If your SaaS blog has zero traffic, the first twenty posts should not target the terms you most want to rank for. Only 1.74% of newly published pages reach Google's top 10 within a year, and 72.9% of the pages already sitting there are more than three years old — a head term is a fight against incumbency you cannot win yet. The twenty posts that work are an ordered sequence: problem-aware long-tail questions first, then comparison pages, then one pillar written last, arranged so each post gives the ones before it another internal link and another reason to rank.
That ordering is the whole idea. Most "what should I blog about" advice hands you a list of topic types and leaves you to publish them in any order. This post gives you the sequence, the reason each slot sits where it does, and the 2026 complications — AI Overviews, half the web being machine-written — that make older advice misleading.
Ahrefs studied one million randomly selected URLs against 1.3 million keywords and found that 1.74% of newly published pages rank in the top 10 within a year. Filter to non-empty English content and it improves to 6.11% — better, but still nineteen in twenty pages never making page one. Zoom out and it's starker: across roughly 14 billion pages in Ahrefs' Content Explorer index, 96.55% get no Google traffic at all.
Read that as a per-post lottery and you'll quit in month three. Read it as a design constraint and it's useful: if one post is a coin flip weighted heavily against you, the unit of planning cannot be one post. It has to be a set, arranged so the failures are cheap and the winners lift each other.
Incumbency explains the bad odds. In the same study, only 13.7% of top-10 pages were under a year old, and the average #1 result was five years old, up from two years in a 2017 comparison. You aren't being outranked by better writing. You're being outranked by pages that have had half a decade to accumulate links and Google's confidence.
One figure there should shape your review habits, though: of the pages that did reach the top 10, 40.82% got there within a month. Breakthroughs, when they come, come fast — which is why Ahrefs advises updating a page that hasn't reached the top 10 after roughly six months instead of waiting on it. That's a decision rule rather than an anxiety loop.
So no, twenty isn't a magic threshold. No credible study shows traffic switching on at post 15, or 24, or 51; those numbers trace back to benchmark blogging from a decade ago. Twenty is a planning unit — two quarters at a weekly cadence, and enough surface area for internal linking to start doing real work.
When your domain is new, Google isn't weighing your content against the competition on equal footing. It barely has anything to weigh. John Mueller put it plainly: "With completely new websites, one of the difficulties that we have is we might not have a lot of signals for those websites so we have to make estimates." Estimates get revised as data arrives. Until then, you're being guessed at.
This is where founders open a keyword tool and misread it. Keyword Difficulty isn't a verdict on your chances; Ahrefs derives KD purely from referring domains pointing at the current top-10 pages. KD 0–5 means those pages barely have backlinks; around 50 means a couple of hundred; 90-plus means thousands, on a logarithmic scale. Their framing: the baseline is universal, but "your personal difficulty depends on the car you're driving." A brand-new SaaS site is driving a hatchback.
So stop reading KD as a score and start reading the SERP for cracks: outdated posts, low-authority domains on page one, forum threads holding real estate, thin content answering half the question. Those are the queries a good page from an unknown domain can take. Ahrefs also notes uncompetitive topics, roughly KD under 20, can rank without links at all — exactly where your first ten slots belong.
One obstacle sits before ranking, too: getting indexed. An analysis of 18 sites and 1.7 million pages via Google's URL Inspection API found quality issues accounted for 88% of monitored indexing problems, with "URL is unknown to Google" and "Discovered – currently not indexed" making up 67% of over 500,000 unindexed pages. A thin post on a new domain doesn't get outranked; it never enters the race. That alone argues for twenty substantial posts over sixty shallow ones.
Now the objection you should be raising. The classic advice is "start with low-volume long-tail questions" — and that is precisely where AI Overviews have taken up residence.
Semrush tracked more than 10 million keywords through 2025 and found nearly 60% of AI-Overview-triggering keywords have 100 or fewer monthly searches, skewing longer and more detailed. The click cost is real too: across 300,000 keywords of aggregated Search Console data, Ahrefs measured that an AI Overview cut position-1 CTR by 58% in December 2025, up from a 34.5% drop in April 2025. Their line: "For every 100 clicks you could historically earn for a top-ranking page, Google now 'keeps' 58."
