Conversion rate optimization: why does my site get visits but not enough calls?

Say your analytics show steady visits month after month but the phone doesn’t ring the way those numbers say it should. The first instinct is to go buy more traffic, and I’d look at the pages before spending a dollar on that, because more of the people reading them now showed up having already read about you, or about somebody else, inside an AI answer. Most sites like that bury the answer, show proof nobody can check or hide the next step, and none of it means much until you measure calls and forms against your own numbers from the same months last year.

What is conversion rate optimization for a service business?

For a service business it means more calls, form fills and booked consultations out of the visitors you already have, and you get there by changing what the pages say, what they prove and how easy they make the next step, then checking whether more people took it.

Your conversion rate is the people who took the action divided by the visits, but for an injury firm or a home builder or a dental group that action almost never is a checkout button. It’s a phone call or a consultation request, a lot of times a few days after the first visit, so the conversion you care about is a conversation with a real buyer. When one new client is worth a lot of money, two more calls a month off one page can pay for a lot of work, and I’d rather fix the five pages that get most of your visits than run a hundred little tests on button colors.

Why does my website get visits but not enough calls?

Usually it’s because the page people land on doesn’t answer their question fast, doesn’t give them proof they can check, or doesn’t make the next step obvious. Traffic is what you see in the report, and what happens after somebody shows up is where a lot of service firms lose the lead without knowing it.

People don’t read your pages the way you wrote them. The Nielsen Norman Group found 79% of its test users scanned any new page and only 16% read it word by word, and in a separate analysis it figured users read about 20% of the text on an average page, so if your best reason to call sits in paragraph six most visitors never see it.

Then there’s where the traffic goes. Plenty of firms get most of their visits from blog posts that answer a general question and just stop, with no path to the service page. A 2026 MySEOConsult study ran into this on the site it looked at, where organic search sent heavy traffic to informational blog posts, and the author put a big part of the gap between organic and AI conversion down to that landing-page mix. And there’s the boring stuff, an eleven-field form, a phone number you can’t tap, a page that crawls on a cell connection, gray text on gray. None of it looks like much from your desk, but every one costs you a few people who were ready to talk.

Do visitors who come from AI answers convert differently?

A lot of times they show up further along, cause the AI tool already summed up their options and named a few firms. What evidence there is points to fewer clicks but more motivated ones, though the studies are young, they don’t agree on how big the gap is and none of them is about your industry.

A Pew Research Center study tracked the browsing of 900 U.S. adults in March 2025, and people who saw a Google AI summary clicked a regular search result in 8% of visits against 15% when no summary showed up, and clicked a link inside the summary just 1% of the time. Fewer people click through when the AI answers first, and the ones who do have already been told something about you.

Google’s own page on AI features says clicks from results pages with AI Overviews are higher quality, but it doesn’t show the data behind that. The MySEOConsult study followed one B2B software company from July 2025 to June 2026, and AI visitors converted to a free trial at 8.79% against 3.54% from organic search. The author is careful to say it’s one site, one kind of conversion, mostly ChatGPT traffic and partly a landing-page mix effect, and I’d read it as a direction, not a forecast for your firm. MySEOConsult also notes some AI visits come in with no referrer and get counted as direct, so the AI number in your analytics is probably a floor.

Where they land matters too. SE Ranking’s July 2026 analysis, as reported by Hartzer, found 60% of AI-referred traffic landed on homepages against 17% from organic search, which makes sense since an AI answer names your company, not one of your service pages. If your homepage is a slideshow and a mission statement, the person who came in already leaning your way has to go hunting for a reason to call.

What should I measure before changing anything?

Track the actions that bring you clients and write down a baseline before you touch a page. That’s calls from the site, form submissions and booked consultations, set up as key events in Google Analytics 4 and broken out by landing page and traffic source, with a year of history if you’ve got it.

Calls are the one most firms miss, and if half your leads phone in and you only track forms your changes will look like they did nothing, so I’d get call tracking on the site number before anything else. Then sort the landing pages report by sessions, and the pages with lots of visits and almost no calls or forms are your first list. I’d also pull out referrals from chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com to see where those visitors land.

When you compare before and after, compare the same months a year apart. A dental group’s March isn’t its December, and going month to month can make a good change look bad. It’s the same reason my own guarantee compares the six months after launch with the same six months a year earlier.

Which page changes bring in more calls?

The ones that help somebody who’s scanning make up their mind faster, so the answer and the offer up top, proof a stranger can check, a short form and a page that’s easy to read on a phone, and most of it is rewriting and moving things around rather than a redesign.

Where the answer and the phone number go

The first screen of a service page should say who you help, what you do for them and what happens when they call, in the words a buyer would use. If people read about a fifth of the text, the top is the part they read, so your clearest sentence goes there with the phone number and the button. It’s the same structure that helps AI tools quote you, cause a page that opens with a straight answer is easier for a model to lift a clean sentence from, and I wrote more on that in the pillar on AI visibility and how AI tools pick who to name.

