Startup 401(k): How to Set One Up and What It Costs

Almost every guide to setting up a startup 401(k) is written by a company that sells 401(k) plans. That does not make them wrong, but it does make them predictable. They explain the four setup steps, they list the tax credits, they mention that plans are more affordable than you think, and they stop precisely where the interesting questions begin.

Here are the questions they stop before. What does the plan cost once you include the part that is not the administrative fee? Can you actually use those tax credits if you are not yet profitable? What work lands on your desk every year, given that you do not have anyone doing benefits full time? And what are you signing when you become the plan fiduciary of a retirement plan for people who work for you?

This guide covers the mechanics that every article covers, and then it covers those four. It is written for an owner or founder who is going to make this decision personally, alongside twenty other decisions, without a benefits team. I build the onboarding and employee record infrastructure that a plan like this sits on top of at FirstHR. This is general information rather than legal, tax, or investment advice, and a retirement plan is one of the places where paying a professional is genuinely worth it.

What Is a Startup 401(k)?

A startup 401(k) is an ordinary 401(k) plan sponsored by a young or small company. There is no separate legal category. The rules that apply to a fifteen-person plan are the same rules that apply to a fifteen-thousand-person plan, which is exactly why the topic feels heavier than it should when you first look at it.

What is genuinely different at startup scale is not the law but the delivery model. A modern bundled provider packages the recordkeeping, the plan document, the investment lineup, and much of the compliance work into one subscription that connects to your payroll, which is why a plan that used to require a benefits manager can now be run by a founder in a few minutes a month.

The tradeoff is that bundling makes it easy to sign up without understanding what you have agreed to. That is fine right up until the first year that something goes wrong, which is why the fiduciary and plan-year sections below exist.

When Should You Offer One?

There is no legal trigger, so the honest answer is that you should offer one when it starts costing you more not to. In practice three signals arrive, usually in this order: candidates begin asking about retirement in interviews, your first hires over forty join the company, and a state mandate lands on your headcount.

That gap is the argument. A candidate weighing your offer against one from a two-hundred-person company is comparing a package, and retirement is one of the few line items where small employers routinely lose by default rather than by budget. Adding it moves you from the minority to the majority on a benefit that costs far less than health coverage, which is why it tends to appear early on any serious small business benefits roadmap.

There is also a timing argument that runs the other way. The startup tax credits are front-loaded into the first three years of the plan, and the employer contribution credit into the first five. Waiting does not preserve them. It just means you claim them later, at a headcount where the per-employee caps may fit you worse than they do today.

The case against starting now is real too, and it is mostly about cash. If you cannot fund a match, a traditional plan with weak participation can fail its annual testing and force refunds back to the founders, which is a bad outcome dressed up as a benefit. Read the plan types section before deciding that a no-match plan is the cheap option.

What It Actually Costs

The cost of a startup 401(k) splits into two numbers that differ by an order of magnitude, and every provider quote you receive will emphasize the smaller one.

The administrative number is what shows up on the pricing page. Setup fees commonly land between a few hundred dollars and roughly $2,500 one time. Annual administration typically runs $1,000 to $2,000 as a base, plus somewhere around $20 to $100 per participant per year. Many providers also take a percentage of plan assets, and that percentage is usually charged to employee balances rather than billed to the company, which is why it rarely appears in the number you are quoted. Against the rest of what benefits cost per employee, that is a small line.

The other number is the employer contribution, and it dominates everything.

Nothing about that arithmetic is unusual. It is what a modestly matched plan costs at that headcount. The point is that the decision you spend a week researching, which provider to use, moves about $1,000 a year, while the decision you make in ten minutes, whether and how much to match, moves nearly $30,000.

Which is not an argument against matching. It is an argument for doing the match math first, then choosing a provider against a budget you have already set. It also reframes what the tax credits are for, because the largest of them applies to the contribution, not to the fees.

