters, and rebuilt the landing page around what had worked for them before turning traffic back on. In the CRM I set the qualification criteria, shortened the routing so leads reached a rep faster, and synced outcomes back into Google Ads.
Result
By September: leads up 36%, qualified opportunities up 2x, valid lead rate 35% to 52%, demos 3 to 11 — on 18% less ad spend. Cost per sign-up fell 84%.
Lesson
In paid search the biggest lever is usually not the bid or the budget. It's what you've told the platform a good outcome looks like.
The budget was never
the problem. The definition
of success was.
the problem. The definition
of success was.
Once a qualified lead was the only thing that counted as a conversion, the algorithm went and found them — and the rebuilt page converted the ones it found. Spending less was a consequence, not a goal.
Optimise for the outcome sales cares about, not the one the platform can see.
Want a paid engine
that sales trusts?
that sales trusts?
Let's talk performance marketing, B2B paid search, or funnel diagnostics.
Vinay Kumar · Performance Marketing