Stop there and a new blog looks pointless. Three findings in the same research say otherwise. First, the damage concentrates at the top: −50.8% at position 2, −46.4% at position 3, and only −19.4% at position 10. You weren't opening at position 1 anyway, and a new post clawing to position 6 loses far less than the incumbent above it — the feature compresses the advantage of pages you can't beat. Second, prevalence hasn't risen in a straight line: 6.49% of queries in January 2025, 24.61% in July, 15.69% in November — and tracking the same keywords before and after an AIO appeared, Semrush saw zero-click rates dip from 33.75% to 31.53%. Third, the visitors who do arrive are worth more: across 500-plus marketing and SEO topics, the average AI-search visitor was 4.4 times as valuable as the average organic visitor by conversion rate.
What this changes about your twenty slots is concrete. Ahrefs examined which of their own pages held up against AI Overviews and found three resilient categories: free tools and utilities ("Summaries weren't enough for users … they required a tool to do it"), keywords that newly broke into page one, and SERPs where the AIO sat below position 1 — about 8.5% of cases, showing click parity or growth. So build one real utility into the twenty, and don't treat commercial pages as a bunker: a follow-up study of 600,000-plus US desktop keywords found the share of commercial SERPs carrying an AI Overview grew 71% between November 2025 and April 2026.
There's also a structural case for a cluster over one mega-post, straight from Google: "Both AI Overviews and AI Mode may use a 'query fan-out' technique — issuing multiple related searches across subtopics and data sources". If one prompt becomes many sub-searches, tightly linked pages that each answer a distinct sub-question offer more retrievable surfaces than one page trying to be everything. The same documentation is blunt about the rest: "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary."
You have no Search Console history, because Search Console only reports impressions your site actually earned — the Performance report defaults to three months, and on a new site those rows are empty. So the topics come from elsewhere.
Harvest questions before touching a keyword tool. Ahrefs documents four ways to mine Reddit with no data of your own: run site:reddit.com [your keyword] to see which threads already rank; push a subreddit URL through an organic-keywords report to see everything those discussions rank for; browse a subreddit sorted by "Top," treating upvotes as demand validation; and monitor brand mentions. Timing bonus: since Google's March 2026 core update, Reddit is a slightly weaker SERP competitor — Amsive's analysis of 2,000-plus domains found Reddit down 64.2 SISTRIX visibility points and YouTube down 567, the pattern being that "platforms that aggregate, list, or comment on other people's content lost visibility, while sites that created or owned the content gained visibility."
Your other sources cost nothing: every question in your support inbox, every question on a demo call, every place your docs were thin enough that someone had to email you. Those become the strongest slots in the map, because they arrive with first-party detail nobody else has.
Then sort the harvest into two piles. This is the most useful distinction in the exercise, and Ahrefs names it: supporting long-tail versus topical long-tail. A supporting phrase is a low-volume variant returning the same pages as the head term — a section inside an existing post, not a new URL. A topical long-tail keyword is one where the long phrasing genuinely is how people ask; that earns its own page. Get it backwards and you'll publish twenty posts chasing one SERP.
The tail is enormous. Ahrefs' US keyword database, refreshed in May 2026 from 110 billion discovered keywords, shows 93% of keywords get fewer than 10 searches a month — 2.3 billion of them — while fewer than 18,000 exceed 100,000. No single one opens floodgates, as Tim Soulo puts it, but "as you address more and more of them, the search traffic will eventually compound."
Don't romanticise zero-volume, though. Testing this on their own rankings, Ahrefs found keywords with estimated volume of ten or fewer delivered an average of 11.3 impressions a month, with fewer than 1% driving more than 100. Two exceptions justify a slot: a trending topic you believe has legs, and one lucrative enough that "even a few visits … can be worth $$$$."
Score every survivor on two axes. Winnability is the first — cracks in the SERP, per above. The second is Business Potential, Tim Soulo's 0–3 scale, which matters more than volume for a pre-revenue blog: 3 means "your product is an IRREPLACEABLE solution to a given problem," down to 0, where "there's no way to shoehorn your product, while talking about this problem." Spend early slots on 2s and 3s only. A 500-visit post at BP 0 is a hobby; a 40-visit post at BP 3 is pipeline.
And de-duplicate before you write, not after. Cannibalisation is narrower than folklore suggests — Ahrefs' test is that it's only real when pages target the same keywords and serve the same intent and performance suffers. We've covered what keyword cannibalization actually costs a small blog and why the cheap fix is a topic record checked before writing rather than an audit afterwards. Spreadsheet or topic ledger, the function is the same: slot 14 should never quietly become slot 3 again.
