By Artem Korsakov ·
In short: a business network launch is best started with a small group that already has a shared task. The first product must deliver value with a limited number of participants. What needs testing is completed actions and repeat use. The number of sign-ups says little here.
A new platform faces an awkward question: why would anyone create a profile if there is nobody there yet? The answer “there will soon be lots of companies” shifts the risk onto the user. It is stronger to offer an action that is useful today: putting together a project history, confirming a role or preparing a clear list of contractors for one purchase.
Choose a limited community
A business network for all companies at once is too vague. Participants may work in different regions, buy different services and never cross paths. The number of profiles grows, but no useful contacts appear.
You can start with the team of an industrial project, members of an industry association or suppliers of a specific type of service. These are segment hypotheses; demand still has to be tested. First find out who is already collecting recommendations by hand and why the current way is inconvenient.
Find value before the network effect
The first version can work as a shared folder of experience: companies describe their roles, attach permitted materials and agree on what can be shown. Even if there are few new partners yet, you get a useful project history.
Another option: a closed list of contractors for a coordinator. They get comparable cards, and participants fill in their information once instead of repeated questionnaires. But these are two different tasks. It is better to choose one and test it before building a general catalogue, feed and chat.
Why manual work at the start is acceptable
In his essay Do Things that Don't Scale, Paul Graham advises personally recruiting the first users and finding a narrow group in which a useful concentration of participants can be built quickly. This is an entrepreneur's advice; it does not guarantee success for any platform.
For testing, you can manually help companies fill in cards and agree on roles. It is important to record where help is needed. If the founder personally creates all the value every week, this must not be passed off as an already working self-sustaining network.
Which scenario to include in the first version
Suppose the participants of one completed project want to use it as proof of experience. A training scenario looks like this:
- the coordinator creates the project and states its boundaries;
- participants add their own roles;
- the client's representative confirms the agreed statements;
- the owner defines open and closed fields;
- a company sends the card to a potential buyer;
- the recipient asks a question about specific experience.
Registration is needed only where it enables an action and access rights. A guest does not have to create a profile to read a permitted description. The invitation should explain what needs to be confirmed and why it is useful to the recipient.
Which mechanics are better postponed
A public overall rating, an endless feed and recommendations based on a large network require data that does not exist yet. They can complicate the first product and create an appearance of activity. First check whether people need the confirmed cards themselves.
At an early stage, AI is useful for preparing drafts and finding incomplete data. It can suggest what information is still needed, but it must not fill an empty network with invented companies or assumed connections. Training data in a demo is always marked separately.
How to measure the first results
Record the whole path: invitation, opening, filling in the role, confirmation, sending the card, a useful reply. This shows at which step participation stops. A refusal to confirm a text sometimes indicates a dispute over content; the interface has nothing to do with it.
Separately measure returns without a reminder and the time spent on manual support. Do not set an attractive conversion target without grounds. Compare several consecutive launches and see whether specific obstacles are being removed.
In the business network prototype, you can walk through a limited scenario before talking about a large platform. A real launch will first need real participants, a clear data owner and an agreement on confirmation. If there is no value in a small group, scaling will not fix that by itself.
Related reading: What a business relationship map can tell you about a company; Contractor work quality: why it is hard to prove before a deal.
Частые вопросы ### Do you need a mobile app right away?
Usually, a page that works well on phone and desktop is enough for testing. An app makes sense if a recurring scenario really requires its capabilities. Having an app does not by itself solve the empty-database problem.
Can you charge the first participants?
Yes, if there is a specific paid benefit and an honestly described scope of service. For example, preparing a closed catalogue or organising confirmations. Promising a future flow of requests without a proven mechanism would be unfounded.