SEO execution means turning keyword research, on-site optimization, and content strategy into concrete, measurable actions. This page collects related practice notes, showing how these actions move forward in real projects and how their results are verified.
A keyword map decides which page should rank for which term. There is only one core rule: one keyword maps to one URL, and paths stay within three levels. Buying-intent terms go to product pages, comparison terms to selection pages, and educational terms to blog posts. Mismatches are the number one reason rankings stall.
Pick 3 to 5 direct competitors, export the keywords where you don't rank in the top 100 but they rank in the top 20, then score each one on search volume, commercial intent, and how hard the first page is to crack.
Branded search volume is an earlier signal of brand health than keyword rankings. A healthy baseline is three consecutive months of month-over-month growth, a 60%+ click-through rate on the top branded result, and an inquiry conversion rate typically 3 to 5 times that of generic keywords.
Writing structured data correctly is only step one. Debugging means clearing two gates: validate a single page with the Rich Results Test, then verify the whole site in the Search Console rich results report.
Commercial investigation keywords (vs, best, review) usually get only a fifth of the monthly search volume of informational keywords, but these visitors are already choosing a vendor, and their inquiry conversion rate is more than 5x that of an educational blog.
In the age of semantic search, models no longer count how many times a keyword appears on a page. They identify "entities" and the relationships between them. The focus of content optimization has shifted from "repeating a word enough times to hit keyword density"
Don't treat AI detection scores as a quality standard. Google never ranks by detector score; it judges whether content has unique value and credible evidence. Detection tools are sensitive to writing style, and factually accurate, well-polished human writing is often misjudged as AI.
A single long-tail keyword often gets only 10 to 50 searches a month. Giving each one its own page leaves you with dozens of thin pages. The right move is to group 30 to 80 long-tail keywords under one topic into a single hub page, build it out to 2,500 words or more,
E-E-A-T is not a direct ranking factor, but it shapes how evaluators judge content quality, which in turn affects rankings. This piece gives you a self-assessment table covering the four dimensions of experience, expertise, authoritativeness and trust; the five must-haves (author byline / About Us / contact details / privacy policy / third-party certifications); ways to raise credibility at the content level (cited sources / labeled data / expert review / update dates); and the criteria for judging that AI-generated content does not automatically hurt E-E-A-T.
Hreflang is the tag multilingual sites get wrong most often, and a single unclosed tag stops every language page from matching correctly. This guide covers how to choose between subdirectories, subdomains, and separate domains, how to use x-default, the principle of reciprocal references, five common mistakes (missing self-reference, wrong region codes, unclosed tags, 404 pages, conflicting declarations), and how to verify everything with the International Targeting report in Google Search Console.
What matters in B2B link building isn't volume, it's mentions from authoritative industry media and partners. This piece lays out a nine-week playbook (audit resources → build a target list → create content hooks → outreach → follow up → publish → monitor), compares the value of DR 70 media against DR 20 directory sites, gives three types of content hooks you can actually execute (data reports, expert commentary, tools and resources), and flags the three red lines: buying links, PBNs, and large-scale guest posting.
Core Web Vitals is one of Google's ranking factors, and since INP replaced FID in 2024, third-party scripts have become the biggest bottleneck. This article covers the three thresholds — LCP ≤ 2.5s, INP ≤ 200ms, CLS ≤ 0.1 — the common causes and fixes for each metric, a nine-week optimization cadence (measure → images → scripts → fonts → caching → CDN → re-test), and why you shouldn't rush to optimize if you get fewer than 1,000 monthly visits.
Structured data doesn't guarantee rich results, but without it you'll never get them. This guide lays out a Schema markup checklist for B2B websites from P0 to P4 priority (Organization/Product/FAQPage/BreadcrumbList/Article), the required fields and common mistakes for each type, how to validate with Google Rich Results Test, and alternative strategies now that FAQPage rich results have narrowed since 2023.
Accessibility isn't a charity issue, it's a compliance item in the procurement chain. For a B2B website, the right approach is to work in tiers: fix the five P0 items first under WCAG 2.2 AA, which takes about 8-20 hours of template-level work. Automated scanners catch only 30-40% of issues, so sign-off must include keyboard walkthroughs and screen reader spot checks, and should produce a conformance statement you can put in a vendor questionnaire.
Video earns its value from being watched; text earns its value from being found and cited. This piece compares the cost of three transcription methods, the four types of errors you must fix when proofreading, the rule for splitting a video into one conclusion per chapter, and how long it takes to turn a transcript into three content formats: subtitles, an article, and source clips. Turn videos with original methods or data into 600–900 word articles; for pure showcase videos, subtitles alone are enough.
Machine translation plus human post-editing can cut the cost of multilingual content to 30–40% of fully manual translation, but only if the termbase is built first: lock down 100–200 industry terms, and post-editing actually means something. Run it in five steps, drop sampling from 100% on the first piece down to 10%–20%, and set the pass mark at 4.0 on a five-point scale. Regulatory, warranty, quotation, and contract content stays out of this pipeline.
Once a site passes a hundred pages, it's time for a full-site audit. The output isn't a report but three tables: a keep list, a merge map, and a takedown redirect table. Tag pages against five switches and decide with four criteria. Typically 15%–25% of pages need merging or takedown. The whole process takes 6–8 weeks, batching no more than 30 pages per week, with checkpoints at 7 and 30 days after each action.
Whether a country page is worth building comes down to whether you have differentiated information. Pages that only swap the price, the contact person and the country name are the closest thing to a doorway violation. Only adapted pages with local case studies, certification specs, currency and shipping details are worth the investment. Start with 3-5 key countries, validate conversion, then expand. Rolling out dozens of country pages at once is a textbook doorway pattern, and if you get hit, the whole set gets demoted together.
