Portfolio
Ten sites, ten bets
Nobody knows in advance which niche wins. The answer is not to guess better: it is to take ten bets and let the data decide.
Short answer
Why ten sites instead of one?
Because nobody knows in advance which niche wins. Ten independent bets turn a single wager into a portfolio: seven can return nothing if three work. The real cost is not hosting, it is review time, and review time is what caps how many sites you can honestly run.
Why ten sites and not one
One site is one bet on one niche. And the truth nobody enjoys saying: even applying the selection method perfectly, you do not know in advance which of your subjects will take off. Competition moves, the algorithm moves, demand moves.
So stop trying to guess. Take ten bets:
If seven sites out of ten return nothing and three work, you have won — because the seven failures cost a few tens of euros each, and the three that work have no ceiling. That is exactly an investor's reasoning: the loss is capped, the gain is not.
It is also what makes this business possible for a beginner: you do not need to be right. You need to be right one time in five, having paid very little for the other four.
The odds
Suppose each site has a probability p of “working” (reaching the “good” scenario on the money page). Here is what ten independent bets give. This is arithmetic — redo it yourself:
| If each site has… | Chance NONE works | Chance at least 1 works | Expected winners |
|---|---|---|---|
| 10% chance | 35% | 65% | 1 site |
| 20% chance | 11% | 89% | 2 sites |
| 30% chance | 3% | 97% | 3 sites |
| 40% chance | 0.6% | 99.4% | 4 sites |
With a single site at 20%, you have a four-in-five chance of ending the year with nothing. With ten sites at 20%, you have a nine-in-ten chance of at least one winner, and you expect two. Same work per article; spreading the risk changes everything.
It assumes your ten sites are independent bets — ten different niches, not ten variants of one — and that you treat them with equal care. Ten rushed sites do not have ten times 20%: they have ten times almost zero. The p does not come from the number of sites, it comes from niche quality and real review. None of these probabilities is measured by us: they are assumptions you set yourself.
What a portfolio returns, in practice
A worked 12-month example, using the scenarios from the money page: 7 failures, 2 good, 1 excellent.
| What you hold | Count | Monthly revenue each at month 12 | Subtotal |
|---|---|---|---|
| Sites that never take off | 7 | ~$0–20 | $0–140 |
| “Good” sites (decent niche, digital affiliate) | 2 | ~$300–900 | $600–1,800 |
| “Excellent” site | 1 | ~$1,800–5,000 | $1,800–5,000 |
| Portfolio total | 10 | ~$2,400–6,900/month |
And if the bet goes badly? Ten failures: you are out roughly $1,300–2,200 for the year plus a lot of hours — and you have learned ten niches. That is a bad result, not a ruin. That is what spreading the risk means.
It is an arithmetic illustration of a 7/2/1 split, with per-site revenues that are themselves calculated, not measured. The real split may be 10/0/0. The table does not say “you will earn this”: it says “here is why you take ten bets instead of one”.
What each site needs
A deliberately simple stack. Every extra line is a line you maintain ten times:
- A domain name. An ordinary .com is fine.
- Shared hosting. One good plan hosts all ten.
- WordPress + a free theme. Astra, Kadence, GeneratePress or the default. Do not buy a premium theme: on a generated site the theme is not the limiting factor.
- AutoBlogging Pro. Free.
- A free SEO plugin. Rank Math or Yoast, free tier.
- One AI API key. Shared across all ten sites.
Premium theme, page builder, paid cache plugin, subscription keyword tool, backlink service. None of them solves the real problem of the first months, which is finding out which niches work.
Setup cost
Assumption: 10 sites, 30 articles each in month one, so 300 articles.
| Item | Detail | Lean | Comfortable |
|---|---|---|---|
| Domains | 10 × .com for a year | ~$120 | ~$140 |
| Hosting | Shared multi-site, 12 months prepaid | ~$65 | ~$270 |
| Theme + plugins | Free | $0 | $0 |
| Content — DeepSeek | 300 articles × ~$0.007 | ~$2 | — |
| Content — Claude Sonnet | 300 articles × ~$0.09 | — | ~$26 |
| Images | 300 × ~$0.03 (Stability Core) | $0 (Pexels, free) | ~$10 |
| Topic clusters | 10 generations | ~$0.11 | ~$1 |
| Setup total | ~$187 | ~$447 |
Content — the part everyone assumes is expensive — is about 1% of the lean budget. Domains and hosting dominate, and neither depends on AI at all. The price of the bets is not what will limit you.
Running cost
At cruising speed: 30 articles per site per month, so 300 a month.
| Item | Per month | Per year |
|---|---|---|
| Domains (amortised) | ~$10 | ~$120 |
| Hosting | $5–23 | $65–270 |
| Content (DeepSeek, 300 articles) | ~$2 | ~$24 |
| Content (Claude Sonnet, 300 articles) | ~$26 | ~$312 |
| Images (300, Pexels) | $0 | $0 |
| Images (300, generated) | ~$10 | ~$117 |
| Total | ~$17–69 | ~$209–830 |
Ten sites, three thousand articles a year, for the price of a phone plan. Money is really not the issue. The issue is directly below.
