Demand generation for SaaS startups

Most demand generation writing assumes things you do not have. It assumes 20,000 monthly visits, a brand people type into Google, a sales team waiting for routed leads, and enough budget to absorb one failed channel test without flinching. A seed stage SaaS company has one marketer, two founders who will do some of the work if you ask nicely, and under $10K a month. That gap is why so many early programs burn nine months proving nothing.

Why the standard playbook breaks below $1M ARR

The playbook everyone publishes is a capture playbook. Rank for category terms, retarget site visitors, nurture the list, defend branded search, score the leads. Every one of those tactics assumes demand is already flowing past you and your job is to stand in front of it.

At seed stage it is not flowing. Branded search is probably under 100 queries a month and a good chunk of that is your own team and your investors. Retargeting pools need roughly 300 to 1,000 matched people before most platforms will serve, and you may not clear that in a month. The email list is 400 names, most of whom signed up on launch day and have never opened anything since.

So the work inverts. You are manufacturing conversations one at a time instead of catching them, which is closer to the distinction drawn in demand generation vs lead generation than to either term used loosely. Get that straight before anyone draws a funnel diagram on a whiteboard.

The three channels that work from zero

Three. Not nine. Under $10K a month with one to two people, the only channels that reliably produce sales conversations from a standing start are founder led distribution, a small set of high intent search pages, and hand built outbound to named accounts.

The ordering matters more than the list. Outbound answers questions fastest, founder distribution compounds, and search takes the longest but is the only one of the three that keeps working when you stop touching it. Run outbound and founder distribution in parallel from week one, and start search in month two so it has time to mature.

Founder led distribution: what it returns in 90 days

A founder posting three times a week on LinkedIn from a 1,200 follower base typically reaches 2,000 to 6,000 impressions per post by week eight, and converts that into somewhere between two and six inbound conversations a month. Those numbers are unglamorous and they are also the highest quality pipeline you will see all year, because the person arrives already believing you know the problem.

What makes it work is specificity about the problem, not the product. Write the thing your buyer argues about internally. If you sell security questionnaire automation, write about why the questionnaire took eleven days last quarter and who in the company actually blocks it. Vanta built most of its early awareness this way before it had any meaningful search footprint.

Podcast appearances are the strongest version of the same motion. A niche B2B podcast with 3,000 to 8,000 listeners is unimpressive as a number and excellent as a channel, because the listener spends 45 minutes with you and the episode keeps getting downloaded for two years. Pitch 20 shows, expect four to accept, and treat the transcript as source material for six written pieces.

Here is the honest cost. Founder distribution eats eight to twelve hours a week of the most expensive time in the company, and it does not transfer. When the founder stops, the channel stops within three weeks. Teams that try to hand it to a ghostwriter usually see engagement fall by half, because the specificity that made it work came from the founder actually having the argument.

High intent search: build 12 pages, not 120

The mistake is publishing volume. A seed stage company does not need 120 blog posts, it needs about a dozen pages that sit on queries where someone is choosing software this quarter. Competitor alternatives pages, head to head comparison pages, integration pages for the two or three tools your buyers already run, and two or three pain point pages that describe the problem in the buyer’s own words.

Category terms are a trap at this stage. “Project management software” has enormous volume and you will not rank for it in year one against Monday, Asana and ClickUp, all of which have a decade of links and a hundred times your topical footprint. Let that go and take the small queries nobody is defending.

One caveat worth stating plainly. Search on a new domain takes three to six months to show anything and nine to twelve to matter, and AI Overviews have taken a meaningful bite out of informational click through since 2025. Comparison and alternatives pages have held up better than how to content, which is another argument for building the twelve rather than the hundred and twenty. The wider picture on what organic is worth now sits in the SaaS demand generation benchmarks.

Hand built outbound: 200 accounts, not 20,000

Build the list by hand. Two hundred to four hundred accounts that genuinely match the three or four customers you already understand, researched individually, with a named person and a reason you are writing to that company specifically this week. Tools like Clay and Apollo help you enrich and sequence, but the research is the product and you cannot buy your way past it at this stage.

Run the arithmetic before you get excited. Two hundred accounts at a 3 percent positive reply rate is six conversations. Of those, maybe three become real opportunities and one closes in the next two quarters. That is a slow engine and it is still faster than waiting for a blog post to rank, which is exactly why it goes first.

Two hundred is also the right ceiling for a reason people miss. Past that number the research quality collapses, because nobody sustains genuine per account work across 800 rows in a spreadsheet, and the moment the research thins the reply rate falls off a cliff rather than degrading gently. A 200 account list worked properly beats a 2,000 account list worked lazily by a wide margin, and it beats it on absolute replies, not just on rate.

