How to structure a business: six decisions, in order

The structuring hub names the sequence and this page turns it into a working method. Six decisions, in a fixed order, and the order is the method: each answer shrinks the menu for the next question, which is why the questions get easier as you go and why answering them out of order means choosing from menus that were never yours.

Take the decisions one at a time, each with the question it asks and the pages that answer it.

First: purpose

One sentence, written down before anything else: what am I protecting, from what, and what is it worth. The foundation page supplies the grammar for that sentence, because it defines what an entity can and cannot protect in the first place. A freelancer protecting $30,000 of annual income from ordinary contract risk and a landlord protecting $2,000,000 of equity from tenant injuries have written different sentences, and every later decision reads that sentence back.

This is also the decision most people skip, and skipping it is what the rest of this page calls running backward. Nobody thinks they skipped it. But a structure chosen by copying a friend or a seminar has inherited someone else’s sentence, and the tell is that you can describe your entities but not the specific claim they were built to stop.

Second: entity type

The purpose sentence in hand, the entity question is short, and for most readers it is already answered. Licensed professionals get their answer from a lookup, not a comparison: the state and the board pick the menu, as the professional practice pattern and the variants page explain. Founders building for venture money get a corporation for the reasons the startup pattern lays out. Everyone else gets an LLC, and the choice of entity page explains why that answer is nearly automatic once the container is separated from the tax label.

Notice what the purpose sentence already did: it sorted you into one of those three doors before the entity question was even asked.

Third: tax election

The label, not the box, and decided on different evidence: not what you are protecting but what you are earning. The election math is annual arithmetic against a profit threshold, redone as the business grows, and the default answer for a new business is usually the default label, because the election’s costs are certain and its savings need profit that a new business does not have yet. The one timing exception is the startup chasing the stock exemption, whose clock rewards deciding early, and it was already sorted through the second door.

Out-of-order damage here has a signature: the election chosen before the purpose. The rental in an S-corp is the canonical case, an appreciating asset locked inside a label chosen to fix a self-employment tax problem that rental income never had.

Fourth: number and layering

Now, and only now, the question everyone wants to ask first: how many entities. The answer is the isolation math from the building blocks and the portfolio pattern: count the liability-generating assets, price each additional wall in real annual dollars, and buy walls only where the equity behind them justifies the rent. The purpose sentence does most of this work too. A business that is you and a laptop generates one liability stream and needs one entity; a portfolio generates one per property and needs a deliberate decision about how many baskets.

The backward version of this decision is the most common expensive mistake in the field: complexity bought first, as if entities themselves were the protection. Structure amplifies discipline and never replaces it, and five entities through one bank account are one entity in a courtroom.

Fifth: jurisdiction

Where, decided fifth on purpose, because by now it mostly decides itself. The jurisdiction page carries the rule: you pay the state where the business actually operates no matter what the filing says, so the default is home state, and exceptions must be earned by facts already established in the earlier decisions. The venture-bound corporation earned Delaware at decision two. A real multi-state operation earned its analysis at decision four. A single-state business that starts at decision five, shopping states before knowing its purpose, is how the freelance designer ends up with the Wyoming holding company protecting a laptop, paying two states forever for an answer to nobody’s question.

Sixth: ownership

Last, the question of whose name goes on the interests, last because everything before it determines what there is to own. Solo, spouses, partners with a real agreement, trusts for the family layer: the trusts page and the family pattern carry the machinery, and the default rules explain what the state writes for every question the owners leave silent. For multi-owner anything, this decision is where the operating agreement stops being a formality and becomes the document that decides the next twenty years, and this site’s operating agreement manual will treat it clause by clause.

Reading a structure backward

The sequence doubles as a diagnostic. When a structure feels wrong, find which decision it actually started from, because backward structures announce their starting point. Started at jurisdiction: out-of-state entities and registered agent bills for a home-state business. Started at layering: an org chart with more boxes than liability streams. Started at the election: payroll filings for a business below the threshold, or an appreciating asset inside a corporate label. Started at entity type: a structure copied from a friend in a different profession. Each of those is a real pattern from the fact patterns in this spine, and each unwinds the same way, by asking the skipped question now: what is this protecting, from what, and what is it worth.

Some backward structures are worth fixing and some are cheaper to live with, and telling those apart is its own craft, covered next in restructuring an existing mess.