A founder called me last spring about a website she said wasn't converting. Six months in, traffic was up 40 percent year over year. Form fills were flat. She was about to fire the marketing agency that built the site and replace them with a fourth one in three years.
We took a look before she made the call. Forty-five minutes into the audit, we found the problem. The site was converting. Form submissions were arriving. The agency had set up confirmation emails to the prospect but no notification email to the firm. The leads were sitting in the database, unread for weeks. Whoever clicked send on the form got an autoresponder. Nobody at the firm got a heads-up.
The site wasn't broken. The pipe between site and CRM was. We rebuilt the routing on top of Hello Automations, our intake platform, and within a week the firm was working a backlog of 87 unanswered leads. Two of them closed at five-figure deal sizes. The agency hadn't been incompetent in any flashy way. They'd just shipped a website without auditing what happened after the form fired.
That's the experience this piece is about. Most complaints that "my website doesn't convert" aren't conversion problems. They're measurement problems. The site is doing what it can. Nobody knows because nobody is watching.
Why the conversion problem is almost never a design problem
The agency narrative goes something like this. Your site looks dated. Your hero is too generic. Your CTA placement is wrong. Your page load is slow. We'll redesign it for sixty thousand dollars and you'll see a 30 percent lift.
Some of that might be true. Most of it usually isn't. Page load is real. CTA placement is real. Hero copy matters. But these are 5 to 15 percent levers, not the difference between a site that converts and one that doesn't. The difference between a converting site and a non-converting site is almost always one of three things.
The form isn't routing to a human who answers. The phone calls aren't being tracked. Or the closed-loop data from the CRM isn't feeding back into the channel that drove the visit, so optimization is operating blind.
Each of those is an instrumentation problem. Redesigning the site doesn't fix any of them. It just makes the broken pipes prettier.
What we actually audit when a client says "my site isn't converting"
There's a thirty-minute audit we run before we even open Figma. It walks through the actual conversion plumbing in the order that matters.
Step one: trigger every form on the site
Every public form. Every "request a quote" button. Every newsletter signup. Every chat widget. We submit fake leads with traceable identifiers ("audit-test-ms-2026" in the message field) and watch where they land. Half the time we find one of these:
The submission triggers a confirmation page but no actual email is sent to the firm. The form was wired up at launch and the SMTP credentials rotated six months later, silently breaking the notification path. The prospect thinks they submitted. The firm has no idea anyone reached out.
The submission lands in a Gmail folder labeled "Website inquiries" that nobody has rules pointing at. It sits unread for days. The auto-responder told the prospect "we'll be in touch within 24 hours." That promise is broken before it's made.
The form is wired to a Zapier flow that pushes to a CRM nobody actively works. The leads are technically captured. They're just rotting in a system the firm checks once a quarter.
A redesign doesn't fix any of these. A 90-minute plumbing audit and a working notification rule does.
Step two: trace every phone call
Phone-tracking is the lever most service businesses miss entirely. The firm with calls coming into the main office line and a receptionist taking notes on paper. The website that lists a tracking number for the paid campaigns but a different number on the contact page. The Google Business Profile that uses yet a third number because it was claimed by a former marketing manager who never handed off the credentials.
What we look at: are inbound calls traceable to source? Does the firm record them, or at minimum log call duration and time? Does the receptionist's qualification routing route by source, with paid clicks getting a different greeting from organic clicks? Does the call data flow back into the analytics layer, or sit in a separate phone log nobody reconciles?
If the answer to any of these is no, the firm is operating blind on the highest-conversion-rate channel in the business. We had a legal client where 72 percent of new clients came from a phone call, not a form. The phone calls were being tracked nowhere. The agency had been optimizing the form on the contact page for nine months. They were optimizing the channel that produced 28 percent of revenue.
Step three: check the closed loop
This is the one nobody does. The closed loop is the data path that goes:
Visitor clicks ad → fills form (or calls) → enters CRM → progresses through pipeline → closes (or doesn't)
For optimization to work, the close (or lose) signal has to flow back into the channel that drove the visit. Google Ads needs to know which clicks turned into closed cases. Google Analytics needs to know which sessions became revenue. The SEO program needs to know which keywords delivered customers, not just which delivered traffic.
Most service-business websites have none of this. The Google Ads account optimizes against form submissions. Half of those form submissions are spam, retargeted bounces, or unqualified leads. The bidder learns to find more of those, because that's the signal it's been given. The CPL on form submissions looks fine on paper. The CPL on closed cases is three to four times higher.
We fixed this for a Jacksonville family law client by running offline conversion uploads from their case management system back into Google Ads, weighted by case value. Their blended CPL on closed cases is $62.79. The state average for the same family-law terms is between $180 and $260. That's not a clever bidding strategy. It's the channel finally seeing the closed-loop signal it needed.
