The 30-day metadata rewrite: the fastest SEO win most sites skip.
Most SEO proposals start with new content. Twelve blog posts a quarter, a pillar page, a content calendar. That work matters, eventually. But it's the slowest thing on the list, and it skips the fastest: the pages you already have, which already rank for something, are almost always promising the wrong thing to the people who find them.
The clearest example I have is a finance client in one of the most competitive niches on the internet, up against the IRS, Fidelity, and Forbes for the same queries. No new content for sixty days. A full metadata rewrite, done carefully, took organic traffic up 363% month over month. The case study has the prompts. This is the process behind it, in thirty days, for any site.
Days 1–5: pull what Google already thinks you're about
Open Search Console. Export the last three months of queries with impressions, clicks, and the page each one landed on. Sort by impressions. What you'll find on almost every site: pages getting thousands of impressions and a handful of clicks, because the title Google shows doesn't match what the searcher typed. A page about "commercial roofing services" that's being shown for "flat roof replacement cost" and losing the click to a result that says those exact words.
Make a sheet: one row per page that matters, the top five queries it's shown for, its current title and description, and its click-through rate. That sheet is the whole project.
Days 6–10: decide what each page is for
For each page, pick the one query cluster it should win. Not every query it's shown for; the one it can actually answer best. If a page is being shown for two things it can't both be, that's a new page, and you write it down for later; you don't contort the existing one.
Then write the promise. A title is a promise about what's on the page, made in the searcher's words, in about 55 characters. A description is the second sentence of that promise, with the specific the searcher wants: a number, a place, a timeframe, a price range. "Flat roof replacement in Central New York, typical projects $40K–$300K, quotes in 48 hours" beats "Quality commercial roofing solutions" every single time, and it's not close.
Days 11–20: draft in batches, check every one
This is where AI earns its keep. Give the model your sheet, your rules, and twenty examples of the voice, and have it draft titles and descriptions for every page in batches of thirty. It'll produce in minutes what used to take a copywriter a week.
Then read every one. Not skim. Read. The model will occasionally promise something the page doesn't deliver, stuff two queries into one title, or invent a number. The job of the person is to catch those, and on a hundred-page site you'll catch a dozen. A rewrite that ships unchecked is how sites lose rankings instead of gaining them.
Days 21–25: ship it, with the structure to match
Push the new titles and descriptions live. While you're in each page, make sure the H1 agrees with the title, the first paragraph delivers what the title promised, and the schema says what the page is. The engines, including the AI ones, read those together; a title that promises a price range on a page with no prices gets ignored.
Days 26–30: measure the right thing
Don't watch rankings. Watch clicks and click-through rate, page by page, in Search Console, against the three-month baseline you exported on day one. Impressions may not move at all in the first month; clicks on the same impressions should. That's the metadata doing its job: the same people were already seeing you, and now more of them are choosing you.
Then keep going. The pages with the biggest gap between impressions and clicks are the next batch. The queries you set aside for new pages are the content plan, and now it's a plan with evidence instead of a calendar.
If you want the sheet built for you, the free SEO audit pulls your queries, finds the mismatched pages, and ranks the rewrite by expected impact. Send me your website and it's in your inbox within 24 hours.
