How do you market a SaaS product when anyone can build one?

By Andrej Ruckij · October 30, 2026

TL;DR: Building got cheap, so distribution decides which SaaS survives, and organic traffic is the channel a founder with no budget can own. But organic is slow. Only 1.74% of newly published pages reach Google’s top 10 within a year, per Ahrefs, and in Supabase’s survey of 2,000+ founders, 56% got their first paying customers through personal networks. So I’d split the first 90 days. Get the first 10 customers by talking to people you can reach directly and posting in one community where your buyers already are. At the same time, build the assets that compound: “[competitor] alternative” pages, one free tool, and a listing in the marketplace your buyers already use. Treat launch days on Product Hunt and Hacker News as a bonus. In the last 30 days only about 4% of Product Hunt launches were featured. An agent-run SEO stack (OpenSEO plus Claude) costs about $30 a month, against €112.52 a month for Semrush’s SEO plan (billed annually) or €119 for Ahrefs Lite as listed. Don’t use it to mass-produce pages: Google’s spam policy names that as scaled content abuse.

Why distribution is the whole game now

My view is simple: anyone can build a piece of software or a SaaS now. What decides whether it survives is distribution, and for a founder without a budget that mostly means organic traffic.

The numbers point the same way. About 560,000 new apps reached the App Store in the first half of 2026, close to the roughly 600,000 in all of 2025, while downloads grew 2% to 17.6 billion (Sensor Tower data, reported by 9to5Mac). Those are consumer apps, but I’d expect SaaS to follow the same pattern. In Supabase’s State of Startups 2026 survey, 62% of startups say more than half their code is AI-written, and 67% of founders have never tried paid acquisition. When the number of products grows that much faster than the number of buyers, better distribution mostly decides who gets the buyers who already exist.

Organic search also got harder. Ahrefs found that the presence of an AI Overview correlates with a 58% lower average click-through rate for the top-ranking page (300,000 keywords: 150,000 with an AI Overview and 150,000 informational keywords without one; December 2025 data compared with December 2023). Pew found people clicked a regular result in 8% of visits when an AI summary appeared and 15% without one, and clicked a link inside the summary in just 1%. More on that: seo/zero-click-strategy.

One detail in Pew’s data matters for tool makers. AI summaries appeared on only 8% of one- or two-word searches, against 53% of searches with ten or more words and 60% of question searches. Short “I need a tool” searches like “invoice parser” are much less exposed than “how do I” questions. That’s my reading of Pew’s numbers, not a claim Pew makes.

What got real small SaaS products their first customers

I looked for founders who published numbers. Most of what ranks for this topic is agency marketing with no baseline, so I’ve kept only cases where someone named the product and the figures, and I’ve labelled them. All of these are self-reported.

”Alternative to [incumbent]” positioning and comparison pages

Plausible Analytics (2020-21). Co-founder Marko Saric wrote that Plausible went from $400 MRR in April 2020 to $10,000 MRR by January 2021, with 1,778 paying customers, on a marketing budget of “$0 if you don’t account for our own time”. They rewrote the homepage to lead with the comparison (“Simple and privacy-friendly alternative to Google Analytics”) because “People are impatient, so if you don’t do that, chances are you lose that visitor.” Google organic search became their top source of trial signups. This was before AI Overviews, so it proves the positioning works; today’s search traffic is a separate question.

OpenSEO (2026). Its GitHub description is the query itself: “Open source alternative to Semrush and Ahrefs”. It’s the only entry on OpenAlternative’s page of open-source Semrush alternatives. It sat on GitHub’s trending lists from late June to the end of August 2026, per Trendshift, and gained about 9,850 stars in the 61 days to November 1, reaching about 26,000. Stars aren’t customers, and I couldn’t verify where its first users came from.

What the conversion data says. Grow and Convert, an agency, analysed 95 of its clients’ articles (123,000+ organic pageviews, 4,687 conversions). Comparison and alternatives keywords had the highest average conversion rate, 8.43%, against 4.85% for category keywords. Their own caveat matters more: “the majority of the 23 competitor keywords we analyzed actually converted at less than 4%.” A few pages carry the average.