Proof a buyer can actually check

The Stanford Web Credibility Project built its guidelines from research with more than 4,500 people, and the first few say make it easy to verify your information, show there’s a real organization behind the site and make it easy to contact you. For a service firm I’d read that as named partners with their real credentials, a physical address, specific numbers you can stand behind and a source link for anything a buyer might doubt.

Reviews help when they’re real. Northwestern’s Spiegel Research Center found purchase likelihood for a product with five reviews was 270% greater than for one with none, and the lift was bigger on higher-priced products, 380% against 190% for cheaper ones. That’s online product buying, not hiring a builder, but I’d expect the pattern of bigger decisions leaning harder on proof to carry over. Spiegel also found purchase likelihood usually peaks at ratings between 4.0 and 4.7 and drops close to a perfect 5.0, so a wall of flawless five-star quotes can look less believable than an honest mix.

Don’t make any of it up either. The Federal Trade Commission announced a final rule on August 14, 2024 banning fake reviews and testimonials, AI-generated ones included, and it can carry real penalties.

How many fields the form should ask for

Give every page one main action and only ask for what you need to start the conversation. In a 2008 case study the web firm Imaginary Landscape cut its contact form from 11 fields to 4 and its form conversion rate went from 5.4% to 11.9% over two-month periods, with submissions going from 10 to 26 and no drop in lead quality. That’s one firm, a small sample and a before-and-after, not a split test, but nobody wants to type in their fax number just to ask a question. Name, phone or email and one box for what they need, then ask the rest on the call.

Speed and contrast on a phone

A 2020 Deloitte study commissioned by Google found a 0.1 second improvement in mobile site speed went with an 8.4% increase in retail conversions, a 10.1% increase in travel conversions and an 8.3% improvement in bounce rate on lead generation information pages. A tenth of a second won’t save a bad page, but a slow one can sink a good page. The W3C’s accessibility standard asks for a contrast ratio of at least 4.5 to 1 for normal text and 3 to 1 for large text, and a lot of stylish sites miss it with light gray copy on white, and if a buyer has to squint at their phone in a parking lot they’ll call whoever was easier to read.

Should I run A/B tests or just fix the pages?

For most service firms I’d fix the obvious problems first and keep A/B testing for the pages with real traffic. A split test only tells you something when enough people go through both versions, and a firm getting a few dozen leads a month can wait a long time for a result it can trust.

I’ve got nothing against testing tools, but a page that never says who it’s for or a form that asks too much isn’t a coin flip between two headlines. You fix those and measure against your baseline, and once the basics are solid on a high-volume page like the homepage, test bigger ideas like a different offer instead of a different shade of blue.

How does conversion work connect to AI visibility and SEO?

I treat them as one job. Search and AI answers decide whether a buyer hears your name and your pages decide whether that buyer calls, and if you only fix one side you either get seen by people who don’t call or build a great page nobody reaches.

Seer Interactive looked at more than 3,100 queries from June 2024 to September 2025 and found sites cited inside Google’s AI Overviews had a 35% higher organic click-through rate and a 91% higher paid click-through rate, while saying plainly it can’t prove the citation causes the higher clicks. Being named and being clicked tend to go together, and the click is only worth something if the page converts.

If you’re wondering whether search still matters at all, I covered it in is SEO dead because of AI, the writing side is in the guide to helpful content, the step-by-step version of this page is how to improve your website conversion rate, and getting named in the first place is in how to get your business cited by ChatGPT.

What it costs to wait

I’m not gonna tell you your leads dry up next quarter. What the evidence does show is that fewer people click through when an AI summary answers first, 8% of visits against 15% in Pew’s tracking, so every visitor who does reach your site is a bigger share of the buyers you were ever going to get, and leaving them on a page that buries the answer costs you more now than it did a few years back.

The common outcome is a slow leak, a few calls a month going to the firm whose page made the decision easier, and you never hear about them because the buyer who didn’t call doesn’t leave a voicemail explaining why. The rarer worst case is a competitor who gets named in the AI answer and also has the page that converts, and that’s a hard gap to close. Page fixes are some of the fastest changes you can make, though, so start with the five pages that get the most visits and the fewest calls.

How I know

This comes from my own ecommerce brand and not a client, so take it as direction and not a forecast for a law firm or a builder. Organic search brings that brand 40% of its revenue from 19% of its visits and organic visitors convert at 9.2% against 1.1% for paid social, though it also runs paid ads, so not all of its growth is organic. Its AI citations in the AI performance report in Bing Webmaster Tools went from 9 a month in April 2026 to 1,275 a month in August 2026. What I took from it is that people who found the brand by searching and landed on pages built to answer their question behaved a whole lot differently from people an ad interrupted. Those numbers are on my results, with how each one was measured, they aren’t typical results, and on a call I’ll show you the live Search Console and Bing reports behind them.

Your options for fixing a site that doesn’t convert

If you’ve only got a handful of pages and a strong writer in-house, start with the do-it-yourself fixes on this page, and you probably don’t need me.