Your customer journey isn't a diagram. It's whether the next step exists at the exact moment somebody wants to take it.
Every article on this subject tells you to map it. Build personas, plot the touchpoints, note the emotion at each stage, find the drop-off in your analytics. That advice was written for companies with enough traffic for a funnel report to mean something, and a researcher on staff to read it.
Here is what the same problem looks like on a real business, and what it cost to fix.
The site
A California real estate school, in business over a decade, with a real office in Carlsbad. They rank well.
Traffic was never their problem. So I stopped looking at analytics and did what a customer does. Searched on my phone, the way somebody deciding on a career change at lunch would, and landed on their San Diego county page. Not the homepage. The county page, because that's where local search drops you.
I scrolled for a price. There isn't one on that county page, or on any county page, or on any state page. The prices exist one level up, $199 through $1,299, but only if you already know to go looking for them.
So I went for the phone. There's a heading on that page that says PHONE, and under it, "P: (858) 329-0999," which is not a link at all. Just text. I tapped the number in the header instead, and my phone wouldn't dial it.
That's where the path ran out. And it's the part nobody inside the business had seen, because nobody who works there has ever had to call their own school from a phone.
The audit
Walking it myself showed me one route. I wanted every route, so I ran a full QA: five buyer personas, plus a crawl of every URL in the site's own sitemap, in a real browser. It read rendered pages and live network requests, not source. 519 pages, zero fetch failures.
The personas were the career changer on her phone in a parking lot, a licensed agent shopping for his broker course, an operations manager at a brokerage told to get the terms in writing, a developer asked by his mother-in-law whether the school was legitimate, and a 66-year-old browsing at 175 percent zoom who would rather phone than fill anything in.
What it found:
tel:+8583290999. A plus sign and ten digits, no country code, so a phone reads it as an international number and refuses. That was live on 318 of 519 pages. The correct string was already on the other 197.Finding one
Every customer journey article assumes the same route. Homepage, then services, then contact. Map that path, find the friction, fix it.
Local search doesn't work that way. Somebody searching on a phone lands on whichever page matched, and for this school that's a county page. She never saw the homepage. She never saw the pricing page. Whatever careful sequence exists across the site, she entered it three-quarters of the way in and had to make her decision from what was in front of her.
So the real test isn't whether the journey works. It's whether every page works as a first page. Take any page that ranks, imagine somebody landing on it cold, and ask whether they can find out what it costs and reach a human. On this site the answer was no on several hundred pages, and the homepage, the one page everyone reviews, was one of the better ones.
Finding two
It starts on a search result or a Google listing, and it keeps going after somebody clicks. This school's ended on three different domains: the marketing site, a forms subdomain, and a separate store where you actually pay.
That's normal enough on the back end. It is not normal to the person holding the credit card. She researched a school with one name and is now being asked to pay a company with another, with nothing on the screen connecting them. One sentence would fix it. Nobody had thought to write it, because internally everyone knows they're the same business.
This is the thing analytics can't show you. Nobody files a support ticket saying "I got nervous at checkout." They just close the tab, and it looks like a conversion rate problem.
The work
The fix list came out ordered by result per hour of work, which put the phone link first. That was one character, in one template, and it reached 318 pages.
Then the four dead links got repointed. The printed phone numbers on the contact and county pages became real tel: links. The meta description that published a number reaching nobody was rewritten. California went back into the 122 state pickers that had lost it. Prices went onto the pages people actually land on.
About fourteen hours of work, quoted at a flat $2,000. That's above our published $1,400 audit and cleanup because of the scale. A local business with a handful of pages and a Google profile is the $1,400 job. Five hundred and nineteen pages across two template generations is a bigger one, and it gets quoted as such rather than run on the clock.
Within four hours of finishing, she was getting phone calls and form submissions again.
Not four weeks. Four hours. Because nothing was wrong with the demand. People had been searching, finding the school, deciding to call, and hitting a dead number. The whole time. The journey was the only thing in the way, and the moment it wasn't, the leads that had always been there arrived.
What it actually was
This is what surprised me, and it's the reason the fix was hours instead of a rebuild.
The site is well built. 3,527 images and only four missing alt text, which is better than almost anything I audit. Meta descriptions on every one of the 519 pages. Canonical and og:url agreeing across all of them, zero mismatches, which almost never happens. The whole crawl returned no errors. Somebody built this carefully.
And the correct version of nearly every defect in the report was already live somewhere on the same site. The working phone string was on 197 pages. The working URLs behind all four dead links existed. The full state list existed. The careful legal wording of the reimbursement offer existed, on the homepage, while 153 other pages said something else.
Nobody had broken anything. Things had been fixed one page at a time and never pushed to the rest, over years, by different people. That's what a broken customer journey usually is on an established site: not bad design, just the good version never reaching every page.
It also means the diagnosis is worth more than the design. You don't need a redesign. You need somebody to find where the site disagrees with itself.
Twenty minutes, on your phone, before you spend anything.
Don't open your homepage. Search for what you sell the way a customer would, and click your own result. Whatever page you land on, that's the page to judge.
Then, on that page only:
Do it on three or four pages that get traffic, not just the one you're proud of. Most people find at least one dead end, and the dead ends are almost never on the homepage. They're on the pages nobody reviews, which are the pages search sends everyone to.
Tell us what's on your plate. We'll come back with what we'd do first and what it would cost, usually same day.
Prefer to talk? Call 442-291-9337 or book a free call →