Free tools

SiteGPT (2023-26). Founder Bhanu Teja told Starter Story (episode published January 8, 2026) that SiteGPT is at about $13,000 MRR. He built “almost 50 free tools”, gets around 50,000 clicks a month from Google, and says almost 90% of those Google clicks come from the free tools. He mentions around 300 leads (a separate write-up of the same case says 200, and neither number is clearly per month) and no paid marketing. He also says SiteGPT reached about $10,000 MRR in its first month, so the first customers likely came from somewhere else; the tools explain the steady inflow since. His claim that a new tool takes him “less than five minutes” is about his own setup; budget a day for your first one.

Product Hunt

IndieMerger (about a year ago). Alexander Belogubov placed 3rd of the day: 1,305 visitors, 210 signups, 7 purchases, $543. He prepared for one day and had about 10,000 followers on X to push the launch. His verdict: “Product Hunt brought decent traffic but not the massive sales many founders expect,” and traffic dropped sharply the next day.

MakerPulse (2026). A metrics dashboard with a $19 Pro plan. After four weeks of building: 3 upvotes in 24 hours, one of them the author’s own, single-digit signups, $0. The author’s read: “The makers who win Product Hunt today have spent months building an audience before launch day.”

The base rate. Hunted.space, which tracks every launch, shows 860 featured launches out of 21,991 in the last 30 days, about 4%. The all-time feature rate is 22.6%. An analysis of 3,869 launches from the first half of 2026 found Sunday launches reached the day’s top three 39.7% of the time against 8.6% on Tuesday, mostly because fewer products launch on Sunday. The author calls these “correlations, not levers.”

Hacker News

ClearNoteLab (December 2025). Jack Brauner built an AI notes-to-PDF tool in 9 days for $200 with Bolt, posted it to Hacker News three days before Christmas, and got 1 upvote (his own), 0 comments and 0 signups. He had no followers on X or LinkedIn. His own conclusion: “Building is easy. Distribution is hard.”

Reddit and communities

A founder who built a Reddit tool, Wappkit Reddit, wrote up how four of his accounts got suspended “for breaking Reddit’s rules against spam”. The causes he lists: links posted from accounts less than about three days old, and the same comment template pasted into about ten threads. What worked afterwards was weeks of plain commenting before any mention, naming the product without a link, and posting in self-promotion subreddits such as r/SideProject. He gives no signup numbers and writes about a tool he built, so take the method rather than the claims.

Reddit counts twice because AI assistants cite it. Semrush found Reddit and LinkedIn among the top five cited domains across ChatGPT, Google’s AI Mode and Perplexity. It’s also unstable: Reddit appeared in close to 60% of ChatGPT responses in early August 2025 and around 10% by mid-September.

Directories

I found no measured multi-directory test that I could read in full. Most “submit to 100 directories” content is written by people selling submission services. My read: a listing where your buyers actually look for alternatives (OpenAlternative for open-source tools, the category pages of the marketplace you sell into) is worth an hour; a mass-submission package mostly buys low-value links. Tag every listing with a UTM parameter so you can see which ones send anyone.

App marketplaces and integrations

Shopify App Store. For most developers, you keep 100% of your first $1,000,000 in lifetime app revenue (counted from January 1, 2025) and 85% above that, after a one-time $19 registration fee. Developers above $20 million in App Store revenue or $100 million company revenue pay 15% on all app revenue. Getting found there is still slow. A team that got a loyalty app approved on August 5, 2026 posted three weeks later that three merchants were using it. Facebook merchant groups gave mostly silence, plus offers of paid installs and paid reviews, which the team declined. A reply from a Shopify-labelled account suggested they “lean in to your past clients”. That points back to the same place as Supabase’s survey: your own network.

ChatGPT apps. By late December 2025, OpenAI had an app directory live inside ChatGPT. Apps are built with the Apps SDK, which runs on MCP, can be called by name, and ChatGPT can suggest them when it thinks they fit. I found no published install or signup numbers for third-party apps, so I’d treat it as a cheap experiment and keep my hours elsewhere. More on why integration matters: marketing/ai-interface-layer.

What each channel costs

Money figures are from the pages cited. Time figures are my rough estimates for a solo founder.

