Startup 401(k): How to Set One Up and What It Costs
When to offer a startup 401(k), what it really costs, tax credits worth up to $15,000 for startup costs alone, and how to run one with no HR department.
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, just predictable: they explain the four setup steps, list the tax credits, mention that plans are more affordable than you think, and stop precisely where the interesting questions begin. This guide covers the same mechanics, then answers those questions.
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 fiduciary of a retirement plan for people who work for you?
I wrote this guide 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. That 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?
No federal law requires one, 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, and it favors starting sooner. 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.
The Other Small Business Retirement Plans
A 401(k) is not the only vehicle, and at a very small headcount it is not automatically the cheapest one. Three alternatives come up in practice: a SIMPLE IRA, a SEP IRA, and a solo 401(k) for an owner with no employees other than a spouse. Each one trades design flexibility for far less administration.
| Plan | Who funds it | What the employer must contribute | Testing and annual filing |
|---|---|---|---|
| 401(k) | Employee deferrals, plus employer money if you add it | Nothing in a traditional plan, a set formula in safe harbor | Nondiscrimination testing unless safe harbor, plus an annual return |
| SIMPLE IRA | Employee deferrals, plus required employer money | A dollar-for-dollar match up to 3 percent of pay, or 2 percent to every eligible employee | No testing, and generally no Form 5500 |
| SEP IRA | The employer only | A uniform percentage of pay for everyone eligible, set year to year | No testing, and generally no Form 5500 |
| Solo 401(k) | The owner, acting as both employee and employer | Whatever the owner decides | No testing while the plan covers only an owner and a spouse |
The IRA-based plans buy simplicity at a real price. Employer money in them is immediately and permanently the employee's, so no vesting schedule is available. A SIMPLE IRA caps employee deferrals below the 401(k) limit, a SEP takes no employee deferrals at all, and both leave you less room to shape eligibility or the contribution formula around your cash position.
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. Published pricing from the providers that list their rates (October 2026) puts the one-time setup fee for a new plan at nothing to about $500. The same price lists show annual administration of roughly $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. Per the Department of Labor, investment fees are generally assessed that way and deducted directly from investment returns, so participants pay them, not the company. That is why they rarely appear in the number you are quoted. Against the rest of what benefits cost per employee, the whole administrative line is small.
The other number is the employer contribution, and it dominates everything.
One-time setup fee (amortized across year one) | $500 |
Base annual administration | $1,500 |
Per-participant fees (10 x $50) | $500 |
Administrative subtotal, year one | $2,500 |
Employer match (4 percent of $700,000 participating payroll) | $28,000 |
Total year-one cost before tax credits | $30,500 |
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.
None of this is 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 math is one afternoon in a spreadsheet. List everyone who would be eligible, put a deferral assumption next to each name, and total what the formula you are considering would actually cost you next year. Then run the same total for the other formulas before you talk to a provider. The workbook below has a sheet for each half of that.
| A | B | C | D | E | F | G | H | |
|---|---|---|---|---|---|---|---|---|
| 1 | Employee or role | Eligible yet | Annual eligible pay | Assumed deferral rate | Match formula applied | Employer cost for the year | Contribution credit may apply | Notes |
| 2 | Replace this row with your first employee | Check your eligibility rule, including long-term part-time staff | Be honest, not optimistic | Employees earning $110,000 or less in 2026 | ||||
| 3 | ||||||||
| 4 | ||||||||
| 5 | ||||||||
| 6 | ||||||||
| 7 | ||||||||
| 8 | ||||||||
| 9 | ||||||||
| 10 | ||||||||
| 11 | ||||||||
| 12 | ||||||||
| 13 | Total |
The Three Tax Credits
Small employers starting a new plan can stack three federal tax credits, and together they can cover most of the administrative cost for the first several years. All three are claimed on Form 8881.
The federal retirement law known as SECURE 2.0 shaped two of them. It created the employer contribution credit and raised the startup cost credit to 100 percent of costs for employers with 50 or fewer employees. The auto-enrollment credit dates from the original SECURE Act of 2019.
