Marketing tactics that don't scale, and work anyway

Guerrilla marketing for software products, including one tactic you can steal from this site and one we learned by getting it wrong in public

Guerrilla marketing lists are usually written by people who have never run any of it. You get "hand out stickers at a conference" next to "make a viral video", with no note about which ones actually produced anything.

So this list is narrower. These are tactics that work for software products with no budget, and one of them is running on the page you are reading.

1. When a tool dies, be the answer

Products shut down constantly, and when one does it leaves behind something valuable: years of accumulated search demand with nothing to satisfy it. People keep searching the name. The comparison pages keep ranking. The subreddit threads keep getting found. Nobody is defending the position, because the company that owned it no longer exists.

This is the most direct example we have, because we did it. GummySearch was the best-known Reddit audience research tool until it closed to comply with Reddit's API policies. So we wrote the page explaining what happened and what to use instead. It is genuinely useful, it is honest about which alternatives do what, and it catches the search traffic of a product that can no longer catch it.

How to run it: watch for shutdown announcements in your category, on changelogs, on Hacker News and in the communities where the tool's users gathered. Write the page within a week, while the announcement is still being shared. Being early matters more than being thorough here, because the first useful page gets the links.

The one rule that keeps this from being grubby: the page has to genuinely help someone who just lost a tool they liked. That means naming alternatives that are not you, including free ones. A page that only says "use us instead" gets ignored by exactly the people it was written for.

2. Write the definitive answer to one question, once

Somewhere in your category is a question people ask repeatedly and nobody has answered properly. Not a keyword: a question, phrased the way a person in trouble phrases it at eleven at night.

Answer it better than anyone has, in one place, with the specifics everyone else leaves out. Real numbers, the failure modes, the thing that goes wrong on step four. Then link it whenever the question comes up, which will be for years.

This is slow and boring and it compounds harder than anything else on this list, because the answer accrues links and citations while you sleep, and because being the person who wrote it becomes part of your reputation in the niche.

3. Go where the unhappy users already are

Every established product has a population of people who are frustrated with it and saying so in public: support forums, review sites, the replies under a pricing change announcement.

Being useful there works, and pitching there does not. The distinction matters more than it sounds: answering the technical question someone asked, without mentioning your product, builds a reputation that eventually makes the mention welcome. Turning up to say "we solve this, try us" in a thread about someone else's outage makes you the villain of that thread.

The uncomfortable truth is that the second version is faster and the first version is the one that works.

4. Do the work manually, free, for five people

Before the product is good, be the product. Pick five people in your target market, do the thing your software is supposed to do for them by hand, and charge nothing.

It does not scale, obviously. That is not the point. You get five people who have seen the value and will tell others, a precise understanding of where the work is actually hard, and testimonials from real users at a stage when you have none. It is also the fastest way to discover that the thing you built automates the easy half of the job.

5. Sponsor something too small to have a rate card

Newsletters with two thousand engaged readers in a specific niche usually have no sponsorship page, because nobody has ever asked. Email the author, offer a modest amount, and you get an audience that trusts the sender, at a price nobody is bidding against.

The same applies to small podcasts, Discord servers, meetups and conference side events. The rule of thumb is that anything with a media kit is already priced efficiently. Anything without one is not.

6. Publish the failure

Most companies publish the wins. The result is that failure posts are scarce, get read, get shared, and make everything else you say more believable.

This is not humility as a marketing pose. It works because specificity is the only real proof of experience, and failures are more specific than successes: "we grew 40%" tells a reader nothing, while "our AI wrote three cold emails citing sources that did not exist, and here is the one whose domain does not resolve" tells them exactly how much you understand about the problem.

We wrote that exact post about our own product, three days after it happened. It is more persuasive than anything we could have written about how careful we are.

7. The physical world, used precisely

Stickers, flyers and swag are usually a waste, because they are aimed at everyone. Aimed at fifty specific people they are cheap and strange enough to work.

A handwritten note to a customer who wrote a thoughtful bug report. A printed one-page brief left where the twenty people at a niche meetup will see it. Something physical arriving from software is unexpected, and unexpected is the whole mechanism.

What they have in common

None of these can be automated, and that is not incidental. Any tactic that can be run at volume gets run at volume, and then it stops working for everyone including the people who automated it. Cold email did this. SEO did this. Reddit comments are doing it now.

What survives is work that requires a specific human to make a specific judgment: which shutdown is worth writing about, which question deserves the definitive answer, which thread you can help in without pitching. The constraint that makes these tactics unscalable is exactly the constraint that keeps them working.

Which is also the bet behind OctoLoops: we do the searching and the drafting, and you make the judgment and press send. If you want to see what that looks like for your product, the plan is free and takes about thirty seconds.