The Three Tax Credits

SECURE 2.0 created a stack of three credits for small employers starting a new plan, and together they can cover most of the administrative cost for the first several years. All three are claimed on Form 8881.

Eligibility for the startup cost credit is narrower than most summaries suggest, and it is worth checking against your own numbers before budgeting around it. Per the IRS, you qualify if you had 100 or fewer employees who received at least $5,000 in compensation in the preceding year, you have at least one plan participant who is a non-highly-compensated employee, and you did not have another retirement plan covering substantially the same employees in the three years before the new plan.

Two limits to keep in view. Employers with 51 to 100 employees claim reduced amounts rather than the full ones, so the figures above describe the smaller end of the range. And the employer contribution credit declines after the second year, at 75 percent in year three, 50 percent in year four, and 25 percent in year five.

The Credit Trap Nobody Mentions

These are nonrefundable credits, which means they reduce income tax that you owe. If you owe no income tax, there is nothing for them to reduce.

That sentence quietly disqualifies a large share of the audience these credits were marketed to. A venture-backed company burning capital has no taxable income by design. It can set up a plan, incur every qualified cost, be fully eligible on paper, and receive nothing in the year it spends the money. The provider selling you on a plan that pays for itself is not lying. They are describing a profitable business.

The credits are not forfeited, though, and this is the part that also goes unsaid. They flow into the general business credit, and unused general business credits can generally be carried back one year and forward up to twenty. A company that reaches profitability within that window can still use them.

The practical consequence is a change in how you frame the decision rather than in whether you make it. Do not tell your board that the plan is free for three years. Model it as a real cash cost now with a tax asset attached that converts when you turn profitable, and confirm the treatment with your accountant, because the interaction between these credits and your specific tax position is exactly the kind of thing that is worth one billable hour.

Traditional, Safe Harbor, or SIMPLE

Three structures are realistically available to a small employer, and the choice is essentially a trade between guaranteed contribution cost and annual testing risk.

A traditional 401(k) has the most flexibility. You choose whether to contribute, you can attach a vesting schedule to employer money, you can offer both pre-tax and Roth deferrals, and you can change the match year to year. In exchange, the plan must pass annual nondiscrimination testing that compares deferrals by highly compensated employees against deferrals by everyone else.

A safe harbor 401(k) buys its way out of that testing. Per the IRS, a safe harbor plan must provide employer contributions that are fully vested when made, and in exchange it is not subject to the complex annual nondiscrimination tests that apply to traditional plans. The standard designs are a match reaching roughly 4 percent of pay for employees who defer, or a nonelective contribution of at least 3 percent to every eligible employee whether they defer or not.

A SIMPLE 401(k) is available to employers with 100 or fewer employees who received at least $5,000 in compensation in the prior year. It also skips testing and requires immediate vesting, but it carries lower deferral limits and less design flexibility, which is why growing companies frequently outgrow it.

For most founders the real question is narrower than the table suggests: how much do you personally intend to defer? If the answer is the annual maximum, a traditional plan at a company where rank-and-file participation is thin will very likely fail testing and hand part of your contribution back to you as a taxable refund. Safe harbor exists precisely to prevent that, and the cost of the required contribution is the price of that certainty.

The Auto-Enrollment Requirement

If you are setting up a new plan, assume auto-enrollment applies to you until you confirm otherwise. Plans established on or after December 29, 2022 generally must automatically enroll eligible employees at an initial rate of at least 3 percent and no more than 10 percent, increasing by one percentage point each year until it reaches at least 10 percent and no more than 15 percent. Employees can opt out or choose their own rate at any time.

The exceptions are the part that matters for a small company, and two of them cover a lot of startups. Businesses that normally employ 10 or fewer people are exempt. So are businesses in existence for less than three years. Governmental plans, church plans, and SIMPLE 401(k) plans are also outside the requirement.

