The Content Treadmill Stopped Working, and AI Search Rewards Provable Credibility Instead

For years the local marketing playbook ran on output. Ten blog posts a quarter, a couple of white papers, a lead magnet, repeat. That machine is sputtering. Businesses are watching their Google clicks slide and finding themselves absent from AI answers entirely, no matter how much content they push. What earns a spot now is evidence, the real, provable credibility a business can hand a search system, and the operators pulling ahead have stopped counting posts and started proving why anyone should choose their client.

Bryan Fikes is the Founder of Bonsai Marketing, where he builds AI search and automation systems for local service businesses like contractors, roofers, and electricians. Across 27 years he says he's worked with thousands of businesses and hundreds of verticals, and in an earlier conversation with this publication he introduced the so-what test, the filter he now runs every claim through. His read on the current moment is that the tools everyone shares have made raw volume worthless and differentiation the whole game.

"Everyone thinks they're going to prompt their way to success," says Fikes. "If everyone has the same tool, what differentiates you from everyone else?" The answer, in his framing, is proof that no competitor can copy, and getting there starts with putting a business's own facts under real scrutiny.

The so-what test goes deeper

Applied to marketing, the test asks what a fact actually earns. A business claiming 25 years and 100,000 customers is stating something true, Fikes says, but neither line tells a prospect why to care. The harder version of the test takes a statement like "this isn't working anymore" and drives into the next question. "The so-what test for a generalized statement like 'I've been in business for 27 years' is one thing," he says. "Now you're talking about a complex dive into that so-what factor: this isn't working, so now what are we going to do?"

That shift moves the work from activity to outcome, and it's why he warns clients off the volume reflex. He can churn out content at any scale a client wants, and it won't matter. "I can automate a hundred posts a day for a client. Are they actually going to be relevant and get found? Absolutely not," he says, because generic content lacks the signals that make a page worth surfacing to either Google or an AI model.

Feed the machine proof, not filler

The signals that count are the ones a buyer looks for when choosing someone: price, reviews, depth, and evidence of real expertise. Fikes builds content that starts from an FAQ and goes further, turning a single service question into a specific, substantive article that plugs in those customary signals. Skip them, he says, and you're left with filler that neither search engine will reward, which is why he leans on reputation over spend as the thing that actually moves a local business up the results, and treats how models surface a business as a credibility problem before a technical one.

Reviews, in his account, are the single most underrated asset, and the resistance to gathering them baffles him. He describes a roofer of 28 years whose business had slowly declined, sitting on seven reviews, afraid that asking clients might surface a bad one. "That's your biggest selling point," says Fikes, who frames a negative review as an opening rather than a wound. His best client ever, he says, was his harshest critic for a year before becoming a superfan, and the resolve to fix that person's problem is what earned him hundreds of referrals over his career.

One hub, many signals

Once he has a full picture of where a business genuinely extends its authority, the little league sponsorship, the leads group, the church donation, the work becomes routing those signals into the right places. The website stays the hub, and everything else feeds it. "Your website becomes the absolute hub for everything you're building as a marketing and outreach program," says Fikes, describing a system where a 30-second video of a community award gets threaded out to LinkedIn, Facebook, and YouTube, then connected back to the center.

The distribution map keeps widening, which is part of why he warns against the templated site builders common in home services and dental. AI reads a cookie-cutter site, sees fifty near-identical ones, and moves on, so that templated approach backfires on the very customer it was sold to. He points to Yelp's deal to become a preferred source inside ChatGPT as a sign of how fast the buckets shift. "That was very smart on their part, because now it made Yelp relevant again," he says, and a sophisticated owner now has to know they're allocating a slice of budget to each channel that matters rather than one.

AI search unlocks the map

The upside of the shift, Fikes argues, is that AI search loosens the geographic cage the map pack enforced. Google's local results effectively capped a business at a five, ten, or fifteen-mile radius. An assistant weighing the best answer to a query doesn't hold that same leash. "AI search kind of releases the bonds of geo-local," says Fikes. "Google said you only get to serve this radius. Now it's, did you want to own nine counties? You can build a strategy to do that easier."

That reach turns search into a budget decision with real money on the line, which is where his work bleeds into financial modeling. He connects a client's data down to what an average roof is worth and what a customer is really worth, so an owner shifting significant spend can make the call with evidence behind it. The stakes echo an earlier era, he says: these are the same weighty investment decisions owners once made leaving the yellow pages, and they need enough information to tie budget to outcomes before they touch a system that's already working.

Build it slow, skip the hacks

The transition to an agentic setup is where Fikes has grown most cautious over the past several months. His firm rule is not to rip out what a client already runs to bolt on something new. "Do not 100% unplug what you're currently doing to go plug into another system," says Fikes. The move from a WordPress site to an automated workflow that produces video and pushes it across channels happens in stages, plugging in a piece, unwinding, testing, and measuring before adding the next.

Rushing it, he warns, gets a business flagged. He likens an overnight jump from last place to first to a NASCAR car that suddenly leads after trailing all race, the kind of thing that gets you pulled over and inspected. "Google is as smart as you are. ChatGPT is as smart as you are. You don't have a new hack," he says. His closing advice is education: owners should do enough due diligence to become literate, gather multiple voices, and remember that under all the tooling the constant is the same one it's always been, telling the right story and delivering it well.