Eligibility for the startup cost credit is narrower than most summaries suggest, so check it against your own numbers before budgeting around it. Per the IRS, you qualify if three conditions hold: 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.
| Credit | Maximum | Duration | Scales with |
|---|---|---|---|
| Startup costs | $5,000 per year, $15,000 total | 3 years | Non-highly-compensated employees eligible for the plan, at $250 each, with a $500 floor on the cap |
| Employer contributions | $1,000 per employee per year | 5 years | What you actually contribute, for employees earning $110,000 or less in 2026 |
| Auto-enrollment | $500 per year, $1,500 total | 3 years | Nothing, it is a flat amount |
Three limits to keep in view. The first is size: an employer with 51 to 100 employees who earned at least $5,000 the prior year claims 50 percent of qualifying startup costs rather than 100 percent, and the contribution credit percentage falls by 2 points for every employee above 50. The figures in the table assume you are below that band.
The second limit is time: the employer contribution credit declines after year two, running at 75 percent in year three, 50 percent in year four, and 25 percent in year five. The third is pay. The wage ceiling on a counted employee is $110,000 for 2026 per IRS Notice 2025-67, and it is indexed every year.
The Credit Trap Nobody Mentions
All three credits are nonrefundable, 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 Section 39 of the Internal Revenue Code lets an unused general business credit go back one tax year and forward for twenty. A company that reaches profitability inside 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 the plan as a real cash cost now, with a tax asset attached that converts when you turn profitable. Then confirm the treatment with your accountant, because how these credits interact with your specific tax position is exactly the kind of question worth one billable hour.
Nick Anisimov, FirstHR founderTraditional, 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 safe harbor 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.
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.
| Traditional | Safe Harbor | SIMPLE 401(k) | |
|---|---|---|---|
| Employer contribution | Optional | Required | Required |
| Annual testing | Yes, deferral and match tests | Exempt | Exempt |
| Vesting schedule allowed | Yes | No, immediate | No, immediate |
| Deferral limit | Standard | Standard | Lower |
| Best when | Cash is tight and participation is broad | Founders want to defer the maximum | Very small and staying that way |
| Main risk | Failed test forces refunds to founders or extra contributions for everyone else | Contribution cost is locked in | Outgrowing it and having to convert |
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.
How the Deferral Percentage Test Actually Works
The test people mean when they say a plan failed is the actual deferral percentage test. It averages what your highly compensated employees defer as a share of their pay, then compares that average against the same figure for everyone else. Owners and their family members almost always sit in the highly compensated group.
Per the IRS, the plan passes if the highly compensated average stays at or below the more generous of two limits: 125 percent of the rank-and-file average, or that average plus two percentage points, capped at twice that average. A companion test applies the same arithmetic to matching contributions.
Failure is fixable but time-boxed. Refunding the excess to the highly compensated group within two and a half months after the plan year ends avoids a 10 percent excise tax, and certain plans with an eligible automatic contribution arrangement get six months.
A corrective employer contribution made within twelve months also avoids that tax for a plan that tests on current-year data. Leaving the failure uncorrected past twelve months puts the plan's qualified status at risk.
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.
The enrollment rate then rises 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 does more to raise participation than anything else you can change, 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, especially on the required notices. Those 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 that 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 $110,000 or less in 2026 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.
| Formula | Cost at full participation | How it feels to employees | Notes |
|---|---|---|---|
| 100% of the first 3%, then 50% of the next 2% | 4% of participating payroll | Generous, and the standard safe harbor match | Satisfies the safe harbor match requirement |
| 50% of the first 6% | 3% of participating payroll | Similar headline, cheaper for you | Requires a 6% deferral to earn the full match |
| 3% nonelective to everyone | 3% of all eligible payroll | Universal, including non-participants | Satisfies safe harbor without requiring deferral |
| 100% of the first 2% | 2% of participating payroll | Modest but real | A reasonable first step for a tight budget |
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.
Vesting and Forfeitures on Employer Money
Employee deferrals are always fully vested. Employer contributions are where you have a choice: in a traditional plan the outside limits are three-year cliff vesting, where nothing vests until three years of service are complete and then all of it does, or six-year graded vesting that phases in along the way.
