Resource Hub

What local service businesses actually need, and what some agencies are quietly selling instead.
By Justine Kingston  |  Just By Design  |  10 min read

The pitch you are hearing right now

A new pitch is circulating with local service businesses. The claim: your existing site, whether WordPress, Wix, or Squarespace, is fundamentally broken for AI search. The fix, conveniently, is a custom-coded rebuild on Next.js and Vercel, with schema injected by hand, ownership of the codebase, and no monthly retainer.

It plays well on a slide deck. There are real technical merits underneath it. And the conclusion most owner-operators are being pushed toward, rebuilding your entire site to be found by AI, conflates platform with strategy.

Those are two different problems. Treating them as one is what makes the pitch sound smarter than it is.

What AI engines actually care about

Strip away the marketing language and what AI engines reward is consistent across platforms:

• Structured data they can parse on the first crawl
• A clear entity graph linking your business, services, location, and credentials
• Server-rendered content that loads quickly and does not require JavaScript execution to read
• Verifiable identity signals across third-party sources (Google Business Profile, Apple Maps, Bing, BBB, licensing bodies)
• Citable content phrased like an answer to a real question

WordPress can deliver every one of those. It does not deliver them by default. Most WordPress sites are built badly, with heavy themes, plugin bloat, and schema that is either missing or stamped out by a generic plugin. That is a craftsmanship problem, not a platform problem.

The same is true in reverse. A custom-coded site can also be built badly. A Next.js build with wrong schema, no entity anchors, and no content cadence is just a slow-decaying app dressed up in modern technology.

The honest comparison is well-built versus poorly built. Not custom versus WordPress.

Four takes on the platform argument

Here is how I think about the tradeoff after twenty-five years of building websites for service businesses across Bend, Portland, Central Oregon, Lake Oswego, and Coeur d’Alene.

Two ways to think about it

Where custom code actually earns its keep

Custom Next.js builds make real sense in a narrow set of scenarios. App-like interactivity, configurators, multi-location franchises with thousands of programmatic pages, content publishers at scale, or businesses with an in-house engineering team that can maintain the codebase. None of those describe a five-person plumbing company in Bend or a real estate office in Coeur d’Alene.

For ninety percent of local service businesses, the right move is a well-built WordPress site with proper schema, fast hosting, edge caching, and a content engine that publishes consistently. That setup wins AI citations every time it is implemented seriously.

The remaining ten percent might need something custom. The job is to know which ten percent you are in before you sign for a rebuild, not after.

But what about the new sites winning AI search in weeks?

A version of this argument is starting to make the rounds. A new domain, a custom-coded site, a clean entity graph, and a few months later it ranks first in AI search panels above competitors with twenty years of history. The case study is real, and the insight underneath it is right.

AI engines do reason differently than Google’s classic algorithm. They reward entity coherence, the cleanness and machine-readability of your identity, more than they reward domain age and link velocity. A twelve-month-old site with a properly built entity graph genuinely can earn AI citations faster than a twenty-year-old site with messy schema and contradictory directory data.

That part is true. The conclusions most people are drawing from it are where the argument breaks down.

A single client in a hyper-specific niche winning AI citations is a successful pilot, not a repeatable methodology. The variables doing the actual work could be very low competition for that exact query set, clean entity work that would have produced the same outcome on any platform, model cold-start bias favoring newer pages with cleaner schema, or the stochastic variance of AI output on a single test. Niche competitiveness changes the timeline as much as platform does. A low-competition local query in a category with few established players is a different game than “plumber in Bend” or “real estate agent in Portland,” because AI engines blend authority signals with entity clarity. They do not ignore authority. They weight it differently than classic Google, and the right weight depends on the vertical.

Ranking first in Perplexity or being cited in Bing Copilot is genuinely valuable. The business question still missing from most case studies is how many real customers found the business through an AI surface, clicked through, and bought. For local service businesses, organic Google search still drives the overwhelming majority of lead volume. AI visibility is the next channel and you should build for it now, but it is not yet a replacement for everything else. And a single morning of screenshots is not durable visibility. AI outputs are stochastic, personalized, and shift as models update. What ranks first in an incognito test today can disappear next month. Durable AI visibility is a system that maintains entity coherence, content cadence, and third-party signals across years. Not a snapshot taken on a Tuesday.

Frequently Asked Questions