At my startup (https://polifinity.com) we do allmost everything manually first and only build when we have "proof" that an idea works.
Examples:
- we do not have onboarding automation, instead a simple make.com trigger is send and then we do onboarding from a team member
- we have not build all parts of the product, but instead use very simple manual processes to simulate a LOT of the more expensive features - only after feedback we actually invest development time
- we have an intelligent reminder feature, that is done using a google sheet + make.com and manual mail; building the feature is not cost efficient anytime soon
- the core of our product is about product strategy. Instead of building a complex interface in the app, we offer (free) workshops that are more effective, but absolutely do not scale at all...
In my past startups we sometimes did even more extreme "do things that don´t scale" stuff:
- giving you personalized recommendations generated within 30 min. from a human
- B2B matching from a web-page, where actually a mail was send and we generated the matches dynamically from different sources => this worked out AMAZINGLY well
- complex reporting from a huge database -> report was generated by a data scientist within 2-3 business days and then based on feedback the feature was never implemented, as every report was different and lots of time required data not actually in the database
Examples: - we do not have onboarding automation, instead a simple make.com trigger is send and then we do onboarding from a team member - we have not build all parts of the product, but instead use very simple manual processes to simulate a LOT of the more expensive features - only after feedback we actually invest development time - we have an intelligent reminder feature, that is done using a google sheet + make.com and manual mail; building the feature is not cost efficient anytime soon - the core of our product is about product strategy. Instead of building a complex interface in the app, we offer (free) workshops that are more effective, but absolutely do not scale at all...
In my past startups we sometimes did even more extreme "do things that don´t scale" stuff:
- giving you personalized recommendations generated within 30 min. from a human
- B2B matching from a web-page, where actually a mail was send and we generated the matches dynamically from different sources => this worked out AMAZINGLY well
- complex reporting from a huge database -> report was generated by a data scientist within 2-3 business days and then based on feedback the feature was never implemented, as every report was different and lots of time required data not actually in the database