First, the mechanic that justifies the ordering. Zyppy analysed 23 million internal links across 1,800 sites and roughly 520,000 URLs, and found URLs with 0–4 internal links averaged 2 clicks from Google while URLs with 40–44 averaged around 8 — four times the traffic, with the correlation reversing past roughly 45–50 links. Pages with at least one exact-match anchor pointing at them had "at least five times more traffic" than pages without. Cyrus Shepard calls this "directionally useful but not necessarily scientific truth," which is the right posture.
Here's why that turns a list into a sequence. On a twenty-post blog, you are the only source of internal links you control. Post 12 can send descriptive links to posts 3, 5 and 9; post 3 can never link forward to post 12 unless you go back and edit it — which is what the sequencing rules below require. Wired both ways, twenty posts don't accumulate linearly. Each new one raises the floor under everything already published.
| Slots | Type | Why here |
|---|---|---|
| 1–3 | Cracked-SERP problem-aware questions | Winnable without links; creates something to link to |
| 4–6 | Questions from your inbox, calls and Reddit threads | Highest first-party detail; hardest to replicate |
| 7 | "[Incumbent] alternatives" | Demand attaches to their brand, not yours |
| 8–10 | Three more spokes | First posts that can link back to six or seven live pages |
| 11 | A free tool, calculator or template | The one format measured as AI-Overview-resilient |
| 12–13 | "Best X for [narrow segment]" | Commercial intent; citation-friendly format |
| 14–15 | Head-to-head "you vs named competitor" | Needs a live product and real differentiators |
| 16 | One opinionated, distribution-first post | Aimed at communities, not a keyword |
| 17 | First-party data or teardown | Originality is what March 2026 rewarded |
| 18 | The pillar | Written last, once there are spokes to link |
| 19–20 | Reserved — filled from Search Console | By now you have real impression data |
Slots 1–3 — three cracked-SERP questions. Pick them by inspecting SERPs, not volume: queries where page one holds a forum thread, a 2019 post, or a domain nobody's heard of. Cheapest wins available to a site with no links, and their real job is structural — they're the first pages later posts will link to.
Slots 4–6 — the questions you already get asked. Every support ticket that took more than two sentences to answer is a candidate. These carry detail nobody can copy from a competitor, which is the property that gained ground in March 2026: created and owned, not aggregated.
Slot 7 — one "[incumbent category tool] alternatives" page. Your first commercial page, and it works pre-launch precisely because demand attaches to somebody else's brand. Nobody searches "[your product] vs" yet; plenty search for alternatives to the tool you're replacing.
Slots 8–10 — three more spokes, each linking back. Here the compounding starts paying rent. Each ships with at least two descriptive links to slots 1–7, and each triggers an edit to an older post so a link comes back the other way. That's three posts and roughly six new internal links.
Slot 11 — a free tool, calculator, template or generator on its own page. The slot most likely to survive AI Overviews, for the reason Ahrefs measured: a summary can't do the job. A meta-description length checker, a pricing calculator, a template with a real worked example. It needn't be sophisticated — just something a paragraph of text cannot replace.
Slots 12–13 — two "best X for [narrow segment]" listicles naming competitors honestly. Narrow does the work: "best time-tracking tools for freelance video editors" is winnable; "best time-tracking tools" is not. Say plainly when a competitor fits better — it's also how you learn what an honest roundup looks like before asking to be in someone else's.
Slots 14–15 — two head-to-head comparison pages, once you have a live product with real differences to describe. Model them on one that works: Plausible's comparison against Google Analytics 4 opens with a framing claim, puts a side-by-side table high on the page, runs five detailed critiques, then — the part most founders skip — concedes what GA4 does better, before objection-handling Q&A, testimonials and repeated CTAs. That concession isn't weakness; it's what makes the other sections believable.
Worried about naming competitors? One data point: a practitioner who built just under 50 standalone comparison pages for a B2B SaaS — producing roughly 1,000–2,000 organic visits a month at 5–10% monthly conversion — reports one cease and desist and one request to correct positioning across all fifty. Small, non-zero, manageable if your tables are accurate and your tone stays generous.
Slot 16 — one opinionated post written for humans, not a keyword. More on this in the case study below; it's the slot most likely to change your trajectory and the one most likely to get skipped.
Slot 17 — one first-party data post or teardown. Your own numbers, your own experiment, your own migration story. This is the slot that gets linked to and cited, and the one thing no volume-based competitor can generate.
