Founder Playbook · The Bootstrapped Founder
10 tactics from Marybeth Alexander
Marybeth Alexander — Knowledgebase Secrets
Watch the full episode“what if we just focused on this segment of the market right like just doing the knowledge-based thing because at the end of that when we started looking for tools to do this there didn't seem to be a lot of other tools doing this exact thing we figured we could just focus on this aspect of it make it play well with these other tools but just focus on doing this one thing really well”
Scope down to one segment of a crowded category and hold the line
KnowledgeOwl started as a full help-desk ambition inside another company, then narrowed to just the knowledge-base slice when an internal technical writer surfaced the gap. Picking one segment and integrating with the rest makes the yes/no calls obvious: not a forum tool, not light ticketing, not a chatbot. The constraint compounds — going narrow reveals how deep even one slice really is.
“we still track like how often because I think should never say never right you should leave your doors open we always said like we're not doing the chatbot right and now here we are in like 2024 like maybe it makes sense for eventually for us to actually have a chat bot even though for years we're like we're not doing that that's like out of scope”
Track every "no" so the tipping point becomes visible later
Out-of-scope feature requests don't get built, but they do get logged. Counting how often a request lands gives a tipping-point signal — what felt obviously off-scope (a chatbot) eventually becomes obviously core as the market shifts. Saying no today without locking the door shut tomorrow keeps optionality cheap.
“whenever you do get a question from a customer and you haven't answered it before rather than just answering that one customer and like sending them an email or sending them a chat it's just write the quick article and then send them a link to the article and then anybody else who has that can find that article”
Write the article first, then send the link — every time
Instead of answering the same question twice, write the article the first time and link to it. The doc deflects future tickets, gets indexed by Google as top-of-funnel, and shows prospects the depth a marketing page can't. A founder who dislikes support actually ships better docs — the avoidance instinct becomes a flywheel.
“I had like the best customer support call the other day because their marketing team was not upset that their knowledge base wasn't getting indexed they were upset that it was outperforming their dub dub dub like people were getting to the documentation”
Public docs outranking your marketing site is a feature, not a bug
A public knowledge base ranks for long-tail 'does tool X do Y' queries that marketing sites rarely answer in depth. Prospects researching a purchase land in the docs, evaluate the product, and convert before ever requesting a demo. One customer's docs outranked their own marketing site — exactly because docs win the specific commercial-intent queries copy can't.
“the same things that will optimize something for like Google SEO are going to optimize it for your knowledge base generally like having keywords in your title and making sure it's in the URL and like using the words it's the same concept”
Write docs with the same SEO rules as blog posts
The optimization rules for ranking on Google and for ranking inside an internal docs search are identical: keyword in the title, keyword in the URL, the words people actually search for in the body. There's no reason to treat docs as a lower-craft surface than marketing pages — one template wins both ranking systems.
“we started doing that I think in sometimes in 2016 2017 and we've just been doing that ever since but it was as much for us to make it easy for us so we can just focus on like building the product for people”
One plan with usage-based add-ons frees up founder time vs feature gating
KnowledgeOwl gives every customer the full feature set and prices on usage (knowledge bases + authors). Feature gating is sales/marketing complexity that distracts from product work. Collapse tiers to a single feature set with usage-based pricing, then charge separately for the compliance and contract work big buyers actually ask for.
“there was that customer and I think I've heard this from other Founders too that was like like basically like we need you to charge us more money like we're not going to be bble to sell this internally unless we have like a higher level package so that's how we first like became like okay maybe we do need like an Enterprise package”
Charge enterprise because they literally need it to be expensive
KnowledgeOwl created an Enterprise package only after a customer told them they couldn't get internal sign-off unless the price was higher. Same product, same features — the premium covers custom contracts, DPAs, security reviews, indemnification, BAAs. Procurement reality, not feature parity, is what sets enterprise pricing.
“we don't want to be in the position where if we lose one of those Enterprise customers it really hurts us in terms of cash flow so what we'd really love to do is just grow that base of like our you know monthly what we call them like DIY customers that you're just covering your expenses and then those Enterprise passers are sort of like icing on the cake”
Let DIY customers fund expenses; treat enterprise as icing
Twenty enterprise accounts pay roughly the same as 170 legacy DIY customers — concentration risk that shapes every feature decision if it grows unchecked. The deliberate move: fund baseline expenses entirely from low-touch DIY signups, so big contracts become upside rather than oxygen. Refusing to move upmarket IS the strategy.
“most people's knowledge bases do not have the depth to create anything close to like a good generative AI experience and I think that's what people that's the expectation because they're used to chat GPT but the reality is is most people do not have enough information”
AI on top of shallow docs is worse than no AI
Customers ask for AI search and generative answers, but most knowledge bases lack the depth to feed a useful LLM. The retention play is corpus density and currency, not bolting a chatbot on top. Audit doc depth and freshness before adding AI — without it, the AI just hallucinates and erodes trust.
“I want to be an employee own business like I want to be able to like sell the company but I want to sell it to the people that are like working in the company so that then they can profit off the company too”
Decide the exit before you need one — destination shapes every decision
Marybeth's stated end-state is selling KnowledgeOwl to its own employees (employee-owned + B Corp). That destination directly justifies staying small, refusing fast growth, and prioritizing calm over chasing every dollar. Define where the company is going before optimizing the daily decisions, and the rejected feature requests stop feeling like missed opportunities.