Beyond compliance, auto-enrollment is the single most effective lever on participation, and participation is what keeps a traditional plan passing its tests. The Department of Labor publishes a plain-language guide to automatic enrollment plans for small businesses that is worth reading before you finalize the design, particularly on the required notices, which are easy to get wrong and consequential when you do.

Should You Match, and How Much?

You are not required to contribute anything to a traditional 401(k). You can offer the plan purely as a place for employees to defer their own pay, and for a company genuinely short on cash that is a defensible starting point.

Two forces push the other way. The first is testing: a no-match traditional plan tends to produce weak participation among lower earners, and weak participation among lower earners is the mechanism by which the plan fails and refunds money to the founders. The second is the employer contribution credit, which reimburses up to $1,000 per employee earning $100,000 or less for five years, at full value in the first two. Both are reasons to treat the match as part of total compensation rather than as an optional extra.

On the design itself, the common formulas are worth knowing because they cost very different amounts for very similar perceived generosity.

The nonelective option deserves more attention than it usually gets from cash-conscious employers, because it behaves differently from a match. It reaches employees who do not contribute, which is often the exact group whose non-participation causes testing problems, and it is far easier to explain to a workforce that does not think about retirement accounts.

Whatever you choose, write down the reasoning somewhere durable and put the formula in your employee handbook in language a new hire can act on. A match that nobody understands well enough to earn is money you have budgeted and failed to spend on the thing you budgeted it for.

How to Set Up the Plan

The mechanical sequence is short. Most of the elapsed time is document preparation, payroll connection, and required notice periods rather than work you personally perform.

On timing, one deadline is worth writing on a wall. A new safe harbor plan generally has to be effective by October 1 to count for that plan year, and the notice must reach employees before the plan year begins. Founders who decide in November that they want safe harbor treatment for the current year discover that the answer is no, and that the next window is twelve months out.

One design question worth settling explicitly at step one is who becomes eligible and when. A plan cannot require more than one year of service as a condition of participation, and separate rules extend eligibility to long-term part-time employees, so assuming that part-time staff are automatically excluded is a common and correctable mistake. Decide the eligibility rule deliberately and make sure your waiting period matches what the plan document says rather than what you assumed.

Choosing a Provider With No HR Team

The provider comparison you actually need is different from the one that is easy to find. Almost every published comparison is authored by one of the providers being compared, and the criteria in those articles tend to be the criteria on which the author wins.

For a company with nobody doing benefits full time, the ranking of what matters is fairly specific. Cost transparency comes first, because opaque fee structures make every other comparison meaningless. Then fiduciary coverage, because that determines how much liability and work stays with you. Then payroll integration, because that determines the monthly time cost, which is the currency that actually matters when you are running HR for a small business alongside everything else. Investment lineup quality matters, but it is rarely the deciding factor between mainstream providers at this size.

Ask all seven in writing and compare the answers side by side. Providers whose model depends on you not asking question one or question three tend to answer both of them evasively, which is itself the information you were looking for.

What You Are Signing Up For

When you sponsor a 401(k), you become a fiduciary under ERISA with respect to that plan. That is a real legal status with a real standard of conduct attached, and it is the part of the decision that gets the least attention relative to its weight.

The Department of Labor summarizes the duties as acting solely in the interest of participants and beneficiaries, acting for the exclusive purpose of providing benefits, carrying out duties with the care and skill of a prudent person familiar with such matters, following the plan documents, and diversifying plan investments. Fiduciaries can be held personally liable for losses caused by a breach.

That sounds heavier than the reality for a well-run small plan, and it should be read alongside the fact that most of the operational responsibility can be transferred to people who do this professionally. Understanding which roles exist and who holds each one is the whole exercise.

The single sentence to take from this: the DOL is explicit that even if you hire a financial institution or retirement plan professional to manage the plan, you retain fiduciary responsibility for the decision to select and keep that provider. So document how you chose them, keep the fee disclosures they gave you, and revisit the decision periodically. That documentation is inexpensive to create and is exactly what you would want to have if the choice were ever questioned.

