Cadence: 71% Click Drop, 3 Channels That Refilled It

Derek SimmonsEleven months number one spot on Google and trial sign-ups off that exact page dropped 71%.

Elena ReyesHold on, it held position one the entire time?

Derek SimmonsThe whole run. Rankings said winning, the funnel said otherwise.

Elena ReyesWell, that's a hell of a way to start a Tuesday.

Derek SimmonsFair enough. Welcome to ARR Autopsy. I'm Derek.

Elena ReyesAnd I'm here, sorry, that number just wiped the greeting right out of my head.

Derek SimmonsHead. Understandable. Today we're taking apart a SaaS founder's funnel that search quietly hollowed out from underneath him.

Elena ReyesBefore it broke, this guy had $340,000 in ARR, a team of four, and most of his trials tied to that one page.

Derek SimmonsWe get into the two moves he made once he finally accepted the channel itself was structurally broken.

Elena ReyesAnd I want to stress test what happened to CAC afterwards because it didn't just improve, it got worse. It got worse

Derek SimmonsWow!

Elena Reyesfirst.

Derek SimmonsUgh, you're gonna love that part.

Elena ReyesThen there's the operational mess once the new channel actually scaled-support load nobody staffed for.

Derek SimmonsI've seen this movie before-fix the top of the funnel and something downstream quietly catches fire.

Elena ReyesWhich is exactly where we're headed. So, Derek?

Derek SimmonsLet's start with the page, the one that won and lost at the same time. So this is one page ranking one for 11 months straight.

Elena ReyesOkay, love is specific number. Keep going.

Derek SimmonsTrial signups off that exact page dropped 71%.

Elena ReyesWait, Wait, wait. 71% while it's still ranking number one?

Derek SimmonsSame page, same position, same green check marks in the rank

Elena ReyesWow.

Derek Simmonstracker the whole time.

Elena ReyesSo the tracker says everything's fine.

Derek SimmonsAnd the trial dashboard's on fire. Two different stories. Different stories, same URL.

Elena ReyesWhich pages and over what window, because SEO is broken and this page died in April are different problems.

Derek SimmonsThat's the spreadsheet our guest is bringing today, month by month, page by page.

Elena ReyesGood. Before we get to their numbers though, I want the industry backdrop. Is this a them problem or an everyone problem?

Derek SimmonsBit of both, and it's moving fast. SparkToro and Similarweb ran a study in July. in June covering searches from January through April.

Speaker 3And?

Derek Simmons68.01% of U.S. Google searches ended with zero clicks. Out of a thousand searches, roughly 276 actually make it to the open web. So three out of four searches never leave Google? Basically. And on B2B specifically, BrightEdge data from earlier this year put AI Overviews on about About 82% of B2B tech queries, 82, of those AI answers, how many pull from page one? This is the part that should scare every SEO person listening. Only 17% of AI Overview citations come from a top 10 organic result. So rank and citation just broke up. Completely. You can hold Position One and still be invisible to the thing. Thing actually answering the question.

Elena ReyesOkay, that's the whole industry though. I went back to the guests page. Eleven months at number one, 71% gone. That's not a traffic story.

Derek SimmonsIt's a funnel rebuild, and this founder didn't guess, they brought the numbers, month by month, page by page.

Elena ReyesSo what did that funnel look like before the page cracked? What ARR, what headcount was writing on one URL? Real.

Derek SimmonsOkay, backing into the wreckage now, what did the business actually look like right before this fell apart?

Elena ReyesRight. Give me the balance sheet, not the vibes.

Derek SimmonsSo this is a team of four, ARR sitting around $340,000, and 68% of trial signups were coming from that one blog channel. Sixty-eight%? That's not a channel. That's a single point of failure. Yeah, Classic problem. Concentration risk with a With a Content calendar attached.

Speaker 4Okay, but let me put on my advisor hat: what did they actually try before admitting the channel was broken?

Derek SimmonsThree things: First, publish more (they went from three posts a week to five);

Speaker 4And what moved?

Derek SimmonsNothing. Flat Trial numbers, more invoices from freelancers.

Speaker 4Great use of six thousand a month!

Derek SimmonsSecond move: refreshed forty old posts; Updated stats, new screenshots.

Speaker 4Did rank change?

Derek SimmonsBarely. Same position, same falling trial count underneath it.

Speaker 4So the page held rank the whole time and nobody clicked. Third thing?

