tl;dr Your funnel converts something like one visitor in 2,700, and the standard fix is to buy more visitors. The more useful move is to work the intent you already paid for: inside that anonymous 98% is a defensible slice of ICP-fit accounts showing buying behavior right now, sitting in tools you already own. The gap isn't seeing those signals, it's running them in the hour they're alive instead of days later, and that is the gap we built Bryn to close.
Before we start: this piece has our product in it. We launched Bryn three weeks ago, and the rationale below is the reason it exists. First the numbers, and they're public. The conclusion is simple math.
Ask your growth lead a different question this week. Don't ask how the campaigns performed, or what the site converted. Instead, ask what happened to the companies that visited your pricing page last Tuesday. Ask for names.
At most 50 to 500 person B2B software companies, this question will produce an awkward pause… The traffic numbers are fine. The conversion rate sits somewhere near the benchmark. But most teams struggle to tell you which target accounts showed up, what they did while they were there, or what your team did about it, because in most stacks the honest answer is nothing.
That's the most expensive "nothing" in your budget.
The 2026 benchmark data is not kind. For a mid-market B2B SaaS ($10M to $100M ARR), the published stage benchmarks average 1.4% visitor to lead, then 41% lead to MQL, 39% MQL to SQL, 42% SQL to opportunity, and 39% opportunity to close. Multiply that chain through and roughly one visitor in 2,700 becomes a paying customer. Meanwhile, the top 10% of companies convert visitors to leads at 8 to 15%, against an average closer to 1.5%. That's not a rounding difference. Compounded down the funnel, it's the difference between a pipeline problem and a pipeline machine, on the same traffic.
Now, let's put those numbers next to your own spend. The pattern may look something like this; your team spends $50,000 a month driving traffic to a website that converts at 1.8%, then you wonder why there isn't enough pipeline. The standard flawed response is to ask for more budget, which buys more visitors, 98% of whom will get the same treatment as the last batch: counted, aggregated into a dashboard, and forgotten.
THE FUNNEL LEAK, ON YOUR NUMBERS (interactive calculator) At the 1.4% mid-market benchmark: 1 customer per about 2,700 visitors. On 50,000 monthly visitors: about 700 leads per month, about 18 customers per month, and 49,300 visitors that never became a lead. Illustrative, from published benchmarks. The defensible ICP-fit slice hides inside that unworked pool. Move the visitor-to-lead rate to the top-decile band (8 to 15%) and it is the same traffic, a different company.Some caveats are important here, of course. Not all of that 98% is pipeline. A good part of it is bots, students, job seekers, and competitors doing their homework (you do it to them too). And person-level identity resolution is genuinely limited: VPNs, privacy rules, and match rates that are worse in practice than the vendor deck implied. I'm not claiming there's a hidden list of every visitor's direct line. There isn't.
But inside that anonymous 98% mass there is a slice you can defend in your board meeting:
You already paid to attract every one of them: the ad spend, the content, the product itself. That intent is an asset sitting on your books, and most of it expires unworked.
The pipeline you're trying to buy next quarter is, at least in part, sitting in the traffic you bought last quarter.
The standard fix for this has been increasing visibility:
And to be fair, the visibility tools mostly work as advertised. I've found that growth stacks at this company size are usually well instrumented but badly worked: the signals land in a dashboard, the dashboard feeds a weekly review, the review produces a spreadsheet, and someone works the spreadsheet when they get to it.
The cost of getting to it later is very well documented. Responding to a website signal within five minutes yields roughly an 8x higher contact rate than waiting 24 hours, and teams that batch-process visitor data weekly lose the timing advantage entirely. Intent decays in hours. A weekly signal review isn't prospecting, it's archaeology.
This matters more than it used to because buyers now run most of their evaluations on their own, anonymously, before anyone fills out a form. By the time a hand goes up, the shortlist is usually formed and their opinions are mostly baked. The timing window where showing up actually changes the outcome is the window your dashboard summarizes after the fact.
Watching signals and running them are different jobs. The first one is mature, competitive, and largely solved. The second one is where the pipeline actually comes from, and at most mid-market companies it's still done by hand, in batches, days late.
WATCHING ISN'T RUNNING (the same signal, on two clocks) WATCHING, read in days: HOUR 0 Signal lands in a dashboard. DAY 3 Dashboard feeds the weekly review. DAY 5 Review produces a spreadsheet. DAY 8+ Someone works it when they get to it. First action: days later. The story gets reconstructed by hand, after the window closed. RUNNING, run in minutes: MIN 0 Pattern named by its parts: pricing, then comparison, then a repeat within seven days. MIN 1 Scored transparently against your ICP. MIN 5 The approved Play runs, story attached. HOUR 1 Rep briefed, every action on the record. First action: the same hour. The first hour of the team's day is done before standup. Same traffic, same team; the delta is whether intent gets worked in the hour it's alive.Concretely, running a signal means the distance from the moment of intent to the first real action is minutes, not days, and the action arrives with its story attached.
The ICP-fit account that hit pricing twice this week gets scored, briefed, and put in front of the right rep the same hour, with the sequence of what it did and why it matters, not a row in a report. The trial that wobbled on day nine gets an assist drafted for the account executive before standup. The competitive evaluation gets caught while it's live, when the next call can still address it, instead of surfacing in the churn postmortem as a mystery.
For the operator, that's the first hour of every workday already done: the triage, the story reconstruction, the who-does-what.
For the executive, it's a different investment case entirely. Same traffic, same product, same team. The delta is whether intent gets worked in the hour it's alive. That's leverage on spend you've already made, and it doesn't show up as a headcount request. Speed to lead stops being something that rests solely on your best reps, and instead becomes a structural property of the system.
None of this requires believing anything mystical about agents. It requires believing that timing and follow-through, applied consistently to signals you already own, compound. The benchmark gap above says they do.
Three weeks ago we launched Bryn, Civic's Signal-Based GTM agent for Growth teams. This post outlines Bryn's reason for existing, so let me be direct about what it does, in terms of the benefits you can expect.
Bryn watches the three surfaces you already own (your product telemetry, your web traffic, your CRM) and runs the patterns your team can't keep up with. It doesn't tell you an account is showing intent and leave the reconstruction to you. It names what it saw, by its parts: e.g. pricing, then comparison, then a repeat visit inside seven days. It scores the account transparently, so your team can see why it scores at the level it does, and then it runs the Play your team approved, end to end, on the record.
Two design choices we made are worth highlighting:
That's the pitch, and I'll leave it there.
You can test the argument against your own company without buying anything. Three questions for your next growth review:
The answers will tell you whether you have a traffic problem or a running problem. In my experience most companies at this stage have the second one while budgeting for the first.
If you run growth at a PLG company and those questions made you wince, I'd like to compare notes, whether or not Bryn is ever part of the answer. Reply here or find me at chris@civic.com.