The Plan Year, Start to Finish

Setup is a project. Operation is a rhythm, and knowing the rhythm before you commit is how you judge whether you can carry it.

With a bundled provider that has accepted 3(16) responsibility and connected to your payroll, your share of that calendar is roughly this: confirm census data once a year, approve any distributions or loans that come up, sign what needs signing, and read the testing results when they arrive. Founders commonly describe it as a few hours annually.

Without that arrangement, the same calendar is yours to run, and the pieces that go wrong are the ones with dates attached. That is the concrete reason the fiduciary questions above are worth pressing on before you sign rather than after.

Two documents sit underneath all of it and should live somewhere your team can actually find them: the plan document itself, which governs everything, and the summary plan description that participants are entitled to receive. Storing both alongside your other employee records rather than in a provider portal nobody logs into is a small habit that saves real time later.

One threshold to watch as you grow: plans that reach 100 or more participants generally require an independent audit attached to the annual filing, which adds a meaningful recurring cost. It is far enough away for most small employers not to worry about, and close enough that a fast-growing company should know it is coming.

State Retirement Mandates

No federal law requires you to offer a retirement plan. A growing number of states do, and for many small employers this is what converts the question from optional to scheduled.

The typical structure is the same across states even though the details are not. Employers above a stated headcount must either sponsor a qualified retirement plan or register for a state-run automatic IRA program that enrolls their employees by payroll deduction. Sponsoring your own 401(k) generally satisfies the mandate and exempts you from the state program.

What differs by state, and differs a lot, is the headcount threshold, the registration deadline, the default contribution rate in the state program, and the penalty for missing registration. Several states have lowered their thresholds over time, which means an employer who was exempt when they last checked may not be now. State-by-state mandate rules change frequently enough that they should be verified against the state program directly rather than against any article, including this one.

If you employ people in more than one state, check each state separately. The obligation follows where the employee works, and a remote team can put you inside three different mandate regimes without anyone noticing until a notice arrives.

A Note on Funding a Business With a 401(k)

A meaningful minority of people searching for a startup 401(k) want something completely different from what this article describes, and it is worth separating the two clearly before someone acts on the wrong one.

Everything above this section concerns offering a retirement benefit to your team. If your question is about your own retirement savings as a source of business capital, the answer requires specialist advice rather than a guide, and the cost of getting it wrong is your retirement.

Mistakes That Cost Small Employers

The failures that actually happen at small companies are not exotic. They cluster in a handful of predictable places.

Budgeting the tax credits as if they were cash is the first, and the credit trap section covers why. The second is choosing a traditional plan to avoid the match cost, then failing testing and refunding contributions to the founders, which achieves neither the savings nor the benefit. The third is treating the administrative quote as the cost of the plan, which understates it by roughly a factor of ten once a match is in place.

Then there are the operational ones. Late deferral deposits, discussed above, are the most common and the most preventable. Using the wrong definition of compensation when calculating the match is a close second, because plan documents define compensation precisely and payroll systems often do not match that definition out of the box. Missing notice deadlines, particularly safe harbor and auto-enrollment notices, is third.

The last one is quieter and harder to correct: setting the plan up and never telling anyone about it properly. A plan with 30 percent participation is a plan you are paying to administer while receiving almost none of the recruiting or retention value you bought it for. Mention it in the offer, cover it during onboarding while the new hire is already filling out paperwork, and say the match formula out loud at least once a year in language that includes an actual number. Treating it as an ongoing benefits communication problem rather than a one-time announcement is what separates a plan people use from a plan people forget.

If you want the current contribution limits at any point, the IRS cost-of-living adjustment page is the source that stays correct, since the deferral limit, the catch-up amounts, and the compensation cap are all adjusted annually.

Frequently Asked Questions