Derek SimmonsSchema markup and citation tweaks, trying to get scooped into AI answers. Ninety pages tagged up.

Speaker 4And did it get cited?

Derek SimmonsBarely moved the needle. Which tracks. As SEOincs been writing about this exactly, that ranking and getting cited by AI overviews is

Speaker 5a top priority.

Derek SimmonsUse is decoupling.

Speaker 4Right, and Mersel made almost the same point in their piece for March: the traffic that's disappearing isn't a penalty, it's buyers finishing their research inside the answer box before they ever click through.

Derek SimmonsRight, so Schema doesn't fix a structural problem, it's rearranging furniture.

Speaker 4Okay, so when did the founders stop calling this seasonality?

Derek SimmonsFebruary. That's the month it stopped being deniable.

Speaker 4What was the weekly number doing?

Derek SimmonsThey'd been running around a hundred and ten Trials a week off that page. By the stretch I flagged it, they were down to thirty.

Speaker 4In how long?

Derek SimmonsAbout ten weeks.

Speaker 4Who caught it first, the Founder or Someone lower down watching the Dashboard? Their growth higher.

Derek SimmonsFounder was still calling it a slow quarter for two more weeks after that.

Speaker 4Always the person closest to the number, never the person with the title.

Derek SimmonsEvery time.

Speaker 4All right, let's stress test the old math. If blog traffic was free, what was the fully loaded cost per trial before this?

Derek SimmonsContent spend was about six grand a month. Writer, editor, couple tools, at a hundred and ten trials a week, that's roughly fourteen dollars a trial.

Speaker 4And after the collapse?

Derek SimmonsSame six grand, thirty trials a week. Cost per trial jumps to almost fifty.

Speaker 4So the channel didn't get more expensive; the denominator just collapsed underneath it!

Derek SimmonsExactly; same spend, same team, three and a half times the cost per lead-which means the free channel story was already a lie before anyone noticed, it was just a little CAC hiding in a big enough number?" o c a c hiding in a big enough number?"

Elena ReyesAnd that's the moment they stopped patching the old engine and went looking for a new one-which is where the replacement moves actually start working. Two of them, and one of them only LOOKS like it worked. So the channels confirmed dead, what actually filled the hole? Move one, comparison pages. Us versus competitor X, built for people already shopping. Wait, comparison pages, not another content sprint? Nope, four pages, three competitors. The first ten qualified signups landed in nineteen days. Nineteen days from zero? What was the weekly cadence once it caught? Pot. Six to eight signups a week. Steady for ten weeks straight. Not explosive, but repeatable. That's the part I care about. Repeatable beats viral every time. Move 2's the bigger swing. They killed the How-To calendar entirely. Uh, entirely? What did that cost them? Six planned articles. Gone. The budget funded an interactive pricing calculator and a migration checklist tool. tool, things you have to use, not read.

Speaker 3Right, because an AI overview can summarize a definition; it can't run your calculator for you.

Elena ReyesExactly the bet. SparkToro's Rand Fishkin has argued branded, local, and high intent transactional queries still earn clicks, while informational and how-to queries get absorbed into the answer itself.

Derek SimmonsWhich tracks with what survived here:

Elena ReyesWow.

Derek Simmonsthe tool pages held...

Speaker 3Old traffic, the How To posts, didn't.

Elena ReyesThe migration checklist alone drove forty-one percent of new signups in month two.

Speaker 3Okay, I need to pump the brakes. Go ahead. Both moves worked, but did they work for the same reason?

Elena ReyesMeaning...

Speaker 3The comparison pages launched in March. The tool launched in March. That's also when a partner co-marketing push ran.

Derek SimmonsYou think the co-marketing push is inflating the tool's numbers?

Speaker 3I think it's possible. Which move do we actually credit to the lift?

Derek SimmonsSo what was the control? Did they hold the tool back from one segment to isolate it?

Speaker 3They didn't. No holdout, no split test. Everything launched the same quarter and the aggregate line went up. So, the sample doesn't isolate anything. That's a real gap. My read, the comparison pages are the real mechanic--direct, trackable, tied to specific searches. The tools forty one percent is real, but we can't say it caused that alone. Which matters, because a founder copying this funds the wrong thing. Chase the pages first, clean signal. Treat the tool as a bonus until you can test it in isolation.

Elena ReyesEven if we can't fully separate them, the funnel refilled. So the real question now is what it costs to refill it.

Speaker 3Uh, here we go.