How to manufacture the first 50 opportunities

You will not find 50 opportunities. You will assemble them from three or four borrowed audiences plus your own effort, and it usually looks like this.

Communities come first. Find the four or five Slack groups, subreddits, or Discord servers where your buyer already complains about the problem, spend two weeks answering questions with no pitch attached, and then keep doing it. This produces a trickle, not a flood: expect one to three conversations a month per active community, and expect to be thrown out of at least one for being too promotional.

Partner audiences are the highest yield and the most underused. Find a non competing tool that sells to the same buyer, offer to write something genuinely useful for their list or run a joint session, and split the registrations. A partner with 8,000 relevant subscribers will typically deliver 60 to 200 registrations against your own list of 400. Do four of these in a quarter and you have built more reach than six months of blogging.

Guest appearances on other people’s content work the same way and take less coordination. A written piece for a partner’s blog, a slot on their customer webinar, a co authored teardown: all of these borrow an audience that already trusts the host. Ask for the registration list or the co marketing rights up front, in writing, because the number of partner sessions that end with nobody agreeing on who owns the leads is remarkable.

Paid newsletter placements sit between the two. A niche B2B newsletter with 10,000 to 30,000 subscribers usually charges $500 to $2,500 for a primary placement and returns 40 to 250 clicks. It is a fine way to test a message quickly, and a poor way to build anything durable. Budget for two placements, not twelve, and put the rest into the channel that showed a pulse. The demand generation budget calculator is the fastest way to see what that split does to your cost per opportunity.

The minimum tracking setup before you spend a dollar

Three things. Not a stack, not a platform, three things that take an afternoon and will still be right in two years.

First, UTM parameters on every link you control, with a written convention stored somewhere the whole team can see. Source is the platform, medium is the type, campaign is the specific thing. Lowercase everything, always, because LinkedIn and linkedin will become two rows in your reporting and you will not notice for four months.

Second, a free text how did you hear about us field on the signup and demo forms. Make it optional and open ended rather than a dropdown, because dropdowns force people into the option you wrote rather than the answer they have. At this stage self reported attribution beats every paid tool you could buy, and it catches the podcast episode and the Slack recommendation that no pixel will ever see.

Third, one CRM object model you commit to. A person belongs to an account, an opportunity belongs to an account, and the opportunity carries an original source field that nobody is allowed to overwrite. HubSpot’s free tier handles this fine and so does Attio. What matters is that you decide the model once rather than discovering in month nine that half your deals have no source at all. The definitions to standardise on are laid out in demand generation metrics for SaaS, and the tooling question is covered in the demand generation software stack.

What $10K a month actually buys

Less than people expect, and the honest allocation looks uncomfortable on a slide because most of it is not media.

That reserve line is the part people delete first and should not. At seed stage the single most valuable thing you can do with budget is double down fast on the one channel that works, and you cannot do that if every dollar is already committed to a retainer. Hold a third back for two quarters.

Notice what is missing. No marketing automation platform, no intent data, no ABM tooling, no agency retainer. Those are all reasonable purchases at $5M ARR and all premature here. The playbooks by ACV band walk through when each of them starts to earn its cost.

When to stop doing things that do not scale

The manual motions above are correct and temporary. The signal that it is time to industrialise is simple: the same manual motion has produced 20 or more qualified conversations, twice in a row, without the founder changing the approach between rounds. Once at seed level, that is luck. Twice, and you have found something.

At that point you do three things in order. Write the motion down so someone else can run it, including the account filter, the message that worked and the objection handling. Hire or contract an operator to run it at two to three times the volume. Then, and only then, put paid budget behind the same message, because now you know what to say and you are buying reach for a proven claim rather than shopping for one.

What you should not do is add a channel. The pull to start a webinar series or a podcast or a partner program the moment something works is strong and it is almost always wrong, because it splits the attention of a team of two across a motion that is not yet finished. The sequencing logic and the kill rules for that decision sit in the 90 day demand generation plan.

Do this in the next two weeks

Pick the account filter and build 200 accounts by hand. Set up the UTM convention, the how did you hear about us field and the CRM object model in one afternoon. Get the founder posting twice a week about the problem, not the product. Book the first four podcast pitches.

Leave paid social alone until one of those has produced a repeatable conversation rate. Write the plan down in a single page so you can tell in 90 days whether it worked, using the demand generation plan template, and keep the wider context of what good looks like at your stage in the SaaS demand generation hub and the B2B SaaS demand generation examples as reference points rather than targets.