A small accounting firm (made up, but you’ve probably met one like it) has held first place on Google for “bookkeeping for restaurants” for two years. The phone used to ring a few times a week from that one page. These days an AI Overview sits on top and does the explaining itself: rough costs, what to ask a bookkeeper, which software restaurants tend to use. The firm’s ranking hasn’t moved. The calls have slowed anyway.
Most ranking-factor articles skip over that wrinkle. Google’s systems still decide who appears, and in 2026 those same systems also shape which pages get quoted in the AI answers parked above the links. If you’re comparing SEO consulting services or trying to handle this yourself, the useful thing to know is which old signals still pull weight and which new shortcuts quietly hurt you.
A quick look under the hood
Google crawls pages, indexes what it finds, then serves results when someone types a query. During that last step, it weighs meaning, relevance, and usefulness, some of it judged page by page and some across the whole site.
There’s no official scorecard with fixed weights. Google has never published one, partly because the weighting genuinely shifts. A search about a tax deadline extension lives or dies on freshness, while someone asking how double-entry bookkeeping works couldn’t care less whether the page was updated last week.
The AI part is simpler than the industry chatter suggests. Google’s own documentation says there are no additional requirements to appear in AI Overviews or AI Mode, and no special optimizations either. A page needs to be indexed and eligible to show in Search with a snippet. Nobody needs a special AI file or new markup. Whatever helps you rank is, as far as Google says, what feeds those answers too.
Content still does most of the lifting
Anyone can write around a keyword in twenty minutes. Answering the worry behind the search takes longer, and that’s the work Google keeps rewarding.
Go back to our accounting firm. A restaurant owner searching for bookkeeping help wants to know who the service suits, how onboarding works, roughly what it costs, and whether the people behind it are qualified. Put those answers on the page, along with one real (even anonymized) client story, and it’ll outperform a long, generic explainer every time. Pages padded with the same phrase over and over rarely hold their ground for long.
Keywords come from the same place good answers do: your customers. Listen to what they ask on the phone, in emails, in reviews. Then let the title tag, main heading, subheadings, opening lines, and URL make the page’s purpose obvious. Related terms should show up because the topic needs them. If a stranger can tell what a page is for within about five seconds, Google probably can as well.
Links and reputation, the slow way
Links from relevant, respected sites still help Google find your pages and gauge how much they matter. The ones that last tend to be earned with something people want to point to, like an original survey, a handy calculator, or a quote a local journalist can use. Paid links meant to game rankings, and those endless link-swap emails, are where trouble usually starts.
You’ll hear E-E-A-T everywhere (experience, expertise, authoritativeness, trustworthiness). Think of it as Google’s description of what it hopes its systems reward, rather than a dial you can turn up.
Freshness and local details sit in a slightly different bucket. Updated information matters whenever it changes the answer, like rates, rules or deadlines. For businesses serving a neighbourhood, an accurate Google Business Profile, contact details that match everywhere online, and location pages with something real to say all help. Structured data can make a page eligible for rich results and helps Google read the content, though it guarantees nothing and Google doesn’t require it for AI features.
The tempting AI shortcut, and why it bites
Here’s the irony. The technology nudging clicks away from search results also makes it nearly free to publish hundreds of pages. Plenty of businesses have tried exactly that, with AI-written suburb pages, “ultimate guides” by the dozen and every keyword variation imaginable.
Google got ahead of it. In March 2024, it introduced a spam policy on scaled content abuse, covering large volumes of content made mainly to manipulate rankings, whether a machine, a person or a mix of both produced it. Its guidance says outright that using generative AI to churn out many pages without adding value for users may violate that policy.
Google is fine with AI-assisted writing that actually helps readers. What gets sites into trouble is quantity with nothing behind it. Sixty near-identical “accountant in [suburb]” pages look like a liability waiting to happen. One properly useful page on local bookkeeping, with real fees and named credentials, is the sort of thing an AI Overview might cite. In practice, winning content for AI search ends up looking a lot like content that was good for search all along.
So what’s a “quality ranking factor”?
You’ll see the phrase in SEO tools and pitch decks. It’s convenient shorthand, but Google doesn’t have a signal by that name. What Google does say is that content quality is central to its ranking systems, working alongside relevance and plenty else.
E-E-A-T isn’t one factor either, and the human quality raters Google employs don’t directly rank anyone’s pages. Their ratings help Google check whether its systems behave as intended.
A more practical test: open your most important page and read it like a skeptical customer about to spend money. Is it clear who wrote this and why they’d know? Are the claims sourced? Is anything stale? Would you trust it enough to pick up the phone?
Turning Ranking Factors Into a Plan with Rise Peak Digital
Start with the handful of pages that bring in actual business and leave the blog archive for later. Make sure they’re indexed and not tripping over technical problems like wrong canonicals, blocked files or painfully slow loading. Then put each page next to what people really type when they’re looking for that service. Quite often the page answers a question the business wanted to talk about, which isn’t the same as the one customers asked.
Rebuilding comes next. Swap outdated numbers for current ones, add details only you can provide (your process, how you price, what happened for past clients), and link related pages together sensibly. Earn links gradually, through work worth citing.
Measurement deserves a fresh look too. Search Console still reports impressions, clicks, click-through rate and average position, but AI Overview and AI Mode activity is folded into the ordinary “Web” totals. A steady position with fewer clicks might simply mean an AI answer now sits above you, and your page could be fine. For a sense of scale, Pew Research Center’s browsing study found people clicked a traditional result in 8% of visits when an AI summary appeared, compared with 15% when there wasn’t one.
Connect organic visits to leads and sales with analytics and conversion tracking, and be honest about each page’s job. Some pages warm people up months before they ever fill in a form, and understanding demand versus lead generation stops you from grading an awareness page on lead-gen numbers.
Some owners run all of this in-house. Others hand it off. Rise Peak Digital’s organic SEO services, for instance, pair classic search work with visibility in AI Overviews. Either way, the order of operations above holds.
One last question
None of the core signals have flipped. Useful pages, a site Google can crawl easily, earned credibility, and accurate local details still carry most of the weight, and Google says the same signals feed its AI features. The payoff has shifted, though. A first-place ranking may bring fewer visits than it used to, while a page that gets quoted might win someone’s trust before they ever click.
Which of your pages would an AI Overview actually want to quote, and which ones would it scroll right past?
Related: Technical SEO in 2026: Is Your Website Invisible to AI Search?