The three things that have to be true for a website to "convert"
I'll be specific about what I think the bar is. After about a thousand audits over twelve years, the pattern is consistent.
One. Every form submission needs to land in a human inbox, get auto-acknowledged to the prospect, and get worked within the SLA the firm has promised on the page. A site that promises "we'll respond within 24 hours" and routes the lead to a forgotten Gmail folder is a worse site than one that doesn't promise anything at all. The promise is a contract. Breaking the contract on a service-business site costs trust at the moment trust is highest.
Two. Every inbound phone call needs to be tracked, attributed, and routed by source. If your paid campaigns and your organic traffic share the same phone number, you cannot optimize either of them. We use Twilio routing inside Hello Automations for service-business clients so the call data flows the same path the form data does. The closed-loop attribution has to cover both channels, not just the one that's easier to instrument.
Three. The CRM has to feed close (or lose) back into Google Ads, GA4, and the analytics layer. Without this, every optimization decision is operating on the wrong target. The bidder is finding more form-fillers. The SEO program is ranking for terms that drive traffic without buyers. The agency is reporting on metrics that don't tie to revenue. Six months of work disappears into the gap between marketing data and CRM data.
If any of those three things isn't true, the conversion problem isn't a design problem. The redesign that fixes the design without fixing the instrumentation is selling you the wrong product.
What it actually costs to fix
This is where the agency conversation usually goes off the rails. The firm asks "how much to fix this." The agency quotes a redesign because that's what they're set up to sell. Nobody mentions that the actual fix is mostly plumbing.
The real fix breaks down into three buckets.
Form routing and notification rules: $0 to $500. It's a configuration job inside whatever CRM and form vendor you already use. Half a day of an experienced operator. Hello Automations clients get this as part of the platform, but you don't need our platform to fix it. You need someone who treats lead routing as a real engineering problem instead of an afterthought.
Call tracking setup: $200 to $500 a month. CallRail, CallTrackingMetrics, or Twilio routed inside your CRM. The number of inbound paths you need to track depends on how many channels you run. For a small service business, three numbers (paid, organic, brand) cover most of it.
Closed-loop attribution to Google Ads and GA4: $1,500 to $3,000 in setup, then ongoing. This is the meaningful budget line. It requires a person who can write conversion uploads against your CRM's API, schedule them, monitor them, and reconcile when something breaks. We do this inside Orbit, our measurement platform, because we needed the discipline anyway and it pays for itself the first month it runs.
Total to fix the actual conversion problem: about five thousand dollars in setup, plus a few hundred a month in tooling. That's an order of magnitude less than a redesign and addresses the actual issue.
When a redesign is the right answer
I want to be fair about this, because I've watched the redesign-pitch cynicism go too far in the other direction. Some sites genuinely need to be rebuilt. The patterns where a redesign actually moves conversion:
The site is on a platform that can't be instrumented properly. WordPress 4.x with a sixteen-plugin Frankenstein page builder is one of these. We can't add proper schema, can't trust the form vendor, can't reliably ship Core Web Vitals. The right answer is a bespoke web platform build on the right stack, and the conversion lift comes from being able to instrument it cleanly.
The IA is genuinely wrong. The practice areas are buried three clicks deep. The contact mechanism on mobile requires pinch-zoom. The conversion paths the visitor would naturally want don't exist. This is rare but real. When it's real, redesign is the cheapest fix.
The brand has fundamentally changed. Acquisition, vertical pivot, new service line. The old site can't carry the new positioning, and patching it costs more than starting over.
If none of these apply, the site is fine. The instrumentation needs work.
Why we run this audit before quoting any web work
The first thing we do when a prospective client asks us about a redesign is run the audit above. About a third of the time we tell the prospect they don't need us. The conversion problem they came to fix is solvable for under five thousand dollars in plumbing, and we'll point them at the contractor or the in-house person who can do it.
About a third of the time the audit confirms what they suspected. The site has structural issues that make instrumentation impossible to layer on cleanly. We scope a real rebuild against the eight-to-ten-week build window plus a week on location for photography and we work for a quarter of the year on one site at a time.
The last third are somewhere in between. Maybe the site is salvageable but the brand needs a refresh. Maybe the IA is fine but the form layer needs to be rebuilt. We scope what's actually needed, not what's easiest to bill for.
The reason we do this is the reason this article exists. We've watched too many service businesses spend sixty thousand dollars on a redesign that didn't move their cases-closed number, because the actual problem was twenty thousand dollars deeper in the instrumentation layer. Telling clients to fix the cheap thing first is bad for short-term billing and good for the kind of long engagements we want to be in.
If you're staring at a website you think isn't converting and you're about to call an agency for a redesign, run the audit above first. Trigger every form. Trace every call. Check the closed loop. If all three are working and you're still not getting cases, then the redesign conversation is the right one to have. If any of the three is broken, you have a cheaper problem on your hands.