← All articles

From 600 Technical Issues to a 72 Performance Score in One Month

Last autumn I audited the website of a leading driving school in Dubai. Solid brand, real demand, decent content — and search performance that didn't match any of it. The initial crawl explained why: 598 technical issues, including 76 hard errors, a performance score of 45, and an LCP of 3.3 seconds.

One month later: performance score 72, average position improved from 12.7 to 11.4, and monthly impressions up from 292K to 326K. Here's the anatomy of that turnaround — because the pattern repeats on almost every site I audit.

What the audit actually found

The scary number — 598 issues — breaks down less scarily once you triage:

• 44 broken internal links leaking authority and crawl budget into dead ends
• 25 hreflang conflicts making the Arabic and English versions compete against each other
• 30 pages missing H1 headings — Google guessing what pages were about
• 1,557 uncached JS/CSS files — the single biggest performance drag
• Duplicate and missing meta descriptions, sitemap errors, unminified assets

Notice what's not on the list: no penalty, no content problem, no mysterious algorithm issue. Just accumulated technical debt — the normal sediment of years of updates by different hands.

Triage: what we fixed first (and why)

The instinct is to fix errors in the order the tool lists them. That wastes the first month. We ranked by impact ÷ effort:

Week 1 — Caching and asset delivery. 1,557 uncached files was one server-configuration fix, not 1,557 fixes. Enabling proper caching and compression moved the performance score more than everything else combined. Biggest lever, smallest effort — always look for these first.

Week 2 — Hreflang conflicts. For a bilingual site in a bilingual market (Arabic/English in Dubai), 25 conflicts meant Google was routinely serving the wrong language version. Fixing reciprocity and self-references stopped the versions cannibalizing each other.

Week 3 — Broken links and H1s. Mechanical, boring, effective. Redirects for the 44 dead ends; proper H1s on 30 pages. No single fix here is dramatic — together they clean up how authority flows and how pages get understood.

Week 4 — Measurement. Re-crawl, re-test, document the delta. If you skip this step, you can't tell your client (or yourself) what worked.

What moved and what it means

Within the same month: impressions +11.6% (292K → 326K), clicks 16.7K → 17.5K, average position 12.7 → 11.4, and competitive non-brand keywords — "driving school", 18,100 monthly searches — climbing toward the top 5. Field-data Core Web Vitals took the usual ~28 days to reflect fully, which is normal: Google's field data is a rolling window, and patience there is part of the plan.

The lesson that generalizes

Almost every underperforming site I audit has the same shape: no single catastrophic problem, but hundreds of small ones compounding. And in almost every case, 20% of the fixes deliver 80% of the recovery — if you triage by impact instead of working down a tool's export alphabetically.

That triage is the real product of a good audit. The crawl takes hours; knowing what matters takes the 15 years.

Want the same diagnosis for your site? My Technical SEO Audit is exactly this process — fixed price, delivered in 10 days, with the prioritized roadmap included. Book a free call.