Elena ReyesVolumes one story, cost per acquisition is a different one entirely. Operator Hat back on. Let's talk the actual bill, not just the traffic.

Speaker 4Mm-hmm. Here we go.

Elena ReyesBlended CAC actually got worse first; post-rebuild, months one and two, it climbed almost thirty per cent over the old blog era number.

Speaker 4Wait, worse? After all that work?

Elena ReyesOld blog CAC ran cheap, call it a hundred and eighty dollars; blended landed near three hundred and ten dollars before it settled.

Speaker 4That's a jump.

Elena ReyesFor two months then the higher converting channels caught up and it dropped to two hundred and ten dollars.

Speaker 4Worse before better, same shape as the volume drop.

Elena ReyesExact same shape, but channel by channel comparison page trials converted to paid nearly double the blog rate.

Speaker 4So fewer people, better people.

Elena ReyesPayback on that channel dropped from roughly five months to under three.

Speaker 3Wow!

Speaker 4Okay, that's real money.

Elena ReyesNow my skeptic hat goes on. Semrush ran research in mid-2025, finding AI-referred visitors converting around 4.4 times the rate of regular organic traffic.

Speaker 4I've seen way bigger numbers floating around.

Elena ReyesSome vendor studies claim ten, twenty times. I don't trust those without seeing the methodology.

Speaker 4Does the guest's own data back up the four point four?

Elena ReyesSort of. His AI-attributed signups converted about three times better than old blog traffic, close to that range, not some moonshot number.

Speaker 4That tracks doesn't confirm the multiplier, though.

Elena ReyesRight, and nobody's cracked the actual attribution problem yet.

Speaker 4Which is-

Elena ReyesSomeone hears the brand inside ChatGPT answer, then types the brand straight to Google a week later.

Speaker 4And Analytics just files that under branded search.

Elena ReyesThe conversation that actually won the customer disappears from the data completely.

Speaker 4So real CAC on that channel might be lower than the dashboard shows.

Elena ReyesCould be nobody's built the model to prove it.

Speaker 4Which means the bill for scaling this thing is still coming due.

Elena ReyesIt came, just not the bill any one expected.

Speaker 3Shifting to what actually broke once this thing scaled, nobody staffed for the mess underneath. What kind of mess? The interactive tool started pulling in triple the memo requests, but a chunk of them were just people playing with the calculator, not buying.

Elena ReyesSo somebody had to sort real leads from noise?

Speaker 3Manually. The founder told us their two-person CS team burned close to 15 hours a week just triaging junk requests.

Elena ReyesBefore a rep ever got on a call. Fifteen hours, that's basically another half a headcount hiding in a growth chart. Exactly the kind of debt that doesn't show up until payroll does, knocked desk, and it gets worse two quarters out. Go on. 90 day churn on the new cohort hit 22%, way above their historical baseline. Okay, but is that a product problem or just bad fit signups from the comparison pages? Yep! That's the question they had to answer fast. They pulled ICP scores against churn and found it wasn't the product, it was traffic from adjacent categories the comparison pages ranked for, people who were never going to convert long term. So the funnel got wider, but the wrong people walked through it. Right, and the fix that stuck was simple: they put a two question qualifier gate before and

Speaker 5after the comparison pages.

Derek SimmonsAnyone could book a demo off the tool or comparison pages.

Elena ReyesDid it actually move the churn number?

Derek SimmonsCohort churn dropped back toward their historical range within one more quarter, and support triage time roughly cut in half.

Elena ReyesWhat would he do differently on day one?

Derek SimmonsBuild that qualifier gate before scaling the channel, not after. If you're between 200k and 5 million ARR and you're about to pour traffic into a new page, Put the filter in front of it this week-not after the support team's drowning. Cheap insurance! Cheapest fifteen hours a week you'll ever buy back! Okay, so if there's one takeaway from today,

Elena ReyesIt's that a page can hold rank one for 11 months and still bleed trials. Rank isn't revenue.

Derek SimmonsRight. And that qualifier gate fix at the end? That's the one I'm stealing for my own funnel.

Elena ReyesSame. Nobody staffs for that support load until it's already on fire. Classic problem. Every single time.

Derek SimmonsAlright, if this saved you from a bad bet, send it to a Founder who needs it.

Elena ReyesAnd subscribe wherever you're listening, YouTube or your podcast app. Leave us a review. That's how we keep getting Founders to hand over their real numbers. Thanks for spending this one with us. We'll see you next time. Bye, everyone.