Semantic keyword clustering: the method that gained 1,200 positions for a West African SME

The "one keyword = one article" myth is dead

For years, SEO consultants sold the same thing: want to rank for "packaging supplier Dakar"? Write an article. Want "cardboard packaging supplier Senegal"? Write a second one. "Industrial packaging Abidjan"? A third one.

Result: blogs with 200 nearly-identical articles cannibalizing each other. Google can't tell which to rank, and ends up ranking none of them properly.

Since 2023, Google's BERT then MUM algorithms understand natural language. They automatically group queries that share the same semantic intent, even if the words differ. That's called semantic clustering, and it became the foundation of modern SEO.

The example of a wholesaler in Abidjan

A B2B food wholesaler contacted us in September 2025. 340 articles published in 4 years, traffic flat at 800 visits/month. We ran a semantic audit with Ahrefs and Surfer SEO.

Diagnosis: across the 340 articles, we identified 42 distinct thematic clusters. But within each cluster, 8 articles on average treated the same intent under slightly different angles.

Concrete example from the cluster "restaurant delivery Abidjan": 11 different articles, all stuck on Google pages 3-5, fighting each other. None on page 1.

The applied method

We proceeded in 5 steps over 4 months:

• Extraction: all keywords the site ranks for (1,847 keywords via GSC + Ahrefs)
• Automated clustering: grouping by semantic similarity via SerpRobot + manual review (3 consultant days)
• Consolidation: for each cluster, keep the best-performing article, merge content from others into it, redirect others via 301
• Enrichment: the winning article is expanded to 2,000-2,500 words, with FAQ, comparison table, examples
• Linking: clusters are interconnected through descriptive contextual links

Cost and return

6-month results:

• Traffic: 800 to 6,400 visits/month (x8)
• Top-10 positions: 38 to 1,247 keywords
• Qualified B2B leads from site: 4/month to 31/month
• Revenue attributed to SEO: 0 to 47 million FCFA over 6 months

ROI on investment: x18 in 6 months.

The tools we actually use

• Ahrefs ($129/mo): extract keywords the site already ranks for
• Surfer SEO ($89/mo): SERP analysis per cluster, semantic term suggestions
• AnswerThePublic (free up to 3 queries/day): search intent discovery
• Google Search Console (free): the underused gold mine, real visitor queries
• Sheets + GPT script: assisted manual clustering when budget doesn't allow Surfer

The mistake to absolutely avoid

Never delete an article without a 301 redirect. We've seen a client lose 60% of their traffic in 2 weeks because a vendor deleted 80 "duplicate" articles without redirecting. Google interpreted it as a site emptying out — authority dropped, rankings collapsed.

Why it's crucial in West Africa

The francophone B2B market (Senegal, Ivory Coast, Mali, Burkina) is far less saturated than the French market. On 80% of commercial queries, the top 10 results are poorly optimized, often foreign sites. Clean semantic clustering launches a local SME onto page 1 without expensive backlinks.

FAQ

How many keywords per cluster ideally?

Between 8 and 25 keywords per cluster, all sharing the same intent. Beyond that, split into sub-clusters.

Should clustering be redone yearly?

Light audit every 6 months, full rework every 18-24 months as SERP evolves.

Surfer SEO or Semrush to start?

Surfer for content focus, Semrush for broader vision (Ads, backlinks, competition). For an SME, Surfer + GSC are enough.

Does clustering work in Wolof or Bambara?

Not yet — BERT struggles with low web-volume languages. Target French + English for African commercial SEO.

---

We can audit your site and identify cannibalized clusters in 48 hours.

WhatsApp: +221 77 596 93 33 — or free quote: /en/free-quote