Slot 18 — the pillar, written last. Practitioner consensus puts a pillar at roughly 10–15 spokes with two-way linking — which you now have. Published first, a pillar is an orphan with a table of contents; published eighteenth, it's a hub with seventeen pages already pointing at it. No study proves pillar-last beats pillar-first, so treat it as a judgement call: internal links are the mechanism, and you can't wire a hub to spokes that don't exist.
Slots 19–20 — deliberately left empty. By now seventeen posts have been live for months and Search Console is showing impressions for queries you never targeted. These two get filled from data rather than guesses.
One thing that is not a rule: a word count. Google's starter guide states there's "no magical word count target, minimum or maximum," and that length alone doesn't matter for ranking. A narrow how-do-I slot might run 900 words; the pillar might run 4,000.
Plausible Analytics is the closest thing to a documented run of this play by a bootstrapped, developer-facing SaaS. Between March 2020 and January 2021, Plausible went from 143 monthly visitors from Google search to 7,167, while MRR went from around $400 in April 2020 to more than $10,000 at the start of 2021 and subscribers from fewer than 100 to over 2,000. Behind it: 32 blog posts, 15 product comparison and differentiation pages, and full docs, at roughly one post a week from April 2020.
Look at the shape. Comparison pages were about a third of everything published — far heavier than "write helpful articles" advice implies — and the cadence held weekly for ten straight months rather than arriving in bursts. Marko Saric describes topic selection as "a mix of focusing on product features … and popular issues our product has a solution for." The best performers addressed GDPR, cookieless tracking and Safari's privacy changes: problem-aware queries adjacent to the product, not head terms like "web analytics."
Then there's slot 16. On 17 April 2020, Plausible published a roughly 2,500-word post arguing people should stop using Google Analytics, and seeded it on Hacker News, Reddit, Lobsters, Indie Hackers, Twitter and Mastodon. In week one it drew more than 48,000 unique visitors — against 27,300 in the previous 15 months combined — plus 166 new trial signups that week versus about 40 across the four prior months. Saric's conclusion is worth taping to a monitor: "The promotional work only starts after the blog post has been published." One deliberate swing among twenty planned pages, not a strategy by itself, and not repeatable on demand.
His caveat is equally honest: "Most of our traffic to date comes from people searching for our brand name and relevant phrases directly." Ten months and 47 pages in, competitive head terms were still uncertain. That's what success looks like at this stage — not ranking for the category term, but a brand with enough gravity that people search for it while the long tail sends qualified strangers.
For the first few months you'll have no clicks, and that isn't a signal. Impressions are. Search Console separates impressions from clicks, defaults to a three-month window, and marks the newest rows preliminary. The first real sign of life is impressions on queries you never targeted — Google testing your pages against phrasings you didn't anticipate. That tells you your pages are indexed and that your niche is broader than your keyword list, which is exactly where slots 19 and 20 come from. Google also confirms AI-feature traffic appears in the Performance report under the "Web" search type.
When clicks do arrive, calibrate. First Page Sage's benchmarks across 100-plus companies put B2B SaaS visitor-to-lead conversion from SEO at 1.9%, ahead of webinars at 1.2% and email at 1.1%. That's the arithmetic behind Business Potential scoring: at roughly 2%, a hundred visitors on a BP-3 topic is two leads, while a thousand visitors on a BP-0 topic is a nice chart.
Three objections are worth answering before someone raises them for you.
"Topical authority isn't real, so a topic map is cargo cult." Partly fair. Asked about it on Reddit, John Mueller endorsed a comment calling topical authority a rebrand of "relevancy" and added, "Don't worry about it." But the mechanisms are documented independently of the label: internal links correlate with clicks, Google recommends internal linking for discovery, and query fan-out rewards covering distinct sub-questions. Call it compounding internal structure and the plan survives the semantics.
"Half the web is AI-written now — twenty more posts is noise." Graphite analysed 55,400 Common Crawl URLs with three detectors and found Q1 2026 at 49.9% AI-generated versus 50.1% human, plateaued near 50% since Q1 2025. Their observation is the important half: "these articles largely do not appear in Google and ChatGPT." Volume alone is worth roughly nothing now. What isn't noise is a support-inbox answer nobody else can write and a number from your own product.
