Four Years to Build Search Authority and Only Six Months to Lose It.

A major financial company's search decline offers a warning for brands replacing editorial investment with cheaper content.

I spent years running a high-volume content program, so I'm not going to argue that

publishing a lot is inherently bad. At our peak, we produced roughly 100 automotive stories a month for a single client, and every one of those stories had to meet the same editorial standards, whether it was the fifth story we published that month or the hundredth.

What made that volume work was pure operational rigor. Experienced, bylined journalists wrote comparisons, maintenance and shopping explainers, along with industry articles, and a full staff of copy and line editors checked every piece before it went live. The work was built around the questions buyers were actually asking about their vehicles, from routine maintenance to shopping for the next one. It's the same playbook brands are now running to earn placement in AI Overviews, only we started running it before AI Overviews existed. Producing at the pace we were overseeing while holding those standards required real investment.

That investment is exactly what many brands are now tempted to skip. Generative AI has made it easier than ever to publish a lot of content quickly. Producing 100 stories is no longer particularly difficult. Producing 100 stories worth reading is.

The publishing industry is starting to bet on the harder path. In the Reuters Institute's 2026 trends report, a net 91% of the publishers surveyed said they plan to invest more in original investigations and on-the-ground reporting, which is the type of work AI can't easily summarize or replicate.

I have a case study that shows what happens when a brand goes the other way. It only came into view after our relationship with the client ended.

A major financial services company spent years earning Google's trust as a source of automotive information. That trust was earned through thousands of pieces of editorial content reported and written by credentialed automotive journalists. When that work slowed down, the traffic followed it out the door. More than a year later, it hasn't come back.

Google doesn't hand out authority to banks.

The company is one of the largest auto lenders in the country. That size bought it a lot of things, though credibility with search engines as an automotive publisher was a serious challenge.

Google treats money and major purchases as high-stakes territory. A car is the second-largest purchase most households make, and a lender writing about which crossover to buy has an obvious commercial interest in the answer. Search engines are built to be skeptical of exactly that setup.

Reviewing SEMrush's history shows how long the skepticism lasted. From 2017 through 2021, the site's estimated organic traffic barely registered. The increase started in 2022 and reached roughly 570,000 monthly organic visits by spring 2023. It kept building until it peaked near 950,000 in the summer of 2024.

That took years of consistent, reported, accurate work. Rankings for automotive terms that a bank had no historical right to own came one story at a time. The site even absorbed a sharp drop in 2023 and recovered within a couple of months, which is what an established publisher looks like when an algorithm update rolls through.

The cost of pulling back

Traffic had already eased from its 2024 peak, as it had for nearly every publisher during a stretch of aggressive core updates and the rollout of AI Overviews. Even so, the site held its ground through the first half of 2025, with estimated monthly organic visits between roughly 600,000 and 700,000.

Traffic Analysis via SEMrush

Then our engagement ended abruptly in the second quarter. Weeks later, Google rolled out its June 2025 core update, which ran from June 30 to July 17. It was the first major quality evaluation the site faced without Telemetry's editorial program behind it. Estimated organic traffic fell to about 410,000 by August and to about 340,000 by September. By December, it was about 260,000. It has hovered between 200,000 and 250,000 through 2026.

Keyword visibility tells a similar story. Just before the update, the site ranked for roughly 440,000 organic keywords. As of September 2026, it ranks for about 160,000, a drop of more than 60%.

The sharpest losses were in AI Overviews. While the publishing volume was intense, it resulted in 160,000 keywords earning the site a spot in AI Overviews or another search feature. Today, that number is under 30,000. That is the shelf space where search engines and AI models decide which sources are worth repeating, and the site lost more than 80% of it.

Search engines and AI models have told us what they reward

One site's charts can be dismissed as an anecdote, but the broader evidence is harder to dismiss.

Google has been explicit about where it's headed. After its March 2024 core update, the company said searchers were seeing 45% less low-quality, unoriginal content in results. That update also folded Google's helpful content system into its core ranking systems, so every core update since has been, in part, a quality review.

In January 2025, Google revised the guidelines its human search quality raters use. Pages whose main content is largely copied, paraphrased or AI-generated with little effort, originality or added value are now supposed to get the lowest rating. Raters don't set rankings directly, but their guidelines describe what Google's systems are built to find.

The AI platforms appear to agree. In an October 2025 study, the research firm Graphite found that 86% of articles ranking in Google Search were human-written, and that 82% of the articles cited by ChatGPT and Perplexity were human-written. When AI-generated articles did show up in Google, they ranked lower. Only 7% of the articles in the No. 1 position were AI-generated. That held even as AI-made articles had come to rival human-written ones in sheer volume across the web.

Graphite was careful to note it did not test AI-assisted work with heavy human editing, and I'd make the same distinction. The tool is beside the point. The question is whether a piece carries a certain level of trust that a reader can't get anywhere else. Large language models are built to summarize what the web already agrees on. The sources they cite are the ones that did the work of getting it right in the first place.

Google's own documentation notes it can take months for its systems to recognize that a site is consistently producing helpful content. In my experience, that recognition runs in both directions. Authority builds slowly, and when the work behind it goes away, the systems notice.

What I'd tell any brand publishing in automotive

If you're a lender, an insurer, a retailer or a supplier trying to become a credible voice on cars, understand what you're signing up for. Google won't take your word for it. Neither will ChatGPT. You earn that standing one accurate, well-reported story at a time, and it can take years.

That standing has real value, and it can start disappearing surprisingly quickly when you stop investing in it. You can build it with a modest publishing cadence, as long as every piece holds up. And once it's lost, you can't simply publish your way back overnight.

This site spent the better part of four years earning its place in automotive search. It gave back most of that ground in about six months. I take no pleasure in pointing that out. I spent a long time helping build what those charts used to show, and I still think the industry needs more of it.