Channel	Money	Time (my estimate)	When it pays off	Named case
Direct outreach to your network	$0	1-2 hours a day for 2-3 weeks	Days to weeks	56% of founders’ first paying customers (Supabase)
One niche community	$0	30 minutes a day, weeks before you post	Weeks	Suspensions when rushed (Reddit case above)
Alternative and comparison pages	$0-30 a month for data	3-5 hours per page	Months	Plausible; Grow and Convert’s 8.43% average
One free tool	$0 plus hosting	1-3 days for the first one	Months	SiteGPT, about 50,000 Google clicks a month
Marketplace listing (Shopify example)	$19 one-time, 0% share on the first $1M for most developers	Days to build and pass review	Months; reviews drive ranking	3 merchants in 3 weeks (Shopify forum)
Product Hunt	$0	1 day of prep, plus the launch day	One day of traffic	IndieMerger $543; MakerPulse $0
Show HN	$0	A few hours	One day, if it catches	ClearNoteLab: 0 signups
Directories	$0 for the relevant few	An hour each	Unclear, measure it	No measured test I could read

What an agent-run SEO stack involves

This is the part that changed this year. Keyword research, competitor gaps and site audits used to need a suite subscription: €112.52 a month for Semrush’s SEO plan billed annually, or €119 a month for Ahrefs Lite as listed. Now an AI agent can pull the same kind of data through an API and do the analysis itself.

The parts:

OpenSEO (MIT licence, about 26,000 GitHub stars): keyword research, rank tracking, competitor insights, backlinks, site audits and AI visibility, with an MCP server that agents such as Claude Code can call (the README also covers ChatGPT, Codex and Cursor setups). Hosted plans are $10, $30 or $200 a month, each including at least that much usage; beyond it you pay as you go. Hosted actions cost what the data costs OpenSEO plus 28%. Per its pricing page: about $0.05 per keyword search, about $0.08 for a backlink domain overview, and a fraction of a cent per rank check.
Self-hosted alternative: run OpenSEO with your own DataForSEO key. DataForSEO has a $50 minimum top-up and no subscription; SERP data starts at $0.0006 per request and keyword data at $0.0012 per task. One critic of OpenSEO adds that backlink data needs a $100 monthly commitment; check that on DataForSEO before you plan around backlinks.
marketingskills by Corey Haines (about 57,800 stars, 53 skills, free): includes competitors for comparison and alternative pages, free-tools, programmatic-seo, seo-audit, ai-seo, directory-submissions and launch. Install with npx skills add coreyhaines31/marketingskills or as a Claude Code plugin. More on the format: glossary/skill and How do you create skills in Gemini (and reuse them in Claude)?
An agent to run it: Claude Pro at $20 a month (or $17 billed yearly) includes Claude Code.
Google Search Console: free, and the only source of what Google actually shows for your site.

What it costs in practice. OpenSEO’s own calculator puts 100 keyword searches at $5.40 and 20 backlink checks at $1.58. So OpenSEO Base plus Claude Pro is about $30 a month for a founder doing research weekly, compared with €112.52-119 for a suite.

What you’d have it do, in order:

Pull the keywords your three closest competitors rank for and list the ones with “alternative”, “vs” or short tool names in them.
Cluster them and pick the 5-10 that a real page of yours could answer better than what ranks now.
Draft each page from your own material: your pricing, your screenshots, real differences. You edit and publish.
Once a week, compare the agent’s numbers with Search Console. Where they disagree, trust Search Console.

Limits. The data is an index snapshot, not a live render. A critic of OpenSEO puts it as “It tells you what DataForSEO’s crawlers saw, when they saw it”, so a JavaScript-heavy site still needs a real crawl. Pay-as-you-go bills grow with use. And once every competitor has the same 53 skills, the skills stop being an edge; what’s left is your positioning, your own data and the judgment about which pages deserve to exist.

Where this goes wrong

Mass-produced pages. Google’s spam policy, last updated August 28, 2026, lists “Using generative AI tools or other similar tools to generate many pages without adding value for users” as scaled content abuse. Also listed: “Creating many pages where the content makes little or no sense to a reader but contains search keywords”. Programmatic pages are fine when each one carries data a reader can’t get elsewhere, such as your product’s real numbers per use case. A thousand template pages with a swapped keyword are what the policy describes. How to check: open five of your generated pages at random; if they read the same apart from the keyword, don’t publish them.

Treating a launch day as a channel. IndieMerger’s traffic fell the day after; MakerPulse and ClearNoteLab got nothing. ClearNoteLab’s founder had no followers, and MakerPulse’s author writes that the makers who win “have spent months building an audience before launch day.” My own rule of thumb: if you can’t name 50 people who’ll see your launch post on day one, you’re not ready to launch.

