Rewriting 100 Articles in One Pass: Content Ops for SEO and AdSense

PakSolarInsights covers Pakistan's solar transition — panel pricing, inverter sizing, battery comparisons, net metering. It launched with 5 articles, which is a brochure, not a content site. The push to 100+ articles, rewritten in one sustained AI-assisted pass, was where I learned that content at scale is an operations problem wearing a writing costume. The MDX parser side has its own post; this one is the editorial system.

Volume is a threshold, not a vanity metric

The blunt reality first: ad networks and search engines both have implicit thresholds. AdSense's most common rejection for young sites is "low value content," and a 5-article site triggers it almost regardless of article quality. Crossing to 100+ substantive guides (each its own route under src/app/blog/) changes the site's category — from "someone's blog" to "a resource" — in both evaluations. Quality still decides the ceiling; volume decides whether you're evaluated at all.

The AI-assisted part is worth being plain about: drafting, keyword expansion, and SEO tweaks were done with Claude in the loop, with a human owning topic selection, facts (local prices, local regulations), and final edit. A hundred genuinely useful articles in weeks instead of a year is exactly what this workflow is for — and the next four sections are what kept it from producing a hundred pieces of slop.

The atomic summary: answer first, earn the read

Every guide opens with a bold two-to-three-sentence atomic summary that directly answers the title's question — how many panels for a 1-ton inverter AC, which battery type for which budget. Sizes, numbers, the verdict. Upfront.

It serves three readers at once: the skimmer gets the answer (and trusts you more, not less, for it); Google gets a featured-snippet-shaped block, which is what position zero is extracted from; and AI search engines get a clean, citable claim — the GEO logic applied per-article. Burying the answer to maximize scroll depth is a 2015 strategy; in the snippet-and-AI-answer era, the answer is the marketing.

Title–H1 alignment, exactly

The unglamorous fix with outsized impact: the listing data (articles.json, which drives the blog's navigation and filters) had titles that roughly matched each article's <h1> — "Lithium-Ion vs Tubular: which battery fits a Pakistani home?" linking to a page headed "Lithium Battery vs Tubular Battery: Which Is Best for Pakistani Homes in 2026?"

Close enough for a human; not for the machines. Mismatched link-text vs. destination-heading reads as a bait-and-switch signal to quality systems and muddies which string ranks. We aligned every listing title to its H1 exactly — and because articles.json is the single source feeding both the UI and metadata, the alignment is now structural rather than aspirational. If your CMS lets the link text and the headline drift, eventually they will.

E-E-A-T: stop inventing authors

Early articles carried plausible-sounding individual bylines. For a site without real public author profiles, that's worse than nothing — unverifiable persons are an anti-signal under Google's E-E-A-T framework, and one reverse-image-searched headshot from undermining the whole domain. We replaced every byline with "PSI Editorial" as an Organization in both the data layer and the BlogPosting schema. An honest editorial-entity claim beats a fake personal one; authority you can't substantiate is a liability, not a decoration.

Freshness got the same consistency pass: articles titled "…in 2026" carried older datePublished values — another small contradiction machines notice — so dates were normalized to match the catalog's actual update.

Layout as an ad-network contract

AdSense injects ad units between your article elements. Default prose spacing lets headings crowd the injected ads — a policy-compliance risk, not just an aesthetic one. Explicit margins (5rem above h2, 4rem above h3) guarantee separation wherever an ad lands, and clearing every React key warning keeps the console clean during automated quality crawls. When a third party renders inside your layout, whitespace is part of the deal.

The system, compressed

Volume past the threshold → answer-first structure → exact internal consistency (titles, dates, schema) → honest entity-level authority → ad-safe layout. None of it is clever; all of it compounds. Content sites look simple until you're touching 100 articles at once — and the ones that win are the ones where touching 100 articles at once is routine.