The real bottleneck: your time
300 articles a month at 30 minutes of serious review each is 150 hours a month. That is a full-time job. Nobody does that, and a plan that assumes otherwise is a plan that fails in month two.
There are exactly four honest ways to solve this. They combine:
| The fix | What it gives | What it costs |
|---|---|---|
| Stagger the launches — 2 sites a month for 5 months | You never review 300 articles at once: you ramp to 60, then 120, while the earliest sites already cruise. | 5 months before all ten exist |
| Calibrated review — 10–15 min per article outside sensitive topics | A fixed checklist instead of a floating read. 300 articles = 50–75 h. | Not allowed on YMYL topics |
| Cull the losers at 90 days | You stop feeding 6 or 7 sites: the load drops to 90–120 articles a month, all on the winners. | You must accept abandoning things |
| Pay a reviewer | The only lever that buys time. A reviewer at $3/article over 300 articles: ~$900/month. | Do it when the portfolio pays, not before |
The combination that works for one person: stagger the launches, calibrate the review, cull hard at 90 days. You land on a sustainable 25–40 hours a month — a light part-time, not a full-time job.
Holding 30 articles per site per month
Thirty a month is one a day. Why that number and not five:
- A site publishing 5 a month does not exist to a search engine. No depth, no subject, no reason to treat it as a reference for anything.
- You need data quickly. 90 articles after 3 months give a usable signal; 15 articles give nothing at all, and you cannot decide.
- The long tail is a numbers game. Out of 30 articles, 3 or 4 will find a question nobody covered. Out of 5, often zero.
The daily rhythm, in practice:
Sunday — prepare
The plugin drafts the week: 7 drafts per active site. You publish nothing yet. Cost: a few cents.
Weekdays — review in batches
One session a day, articles back to back with the same checklist. You read fast because you always read in the same order: the order is here.
Publish on a spread
One article a day per site, not thirty at once. A burst publish on a new domain is the crudest signal there is.
Friday — look
Search Console, ten minutes per site. What is getting impressions? That is the week's only question.
The word that matters in “30 quality articles” is quality. Thirty unreviewed articles a month is the fastest, cheapest way to get ten sites demoted at once — see the risks. If you cannot review 30, publish 15 reviewed ones.
The 90-day cull: the decision that makes the return
This is the most important moment in the portfolio, and almost nobody does it.
| What Search Console says on day 90 | What it means | What you do |
|---|---|---|
| Impressions rising, a few clicks | The niche is responding. Probable winner. | Double down: take this site to 40–60 articles/month |
| Impressions flat but indexing fine | The niche is too competitive for a new site. | Hold at 10 articles/month, decide again on day 180 |
| Almost nothing indexed | Technical problem, or quality judged insufficient. | Fix the technical side; if it is already clean, stop |
| Nothing at all, on a genuinely weak niche | The bet was wrong. | Stop feeding it. Keep the domain, it costs $12/year |
You delete nothing: a stopped site still exists and can recover on its own. You reallocate your hours from losers to winners. That reallocation, not the number of sites, is what separates a portfolio from ten abandoned blogs.
The 12-month calendar
| Period | Active sites | Articles / month | Hours / month | What you do |
|---|---|---|---|---|
| Months 1–2 | 2 | 60 | ~15 h | Launch the first two. Learn your real pace. |
| Months 3–4 | 4 | 120 | ~30 h | Two more. The first two now have 90 days of data. |
| Months 5–6 | 6–8 | 150–200 | ~35 h | First cull: stop the dead, double the living. |
| Months 7–9 | 10 | 150–250 | ~40 h | Full portfolio. The cull freed the hours. |
| Months 10–12 | 3–5 (active) | 150–200 | ~35 h | Almost everything goes to the winners. The rest sleep. |
Look at the last row: by the end of the year a healthy portfolio is no longer feeding ten sites. It feeds three or four — the ones that proved something. The ten existed to find those three.
The 12-month total
| Scenario | Spend | Hours | Real cost at $27/h |
|---|---|---|---|
| 10 sites, DeepSeek, Pexels images | ~$325 | ~400 h | ~$11,125 |
| 10 sites, Claude, generated images | ~$830 | ~400 h | ~$11,630 |
| 10 sites, paid reviewer from month 6 | ~$5,400 | ~150 h | ~$9,450 |
The money stays trivial; time is the real capital invested. So the question is not “do I have $300?” but “do I have 30 hours a month, every month, for a year?” If the answer is no, launch 4 sites instead of 10: the portfolio reasoning already holds at four or five bets.
Choose 10 niches → SEO without learning it Install the plugin
Prices quoted are order-of-magnitude figures in US dollars, taken at this page's update date, for entry-level consumer offers. The probabilities and revenues on this page are calculation assumptions, not measurements. Verify pricing and rates before building a budget on them.
Updated 2026-09-10