What We Did
30 days. 176 service reviews, 94 comparisons, 32 guides. 6 languages. ~1,800 published pages.
Recommended platforms: awin.com, cj.com
This wasn’t just a content experiment. We were testing a hypothesis: can systematic, well-structured content about SaaS services generate measurable affiliate revenue? The answer is yes — with nuances.
Part 1: What Works — Formats
Comparisons Are the Conversion King
Comparison articles (Stripe vs Paddle, Midjourney vs Leonardo AI, Bubble vs Adalo) showed the highest buyer-intent density. Someone searching “X vs Y” is already in the decision phase. You don’t need to explain what a payment gateway is — they’re choosing between two specific options.
Metrics: CTR on affiliate links in comparisons is 2-3× higher than in reviews. Average time on page: 4:20 (vs 2:50 for reviews). People read carefully.
Verdict: comparisons are the most effective format for affiliate content. If your content budget is limited, start here.
Guides Bring Traffic, Not Conversion
Guides (how-to-choose-*) attract organic traffic but convert worse on affiliate clicks. The reader comes for knowledge, not to buy.
Solution: guides should funnel to comparisons and reviews via internal links. Don’t try to sell directly in a guide — it’s the top of the funnel.
Reviews Are the Foundation
Individual service reviews are the basic building block. Without them, comparisons have nothing to link to, guides lose depth.
Mistake we almost made: templated reviews. If all reviews look identical (“Pros”, “Cons”, “Verdict”), readers tune out by the third article. We introduced style rotation (analytical, practical, narrative, technical, problem-oriented) — time on site increased by 22%.
Part 2: Categories — Where Is the Money
High organic traffic categories:
- AI services — the main search traffic driver. AI hype keeps search demand at peak.
- Payment systems — stable evergreen traffic. People always search for payment solutions.
- No-code platforms — growing trend, +40% YoY demand.
High affiliate revenue categories:
- AI writers (Copy.ai, Jasper, Writesonic) — high commissions (30-45%) + growing market = maximum RPM.
- Marketplaces (Bonanza, Mercari, Zazzle) — % of product sales, higher average order value than SaaS.
- VPS/Hosting — low commission ($25-100/sale) but high conversion (reader is often ready to buy).
Surprise:
The “Payment Systems” category showed high traffic but low affiliate conversion. Reason: people read to understand the topic, but buying a payment gateway is a high-inertia decision (integration takes weeks). Affiliate links get clicked, but purchases are delayed. Attribution window: 30-60 days, not 7-14 like AI writers.
Part 3: Multilingual — Is It Worth It
The hypothesis: content in 6 languages = 5-6× addressable audience. Reality is more complex.
What worked:
- German (DE): strong organic traffic. German-speaking audiences actively search for SaaS solutions in their native language. Lower search competition than English.
- Spanish (ES): huge audience (500M+ speakers), rapid growth.
What didn’t work (yet):
- Chinese (ZH): near-zero traffic. Baidu doesn’t index sites without an ICP license. Chinese audiences use Baidu, not Google. Needs a separate strategy.
- French (FR): average results. French users prefer .fr domains.
Recommendation: base multilingual strategy: EN + DE + ES. FR if you have resources. ZH only after understanding how to attract Chinese traffic.
Part 4: SEO Lessons
What worked:
- Cluster structure. Hub pages (category overviews) + spoke pages (individual reviews) + internal linking. Google approved: indexing took 3-5 days, not 2-3 weeks.
- Featured images and alt-text. Every article with a cover image and meaningful alt-text. +10-15% CTR in Google Discover.
- FAQ sections at article end. 40% of search queries are questions. FAQ sections capture featured snippets.
What didn’t work:
- Over-aggressive internal linking. First versions had 8-10 “read also” links per article. Google may see this as link scheme. Reduced to 3-4 relevant ones.
- Cross-language content duplication. Hugo generates hreflang tags, but if the translation is too close to the original, Google may flag it as duplicate.
Part 5: Technical Lessons
The 30-day experiment proved: the content pipeline must be semi-automated:
- Article generation via AI (saves 80% of time)
- Manual verification: facts, figures, relevance
- Automatic deployment via Git + Hugo
- Automatic cover generation
Bottleneck: covers. Generating 9 covers daily took more time than writing 9 articles. Solution: parallel cover generation alongside translation, not sequentially.
Git as the source of truth: all articles in Git. Complete change history, rollback capability, automatic deployment via post-receive hook.
Part 6: Affiliate Revenue — The Reality
| Category | RPM (per 1000 visitors) | Best Program |
|---|---|---|
| AI Writers | $15-25 | Copy.ai (45%), Jasper (30%) |
| VPS/Hosting | $8-15 | Kamatera ($44/signup) |
| No-Code | $5-12 | Bubble (long deal cycle) |
| Marketplaces | $10-20 | Zazzle (% of sales) |
| CPA Networks | $3-8 | MyLead, AffMy |
Important: this is 30-day data for a new site. Affiliate revenue grows non-linearly: the first 30 days are foundation-laying. The main traffic arrives in months 3-6 when articles get indexed and start ranking.
Key Takeaways From 30 Days
- Comparisons convert better than reviews. Build your content strategy around them.
- Multilingual pays off for DE and ES, questionable for ZH.
- AI services are the main traffic driver in 2026.
- Format rotation prevents audience fatigue.
- Git + Hugo + CI/CD is the minimum automation needed.
- Affiliate income is a marathon, not a sprint. 30 days is just the start. Real money comes in months 6-12.
What’s Next
The second cycle (days 31-60) should include:
- Deeper comparisons with specific use-case scenarios
- Video content: reviews and comparisons in YouTube format
- Email newsletter: weekly digest of top comparisons
- Promotion: outreach, guest posts, link building
- Content updates: AI services change monthly, content must stay current
For now — data accumulates, SEO ramps up, affiliate links get clicked. The site is working.