Does schema markup help Charleston local SEO?

Twenty-nine lawn care companies just settled an argument that has been running in SEO for about a decade, and the answer is going to annoy roughly everybody who has ever sold you a website.

Schema markup does not move your Google ranking. Not organic, not Maps, not the local pack. It was tested properly this spring on real service businesses, and the result was a flat zero.

I want to be careful here, because this is the kind of finding that gets flattened into a bad headline and then a Charleston contractor rips working code off his site. So let me walk through what actually got tested, what the second study found, and what I still do on every site we build.

The test that used actual service businesses

Jake Hundley at Evergrow Marketing ran it and published on June 3. What makes it worth your time is the setup: 29 domains covering 36 locations, all of them landscaping and lawn care companies. Not SaaS blogs. Not ecommerce. Guys with trucks.

Thirteen domains (18 locations) sat in the control group and got nothing. Sixteen domains (18 locations) got LocalBusiness schema added. Five weeks of benchmark, then five weeks of test, February 27 to April 4. Rank tracking through Moz Pro, Maps and AI tracking through LocalFalcon, results run through a one tailed Welch's t-test with a 90 percent confidence bar before anything counted as real.

He also had the methodology reviewed before he ran it, by Jarno Van Driel (a Schema.org contributor since 2008), Alex Moss and Carolyn Shelby at Yoast, David Hunter at LocalFalcon, Darren Shaw at Whitespark, and Joy Hawkins at Sterling Sky. That is most of the people who would normally show up in the replies to tear a schema study apart, brought in before publication instead. I respect that a lot more than the usual "we saw a 47% lift" case study with no control group.

The results

On search rank, nothing. Google, Google mobile, Bing, and Yahoo all came back under an 85 percent confidence level, which is statistician for "this is noise."

On Google Maps, worse than nothing. The control group that did nothing at all improved by 0.22 positions. The test group that added schema declined by 0.62 positions. Share of Local Voice showed no significant movement either. Adding LocalBusiness schema did not help these companies rank in the map pack, full stop.

Then ChatGPT.

Position improved by 3.33 spots at 92.91 percent confidence. Share of AI Voice went up about 10 percentage points at 91.51 percent confidence. Both cleared the bar. Google AI Mode, AI Overviews, Gemini, and Grok all came in under 90 percent confidence, meaning no detectable effect.

So: nothing for Google, something for ChatGPT, and only ChatGPT.

Except a much bigger study says the ChatGPT part is noise too

Here is where it gets messy, and I would rather show you the mess than pick the study that supports whatever I am selling.

Ahrefs published the other side of this on May 11. Louise Linehan and Xibeijia Guan tracked 1,885 pages that added JSON-LD between August 2025 and March 2026, matched each one against three control pages that never added schema (4,000 controls total), and measured citations 30 days before and 30 days after.

AI Overviews went down 4.6 percent, which was small but statistically significant. AI Mode came in at +2.4 percent and ChatGPT at +2.2 percent, both close enough to zero to be nothing. Ryan Law, who reviewed it, put it on X about as bluntly as you can: schema shows up more often on heavily cited pages, but that is correlation, because schema also shows up more often on pages that are just well built in general.

At this point a lot of people are thinking the two studies cancel out and the whole topic is a wash. They do not cancel out, and the reason is buried in the Ahrefs caveat.

Every page in the Ahrefs sample was already getting cited heavily by AI. The authors say so themselves. That study answers "does schema push an already visible page higher," and the answer is no. It does not answer "does schema help a page that no AI system has ever noticed get parsed and understood in the first place." Different question. The Evergrow sample, a pile of lawn care companies in secondary markets, is much closer to the second question, and that is where the ChatGPT lift showed up.

If you are a Mount Pleasant landscaper with no AI visibility at all, you are in the Evergrow sample, not the Ahrefs one.

What I actually do, and why I have not changed it

We put structured data on every Charlestowne build. LocalBusiness, Service, FAQPage, BreadcrumbList, the whole stack. I have not removed a line of it since reading either study, and I am not going to.

Not because I think it is a ranking factor. It is not. That question is now answered and I would tell any client who asks that schema markup will not move them up the map pack one position.

I keep it because it is the cheapest legible description of your business that exists. It costs an hour once. It never needs maintenance. It makes your name, address, phone, hours, service area, and price range machine readable in a format every retrieval system already parses. And the only measurable effect anybody has found in a controlled test was on ChatGPT, which is the surface where Charleston service businesses are the most invisible right now. I went through that visibility gap in detail in why Charleston's map pack winners are invisible in AI, and nothing in these two studies changes that picture.

What I will not do is charge you for it as a growth strategy. (If an agency's proposal has "schema markup implementation" as a line item with a monthly number attached, ask them which study they are basing that on. Then watch what happens.)

Where your effort should actually go

If schema is a one hour job and not a lever, the honest follow up question is what is. Based on everything I have tracked this year for Charleston service businesses, in order:

1. Reviews with recency, not just volume. This is still the loudest local signal and the one most owners neglect. I broke down the recency shift in why fresh Google reviews now beat review count.
2. Being on the lists other people write. Ahrefs' own citation research found "best of" listicles are the single most cited page type in AI answers, and you do not own those pages. You get on them.
3. NAP consistency across every profile you have. Schema that says one address while your Yelp says another is worse than no schema. Worth checking whether the agency you hired is even in Charleston while you are at it.
4. Pages built for how people actually search. A page about hail damage claims in Summerville beats a page about "roofing services" every time. That is most of what we do on the Charleston SEO side.
5. Then schema. Once, correctly, and then stop thinking about it.

Do this week

Open your site's LocalBusiness JSON-LD and check three fields against your Google Business Profile, character for character: business name, street address, phone number. Not "close enough." Identical. Then run the page through Google's Rich Results Test and confirm it parses clean.

That is it. That is the whole job. If the fields do not match, fix them, because inconsistent structured data is one of the few ways schema can actively work against you.

And if somebody quotes you a retainer to keep doing that every month, send them this post.