The Channel That Stopped Working

San Francisco · March 24, 2025

For eighteen months, Google Search ads were our golden channel. We spent $47,000 a month on SEM, generated 620 qualified signups, and converted 74 of those into paying customers. CAC: $635. LTV of a search-acquired customer: $4,100. LTV-to-CAC ratio of 6.5:1. Those are the kind of numbers that make a growth person sleep well at night.

Then, in January 2025, the numbers started to move. Not dramatically — not the kind of sudden cliff that makes you check if your tracking is broken. Slowly. The way water erodes a cliff face.

January: 580 signups. Down 6.5%. February: 512 signups. Down 11.7% from December. Cost per click went from $4.20 to $5.80. Conversion rate from click to signup dropped from 3.2% to 2.7%. CAC climbed from $635 to $891.

By March, I'm staring at a channel that went from our best performer to our most expensive in ninety days.

There are three possible explanations for a channel decline: your targeting got worse, your creative got worse, or the channel itself changed. I need to figure out which one — or which combination — is responsible.

I start with targeting. Our keyword set hasn't changed meaningfully. We're bidding on the same terms we've been bidding on: category terms, competitor terms, and long-tail problem-aware queries. I check the search terms report to see what queries are actually triggering our ads. The distribution looks normal — no weird irrelevant queries eating budget, no keyword cannibalization between campaigns.

Next, creative. Our ad copy hasn't changed in four months, which is either a sign of stability or laziness, depending on your perspective. I run a quick analysis on ad fatigue — comparing click-through rates over time for the same ad variations. CTR has declined about 8% across all ads, which suggests some fatigue but not enough to explain a 38% increase in CPC.

That leaves the channel. And this is where it gets interesting.

Google made three changes to its search results page between November 6, 2024 and February 26, 2025. First, the AI overview expanded to cover more commercial queries, pushing paid ads below the fold on certain searches. Second, they increased the number of ad slots from four to five on competitive queries, which means more advertisers competing for roughly the same volume of searches. Third, and this is the one that hurts the most, a new competitor (the same one that launched in October — the one I wrote about) started bidding aggressively on our brand terms and category terms.

Three competitors joined our keyword auctions in ninety days. That's not bad luck. That's a market signal. The category is getting crowded, and SEM is the first place crowding shows up.

I run the numbers on the competitor's ad spend. I can't see it directly, but I can estimate it from impression share data and auction insights. They're spending approximately $30,000-35,000 a month on the same keywords we're targeting. Before they entered, there were four serious bidders in our category. Now there are seven. The auction math is brutal: more bidders, same inventory, higher prices.

I lay out four options for the growth team. Each has trade-offs, and none of them is a clean win.

Option 1: Outspend the competition. Increase our SEM budget from $47,000 to $70,000/month to maintain volume at the new CPCs. Pro: maintains our top-of-funnel volume. Con: CAC goes to $946, which puts LTV-to-CAC at 4.3:1. Still profitable, but uncomfortably close to the 3:1 threshold where payback periods start stretching past eighteen months.

Option 2: Shift to long-tail keywords. Reduce spending on high-competition head terms and move budget to longer, more specific queries with lower CPCs. Pro: cheaper clicks, more qualified traffic. Con: lower volume — there are only so many people searching for "B2B SaaS onboarding analytics tool for mid-market companies."

Option 3: Diversify channels. Take $20,000 of the SEM budget and invest it in channels we've underinvested in — LinkedIn ads, content marketing, integration partnerships. Pro: reduces dependency on a single channel. Con: new channels take 3-6 months to mature, meaning a short-term volume gap.

Option 4: Do nothing differently, accept the new economics. The channel still works — it's just more expensive. Keep spending, absorb the higher CAC, and focus on improving downstream conversion to compensate. Pro: no disruption. Con: it assumes the decline has stabilized, which is an assumption I can't validate.

We go with a combination of Options 2 and 3, which is the kind of non-answer that sounds strategic in a meeting and feels chaotic in execution.

Here's what we actually do:

We cut the SEM budget on head terms by 40%, from $47,000 to $28,000. We use the savings — $19,000 — to fund three new channel experiments. Seven thousand goes to LinkedIn ads targeting specific job titles at companies in our ICP. Seven thousand goes to a content SEO sprint — twelve articles targeting bottom-of-funnel keywords that we've identified but never written for. Five thousand goes to an integration partnership pilot with our three most popular integrations.

We also build what I call a "long-tail machine" — a system for identifying and bidding on hyper-specific search queries that have low competition and high intent. Priya writes a script that pulls search terms from Google Search Console, cross-references them with our conversion data, and identifies queries where we rank organically between positions 8-20 (close enough to be relevant but not close enough to get clicks). We bid on those terms at low CPCs and create dedicated landing pages for each.

The long-tail machine is the kind of thing that only works if you're willing to manage 200+ keywords with individual bids and dedicated landing pages. It's operational intensity, not strategic brilliance. But at $1.40 per click versus $5.80, the math is hard to argue with.

It's now early May. Here's the scorecard:

SEM (head terms): $28,000 spent, 340 signups, 38 conversions. CAC: $737. Higher than the golden era but stabilized. The reduced budget means we're no longer trying to win every auction, and the cost pressure has eased slightly.

SEM (long-tail machine): $4,200 spent, 180 signups, 27 conversions. CAC: $156. These are small deals — average ACV of $780 — but the efficiency is remarkable. The long-tail customers are also more likely to be self-serve, which means lower support costs.

LinkedIn ads: $14,000 spent over two months, 89 signups, 7 conversions. CAC: $2,000. Too early and too expensive. But the leads are higher quality — larger companies, higher ACV potential. I'm giving it another sixty days before I make a call.

Content SEO: 12 articles published. Organic traffic to those articles: 4,200 visits. Signups attributed: 31. Conversions: 4 so far, with 12 in active trials. CAC: effectively $583 (amortized content cost over projected conversions). The content takes time to rank, so I expect this number to improve over the next six months as the articles gain authority.

Integration partnerships: 3 co-marketing campaigns launched. 142 signups. 19 conversions. CAC: $263. And the early retention data matches the historical pattern — integration-sourced customers stick around longer. This might be our new golden channel.

Integration partnerships might be our new golden channel. CAC of $263 and customers who stick around. Sometimes the best channel is the one you've been ignoring.

Every channel has a lifecycle. Discovery, growth, maturity, decline. SEM was in growth for us in 2023 and 2024. It hit maturity in late 2024. It entered decline in January 26, 2025 — not because we did anything wrong, but because the market changed. More competitors, more bidders, higher prices. The physics of auction-based advertising.

The mistake would be to wait until the channel is fully dead before diversifying. By then, you've built your entire acquisition model on a single foundation, and when it cracks, everything cracks with it. The right time to diversify is when the channel is still working but showing signs of decline. You're not desperate yet. You have budget. You can experiment.

We're not out of the woods. The total signup volume is still 15% below where it was six months ago, and it'll take another quarter for the new channels to mature. But we're no longer dependent on a single auction for our survival. That's progress.

The golden channel stopped being golden. It happens. The question isn't how to make it golden again. It's how many channels you need before you stop caring about the color of any single one.