The output of a gap analysis isn't a report, it's a 12-week schedule. Build a query set from 20 product terms plus 10 buyer-question terms, then check each one against the top 10 results for you and two competitors. Gaps come in only three kinds: no content, weak content, or wrong format. Rank what to fill first by search volume, commercial value, and how hard it is to produce, and start with the ones that are valuable and where competitors are weak too.
FAQ is the content asset you reuse more than any other: one answer serves search, AI engines, and your sales team at the same time. Collect questions from three channels, aim for 60 first, then dedupe and group them. Organize pages into three tiers by question count. The first sentence of each answer should stand on its own in 40–60 characters, with conditions and limits added after. Always phrase questions in the buyer's own words. Add new questions monthly and retire ones nobody asks every quarter.
White papers are the highest-converting content format for lead capture on B2B sites, but most fail because the topic is too broad, the research is cobbled together, and the form asks for too many fields. This piece lays out a six-step production process: a nameable problem, a 15–25 page length ratio, citation standards for four types of sources, a three-field lead capture form, four distribution channels, and the normal ranges and review cadence for three key metrics: download volume, form completion rate, and sales follow-up conversion.
Refreshing old articles recovers results faster than writing new ones, but most teams refresh the wrong ones. Judge by four signals only: ranking dropped off the first page, traffic down more than 30% over six months, content contains outdated data, search intent has shifted. Three ways to handle it: minor update if the data is stale, rewrite if the writing is dated, merge if it competes for the same intent as another article. After refreshing, check results at 14, 30, and 60 days.
When AI writes a summary, it grabs verifiable specifics, not adjectives. Think of your site content in three tiers: the fact tier has scope, conditions and numbers; the opinion tier offers judgments and trade-offs; the promise tier is guarantees, leadership claims and 100%. The first two can be excerpted; the promise tier almost never gets cited, and it drags down the credibility of the whole page. The smallest fix is to add a condition and a number to every adjective.
Cannibalization isn't judged by repeated keywords, but by overlapping intent. The most reliable signal is two URLs alternating between positions 5 and 20 for the same query, each with more than 30 impressions. There are only three ways to handle it: merge, split, or rewrite. The dividing line is whether the two pages can each hold a distinct intent. After merging, watch for 60 days and change only this one variable.
When AI judges whether a supplier is trustworthy, it does not look at adjectives like "strong and well-established." It looks for verifiable, specific statements: year founded, number of machines, certificate numbers, issuing bodies. This article sorts the evidence on trust pages into five categories and spells out what belongs on About, Team, Certifications, and Quality Control pages, plus four types of information you should not make public.
Most B2B case studies read like testimonials: adjectives throughout, nothing you can check. A case study that works only has to get three things right: cover all six sections, give every result a measurement basis, and anonymize enough that the customer will sign off. This article gives you the word-count ratios, four anonymization methods, and before-and-after rewrites.
Content is scattered points; internal links are the structure. Once you have a lot of content, the real bottleneck isn't that you can't write more — it's that new content can't get in and old content can't get out. Internal linking has to solve three things at once: let crawlers find pages, let authority flow, and let readers keep moving. Do five fixed things every time you publish, and you'll keep your internal linking debt under control.
The biggest waste on a website isn't paying too much to build it. It's having no one look after it once it's live. Content stops, rankings slide, nobody reads the inquiry form, and a year later the site is effectively dead. We break maintenance into six monthly tasks, four quarterly tasks, and three annual tasks: a fixed 4 hours a month, and every task leaves behind a record you can check.
Most overseas-facing sites only have an English version, yet buyers in German, Spanish, Portuguese and Arabic markets are asking AI their sourcing questions in their own languages. Language priority isn't about market size, it's about four variables: where inquiries come from, certification coverage, competitive density and content reuse rate. Doing two languages properly in year one beats spreading yourself across six half-finished ones.
A B2B email isn't a promo channel. It's a low-noise way to stay present. Going from zero to five hundred subscribers, a sensible cadence is one issue every two weeks for a full year. Each issue follows a fixed structure: one main point of view, two industry updates, and one downloadable. A healthy open rate sits between 20% and 35%, and the unsubscribe link goes in the footer.
WhatsApp is already the de facto first point of contact for business communication in Latin America, the Middle East, Southern Europe, and Africa, but it is not your personal WeChat. What really determines efficiency is not polished scripts, but three things: the customer leaves their number voluntarily, you respond within 5 minutes during business hours, and chat records can be saved into your CRM. Cold blasting strangers violates the rules and won't win you orders.
90% of multilingual site problems show up "after translation is done." Inconsistent terminology, metadata that was never rewritten, misconfigured hreflang, units and dates that were never localized, and no one checking after launch — miss any one of these five gates and all the upfront work on a language version goes to waste.
For B2B factories, YouTube's value isn't in view counts — it's in "long-tail search + credibility." A factory channel hitting 100,000 views over three years is normal, but the leads from that traffic are far higher quality than the same volume from social media. The key is the topic bank — lay it out across four categories, starting with 30 videos.
LinkedIn's role in B2B exporting isn't a traffic platform, it's a due diligence platform. What you can achieve in 90 days isn't a flood of inquiries, it's a credible trail that overseas buyers find when they search for you. Your first qualified inquiry usually comes from someone who saw your content and then looked you up on their own.
AI can handle about 70% of the work in content production, but three things must never be handed to it: factual accuracy, a point of view, and the details that come from experience. The test is one question: if this sentence is wrong, who gets hurt?
