Let me set the scene honestly. The site had 140 posts, about 9,000 monthly organic visits, and a decade of accumulated neglect. Half the posts hadn't been touched since they were written. Some were written before facts changed and still confidently claimed otherwise. Nothing had been penalized. The site was just… tired.

Here's what actually moved the needle, in order.

Week 1-2: the pruning spreadsheet

I exported every URL with impressions, clicks and position from Search Console. Anything with fewer than 10 monthly impressions and no rankings inside the top 50 went into one of three buckets: update, merge, or delete. That sounds aggressive until you see the numbers — 41 of 140 posts fell into those buckets.

Nine got deleted outright (duplicates, dead topics, posts I'm not proud of). Seven were merged into stronger pages with redirects. The rest went on the refresh list. Within two weeks of the deletions, average crawl stats improved noticeably. Google stopped wasting time on the junk and started revisiting the good stuff more often.

Week 3-9: twelve surgical refreshes

I refreshed two posts a week. Not "changed the date" — properly: new sections answering questions the post ignored, updated screenshots, removed steps that no longer exist, tightened intros so the answer lands in the first screen. Each refresh got one internal-link push too, with descriptive anchors from three or four stronger pages.

By week six, the first refreshed post had jumped from position 17 to position 6. Two more followed in weeks eight and nine. Refreshed pages, on average, outperformed brand-new pages I'd written that quarter — by a lot. The audience was already telling Google these URLs mattered. The updates just gave the signal weight.

Week 10-12: compounding

With refreshed pages climbing, I added "related questions" sections mined from People Also Ask. Cheap to build, and they pulled in long-tail queries the main content never targeted.

The final number, day 90: 18,400 monthly visits. Double. Zero new backlinks. Zero design work. The whole project was maybe 40 hours. If your site has more than a hundred posts and a year of silence, this playbook will probably work for you too — it's just not glamorous enough to sell courses.