"My blog might not even be the main lever." This one has teeth. Ahrefs studied 75,000 brands and found branded web mentions correlated most strongly with AI Overview brand visibility at 0.664, ahead of branded anchors at 0.527 and backlinks at 0.218. Correlations were moderate-to-weak, but the direction matches what one bootstrapped SaaS reported: picdrop logged 12,539 LLM referral sessions in 2025 against 1,473 in 2024 and was 6.5 times more likely to be cited via third-party sources than via its own domain. So run directory listings, review profiles and roundup outreach in parallel with slots 1–20.
One last question, especially if an agent drafts any of this: where's the line? Google's spam policies draw it directly. "Scaled content abuse is when many pages are generated for the primary purpose of manipulating search rankings and not helping users," explicitly including generative AI used "to generate many pages without adding value for users." The operative phrase is "no matter how it's created" — the offence is unoriginal, low-value volume, not the tool. Our longer read on what Google's scaled content abuse policy really bans covers the enforcement mechanics; the short version is that twenty sequenced, verified posts carrying first-party detail sit comfortably on the right side of it.
Which is why the constraints in this map double as safety rails. Business Potential scoring caps how many topics qualify. The cracked-SERP requirement forces you to inspect each SERP by hand. The inbox slots guarantee original material. The six-month review means you maintain rather than accumulate. And one post a week — the pace Plausible held for ten months — is the shape of publishing that compounds instead of spiking; for the benchmarks behind that cadence, see how often a small blog should publish.
That cadence is what Magic Share is built for: the agent researches a topic, drafts it, confirms every cited URL resolves before the draft is saved, attaches a verification report, then waits for you to approve it. The topic ledger remembers all twenty slots, so post 14 never repeats post 3. Nothing publishes on its own — which matters most in exactly the situation described here, where slots 19 and 20 depend on a judgement only you can make from your own data.
How many blog posts before a new SaaS site gets traffic? There's no verified threshold. The widely repeated "15–20 posts" and "24–51 posts" figures trace back to decade-old benchmark blogging. What the data supports: 1.74% of new pages reach the top 10 within a year, and 40.82% of pages that eventually get there do so within a month. Treat twenty as a batch you evaluate after two quarters at a weekly cadence, not a finish line.
Should I write comparison pages before my product launches? Write "[incumbent] alternatives" and "best X for [narrow segment]" pages, not "[your brand] vs [competitor]" pages. Pre-launch, nobody searches your brand — but plenty search for alternatives to the tool you're replacing. Save head-to-head pages for slots 14–15, and include an explicit section on when the reader should choose the other tool, as Plausible's GA4 comparison page does.
Do AI Overviews make long-tail blogging pointless? No, but they change the maths. Nearly 60% of AIO keywords have 100 or fewer monthly searches, and an AIO cuts position-1 CTR by 58%. Losses shrink down the page (−19.4% at position 10), AIO prevalence fell from 24.61% in July 2025 to 15.69% in November, and AI-referred visitors converted 4.4 times better in Semrush's measurement. Adjust by including a free tool and first-party data among your twenty.
Should I publish the pillar post first or last? No study settles it. The argument for last is mechanical: a pillar's value comes from two-way internal links, and it can't link to spokes that don't exist. Publishing it eighteenth means it launches with seventeen pages already pointing at it, in a niche where you now know what readers actually ask.
How do I find keywords with no Search Console data at all?
Harvest questions first, validate second. Use site:reddit.com [keyword] to see which threads rank, run subreddit URLs through an organic-keywords report, and read your support inbox and demo-call notes. Then run each candidate through a volume and difficulty check, split supporting long-tail from topical long-tail, and score what's left on Business Potential 0–3.
A blog with no authority doesn't lose because its posts are bad. It loses because each post arrives alone, into a SERP where the average #1 result has had five years to settle in. Twenty posts published in a deliberate order fix the "alone" part: cracked-SERP questions first to build something worth linking to, inbox-sourced posts for detail nobody can copy, one alternatives page, a free tool, honest comparisons, one swing aimed at people rather than a keyword, one post made of your own data, then the pillar — and two slots kept empty for whatever Search Console tells you that you didn't know.
The hard part isn't the map. It's holding a weekly cadence for two quarters while the clicks sit at zero and the impressions creep up. If that's the part you'd rather stop doing by hand, Magic Share drafts and fact-checks posts on your schedule and holds every one as a draft until you approve it — your first three posts are free, no card required. Bring the twenty slots; the agent will bring the drafts.
Magic Share researches, writes and fact-checks posts like this for any site — point it at your URL and review your first draft today.
Plant your first post