Technical SEO for businesses: the 12 things to fix first
The technical gaps to close before you spend a lira on content. A prioritised, measurable checklist you can work through in order.
There is one question to ask before allocating budget to content: is the content you publish actually being indexed? On most sites the answer is “partly”, and the missing part usually has a technical cause.
The list below is what we look at in a technical SEO audit, in priority order. The items at the top matter far more than the ones at the bottom — work through them in sequence.
Layer 1 · Accessibility and indexing
A defect in this layer makes everything else irrelevant.
1. Accidentally blocked pages
Check robots.txt and noindex tags together. The most common failure we see is a staging noindex shipped to production. Open the Pages report in Search Console, expand the “Excluded” group and read every reason one by one.
2. Canonical inconsistencies
Every page needs a self-referencing <link rel="canonical">, and that address must be byte-identical to the one in the sitemap. A trailing slash, a www prefix and a protocol difference each count as a separate URL.
<link rel="canonical" href="https://boygantech.com/en/services/seo-growth" />3. Sitemap and internal linking out of sync
Pages that appear in the sitemap but are linked from nowhere (“orphan pages”) get indexed late or not at all. A sitemap is a suggestion list; internal links are an actual signal.
4. A badly built multilingual setup
On a bilingual site this is the most expensive mistake. Every page must point to all language versions including itself, and x-default must be declared:
<link rel="alternate" hreflang="tr-TR" href="https://boygantech.com/hizmetler" />
<link rel="alternate" hreflang="en" href="https://boygantech.com/en/services" />
<link rel="alternate" hreflang="x-default" href="https://boygantech.com/hizmetler" />Reciprocity is required: if the Turkish page points to English but the English page does not point back, Google ignores the tags entirely.
Layer 2 · Speed and experience
5. Get LCP under 2.5 seconds
The largest contentful element is usually the hero image. Three steps usually suffice: produce it in a modern format (WebP/AVIF), set fetchpriority="high", and do not lazy-load it (putting loading="lazy" on the hero image is a common and costly mistake).
6. Remove unnecessary JavaScript
Every kilobyte of JavaScript that runs on the page lands in the INP measurement. On most corporate sites half the bundle is a slider used once or an analytics stack never used at all. Measure, then delete.
7. Self-host your fonts
A third-party font CDN means an extra DNS lookup and connection setup. Serve fonts from your own domain with font-display: swap.
8. Take layout shift (CLS) to zero
Give images width and height, reserve space for embeds, and place the cookie banner below the content rather than above it.
Layer 3 · Meaning and structure
9. Add structured data
Rich results raise click-through rate directly. The minimum set for a corporate site:
Organization— brand information, logo, social profilesBreadcrumbList— path display in search resultsService— for service pagesFAQPage— for FAQ sectionsBlogPosting— for articles
Validate with the Rich Results Test after adding it; malformed schema is worse than no schema.
10. Fix the heading hierarchy
There should be exactly one <h1> per page and it should state the page’s actual subject. <h2> and <h3> should descend without skipping levels. This serves both accessibility and content comprehension.
11. Build an internal linking architecture
Every service page should link to related services, related blog posts and case studies; every blog post should link to at least one service page. Link text should be the subject of the destination page, not “click here”.
Layer 4 · Measurement
12. You cannot improve what you do not measure
Rank tracking alone is not enough. At minimum, you need:
- Clicks, impressions and average position in Search Console
- Conversion tracking on organic traffic (form, call, purchase)
- Field-based Core Web Vitals monitoring (real user data, not lab tests)
- A weekly count of indexed pages
A realistic timeline
| Period | Work | Expected result |
|---|---|---|
| Weeks 1–2 | Audit, Layer 1 fixes | Indexing errors drop |
| Weeks 3–6 | Layers 2 and 3, structured data | Impressions rise, CWV turns green |
| Months 2–3 | Content cluster production | First rankings on long-tail queries |
| Months 4–6 | Internal linking, refreshes, expansion | Movement on competitive queries |
This timeline is typical, not pessimistic. If you have a proposal promising faster, ask which item was skipped.
A common mistake
Teams usually start at the end: they allocate budget to content production and skip Layer 1. The result is dozens of articles that are not indexed or that compete against the wrong page on their own site.
Order matters. Firm up the ground first, then build on it.
If you want to see which layer your own site is stuck on, take a look at our technical SEO audit — the output is a prioritised fix list with estimated impact written next to each item.