Counting signups instead of users. IndieMerger’s 210 signups became 7 purchases. Look at activation and payment per channel, not signups.

Burning a community. Links from new accounts and repeated comment templates got four accounts suspended in the Reddit case. Read each subreddit’s rules first and comment for a few weeks before you mention anything.

Judging AI referrals by visits. At Ahrefs, AI search sent 0.5% of visits in the last 30 days but 12.1% of signups. That’s one site, and Ahrefs says so, but it means you should track chatgpt.com, perplexity.ai and gemini.google.com as a separate channel, judged on signups.

The counter-evidence

Some of what I found cuts against my own take, and it changed my advice.

First customers rarely come from organic. 56% of founders in Supabase’s survey got their first paying customers through personal networks. My read: organic brings the next hundred customers and rarely the first ten.
New pages rank slowly. Only 1.74% of newly published pages reached Google’s top 10 within a year in Ahrefs’ study, and the average #1 page is five years old. On the other side, 40.82% of pages that did reach the top 10 got there within a month, which is mostly low-competition queries. That’s why the plan below starts with alternative and short tool queries.
AI-written content isn’t penalised as such. Ahrefs analysed 600,000 ranking pages and found 86.5% contain some AI content, with a correlation of 0.011 between AI use and position. What gets sites in trouble is the policy above: scale without value.
Google disputes the click collapse. Google’s Liz Reid wrote that “total organic click volume from Google Search to websites has been relatively stable year-over-year”, and that clicks are moving toward forums, videos and first-hand perspectives. Google has an interest here, but it supports publishing your own numbers rather than generic guides.
“Anyone can build” is overstated. 78% of Supabase’s respondents are technical. Building got cheaper; it didn’t become free of skill.

Who this is for

For: a founder with a working SaaS, few or no paying customers and little money for ads, selling self-serve or low-touch to small businesses, developers or other founders. Not for: enterprise products sold through long sales cycles, where a handful of relationships matter more than search, and products nobody has yet said they’d pay for. Distribution won’t fix that; talk to buyers first.

How to do it: the first 90 days

Days 1-30: the first ten customers.

Set up measurement (an hour). Add Search Console and your analytics, a UTM tag on every link you post, and a segment for referrals from chatgpt.com, perplexity.ai and gemini.google.com. Check: a test visit from a tagged link shows up.
List 50 people who have the problem (a day). Past clients, colleagues, people in your industry. Write to each one personally, offering a call or free setup. Check: 10 conversations booked by the end of week two.
Pick one community (ongoing). The subreddit, Slack, forum or Facebook group where your buyers already ask about the problem. Comment usefully for two to three weeks, then post once as “I built this to solve X”, and answer every reply.
Write three comparison pages (a day each). “[Competitor] alternative” for your three closest competitors, built from your real pricing and differences. Put the comparison in your homepage title too, the way Plausible did. Check: search the phrase a month later and note where you appear.

Days 31-60: build what compounds.

Set up the agent stack (an afternoon). OpenSEO hosted or self-hosted, connected to Claude Code, plus the marketingskills pack. Have it find short tool queries and alternative queries your competitors rank for.
Ship one free tool (one to three days). It should solve one small piece of the problem your paid product solves, with a short, searchable name (“invoice parser”, “backlink checker”), and link back to the product. Check: Search Console impressions for its query after four to six weeks.
List where your buyers already shop (days). The Shopify App Store, a CRM’s marketplace, OpenAlternative if you’re open source, or a ChatGPT app if your product works inside a chat. Ask your first customers for reviews, since that’s what marketplace ranking runs on.

Days 61-90: launch, then cut.

Launch once you have proof. Product Hunt once you have a few real testimonials (I’d aim for five, my own rule of thumb) and a list of people who’ll see it on day one; Show HN if developers are your buyers. Sunday has had the least competition on Product Hunt, though the analysis behind that calls its findings correlations, not levers. Expect one day of traffic.
Review by channel (two hours). For each channel: hours spent, signups, paying customers. Keep the two that produced paying customers, drop the rest, and put more hours into the compounding ones that show rising impressions.

The test at day 90: you know which two channels bring paying customers, and you’ve stopped doing the others.

More on getting your brand into AI answers: glossary/share-of-model.

Sources

More of this every day on the channel: https://t.me/aidaily_business