Google Timed Itself. Your Page Edit Is Thirty Days Behind.

Gary Illyes stood up in Barcelona on October 2 and put numbers on the clock. Google almost never does that.

At Search Central Live Deep Dive Europe, Illyes showed Google's own internal timing data for crawling, indexing and serving. Fastest, typical and slowest for every process. John Campbell of ROAST was in the room and published the tables that afternoon. Search Engine Journal picked them up two days later. Google's own slides still aren't on the Search Central events page, so Campbell's recap is the record.

Two lines in those tables. They should change how your brand's website works.

Discovery of a new URL: about 20 hours, typical. Refresh of a URL Google already knows: about 30 days, typical. Same worst case on both, written in the recap as "weeks to never."

Publish a page. It gets found tomorrow. Edit that same page and you're waiting a month for a re-read that might never come.

Every food brand does this backwards

The sequence goes about like this. The formula changes. Or the co-packer changes, or the oil changes, or a certification finally lands after eight months of paperwork. Somebody opens the product page, swaps the paragraph, updates the ingredient list, drops in a new photo. Then they post it to Instagram and wait.

The Instagram post does its work in forty-eight hours and dies, which is what Instagram is for. The website edit just sits. Google already has that URL. It has a refresh schedule for it built on how often that page has historically changed, which for a brand product page is close to never. Thirty days is the typical case. That number is Google's, not mine.

What you needed was a page for the shopper typing your brand name next to "seed oil." That page doesn't exist as its own URL. It's one sentence buried on a page Google has no reason to look at twice.

The rest of the clock

None of the other ranges are fast either.

A canonical change runs one to three weeks, months if your signals conflict. A site move runs one to three months, six months to over a year at the slow end. Title and snippet changes are the quick ones at one to two days, and even those stretch to several weeks. Core update recovery is three to six months typical, and six months to a year if you end up waiting on the next core update.

Campbell flagged the detail most people read past. The word "never" shows up five times in the slowest column. Three of those five have "quality" in parentheses next to them: sitemap processing, end to end indexing, structured data updates. Google is saying out loud that a clean technical fix does not buy an index slot for a page that doesn't deserve one.

That lines up with the other thing Illyes said on that stage. Google filters out 40 billion spam pages a day, and in his telling scaled content has now passed link spam as the bigger problem. If your product descriptions got bulk-generated last quarter because somebody found a tool that would write two hundred of them over a weekend, that is the category he named.

Duy Nguyen defined it right there in the room. Effort, originality, talent or skill, accuracy. He's a search quality analyst at Google, and his good version was a shoe review that ran its own measurements, shot its own photos, and did the research. Four criteria. Three of them just check whether somebody actually did the work.

What my own archive taught me about the slow half

gallucci.net has been mine since 1996 and there are 435 posts on it. I publish to it every day, so I've been running a live experiment in this asymmetry whether I meant to or not.

New work gets found fast. Thirty years of history plus a daily habit trains Google to keep coming back, and it does. Going backward is the slow half. Every time I've gone into the archive to fix an old post, correct something that went stale, repair a link that rotted, that work has sat quiet a good while before search caught up to it. I spent a long time assuming that was Squarespace being Squarespace. It wasn't. It's the thirty day refresh line, and now somebody has written it down.

Publish, don't revise

The website is a place you publish. It's not a document you keep editing. Ship the page, then leave it alone.

When the claim on the bag changes, give the new claim its own page with its own date, and point the old page at it instead of quietly rewriting the old page in place. New URL, roughly 20 hours into the queue. Edited paragraph, roughly a month.

Build the page before the thing happens. New SKU hits shelves March 1, that page needs to already exist in January carrying whatever is true in January. You're buying crawl time. It's the one input in this business nobody will sell you.

Run your claims as a calendar. Non-GMO, no seed oils, single origin, family owned, whatever you print on the front panel gets its own page and its own proof. I've written before about building a proof post for every claim on the bag, and this is the search-side version of the same discipline. It's also a content strategy for food and beverage brands that survives contact with Google's actual timing, which is more than most calendars manage. Most of what a food and beverage marketing agency hands you in month one is a calendar. Whether the pages on it are new URLs or edits to old ones never comes up in the kickoff meeting, and it decides when any of it shows up.

I laid out the version that runs a whole content operation, not one page. It's an HACCP plan for AI content. Control points, not vibes.

Stop blaming the social team for the dip

Organic traffic falls and the meeting goes looking for somebody to explain it. With these ranges published you can date the cause instead of defending the channel. A site move unsettled at ten weeks is normal. A core update recovery at month two is normal. A page you edited three weeks ago that hasn't moved is normal. None of those are a social problem, and none of them get fixed by posting more.

Attribution is getting easier in the same month, which helps. Google started tagging Gemini's outbound links a few days ago, so traffic that had been piling up in direct now has a name on it. Better numbers and honest timelines at once is about as good as this job gets.

One more from Barcelona. Pablo Pérez, a research manager at Google, showed UK search interest for "is [brand] legit" climbing steadily since 2020 and hitting its highest point in 2026. People check whether you're real before they buy. If the page that proves you're real is a paragraph you edited last Tuesday inside a page Google reads once a month, you are invisible at the exact moment somebody wanted to believe you.

Open your product pages and find the last claim you changed. See whether that change lives on its own URL with its own date, or whether it got swallowed into a page that was already sitting there. If it's the second one, you now have a rough idea how long it's been in the dark.

I build the claim-to-proof map and the publishing calendar that goes with it for a handful of food brands a quarter. If yours is next, message me on LinkedIn.