Standard safe harbor money vests immediately, with one exception. A safe harbor plan built on automatic enrollment, called a qualified automatic contribution arrangement (QACA), may hold its safe harbor contributions to a cliff of up to two years of service.
When someone leaves before vesting, the unvested employer money becomes a forfeiture, and a forfeiture is not a refund to the company. Plans may use it to pay plan administrative expenses, to reduce future employer contributions, or to increase other participants' accounts.
Under IRS regulations proposed in 2023, which plans may rely on ahead of a final rule, a forfeiture must be used no later than twelve months after the close of the plan year in which it arose, so do not let it sit unallocated.
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, because the regulation lets a newly established plan run a short first plan year only if that year is at least three months long.
Founders who decide in November that they want safe harbor treatment for the current year discover that the answer is no, and that the earliest start is the following plan year.
For a safe harbor match plan, the annual safe harbor notice runs on its own clock. It counts as timely when it reaches each eligible employee at least 30 days and no more than 90 days before the plan year begins.
An employee who becomes eligible after that 90-day mark has passed can instead get the notice no earlier than 90 days before eligibility and no later than the eligibility date itself. That alternative window is what covers a plan starting midyear.
A standard nonelective safe harbor plan, which gives every eligible employee 3 percent instead of a match, no longer has to send that notice: the SECURE Act dropped the requirement for plan years beginning after 2019. An existing plan without a safe harbor match can even switch to that design by an amendment adopted before the 30th day before its plan year ends, or later, before the last day of the following plan year, if it raises the contribution to 4 percent.
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 before an employee may start deferring. 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: which formal roles the provider will accept in writing, such as plan administrator or investment manager. That determines how much liability and work stays with you.
Third comes 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.
Switching 401(k) Providers
Changing providers is a conversion rather than a signup, and the piece that catches employers is the blackout period. While balances move from the old recordkeeper to the new one, participants cannot direct investments, take loans, or request distributions, and that window commonly runs a few weeks.
The Employee Retirement Income Security Act (ERISA) requires written notice of a blackout to participants at least 30 days and no more than 60 days before it begins, so the notice date sets the whole timeline rather than your target go-live.
Ask both providers in writing who drafts that notice, what the exit and asset transfer fees are, and who reconciles balances once the money lands.
Dwight SchruteSales3 people
Darryl PhilbinWarehouse1 person
Kelly KapoorCustomer Service1 personWhat 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, paying only reasonable plan expenses, 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, because 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 Department of Labor is explicit that even if you hire a financial institution or retirement plan professional to manage the plan, you retain some fiduciary responsibility for the decision to select and keep that provider.
Document how you chose the provider, keep the fee disclosures it 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.
Here is what that documentation looks like in practice. Record the providers you compared, their written answers to the questions above, the reason you picked the one you picked, and where the fee disclosures are stored. Then add a dated line to the monitoring log each time you revisit it. Fill it in while you still remember the calls, not two years later when someone asks.
The Deposit Rules for Employee Deferrals
The rule has two halves, and the half that binds you is the first one. Deferrals must reach the plan as soon as they can reasonably be separated from company assets. The 15th business day of the month after the withholding is an outer limit the Department of Labor is explicit about, not a deadline you are entitled to use.
Plans with fewer than 100 participants get a safe harbor on top of that. Deposit within seven business days of the withholding and the amount is deemed timely, with no argument about what was reasonable in your case. Nearly every startup plan fits inside that window, so make seven business days the internal standard and let the payroll connection enforce it.
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. It adds up to a light annual workload.
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: a plan that starts a plan year with 100 or more participants holding account balances generally requires 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 asking about 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.
The IRS has studied this structure directly. Its Rollovers as Business Start-Ups Compliance Project reported that most ROBS businesses either failed or were heading that way, with high rates of bankruptcy, liens, and corporate dissolution, and that many sponsors never grasped that the plan is a separate entity with its own filing duties.
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, head the list because they are so easy to prevent. Using the wrong definition of compensation when calculating the match comes next: plan documents define compensation precisely, and payroll systems often do not follow that definition out of the box. A third is missing notice deadlines, particularly safe harbor and auto-enrollment notices.
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. The cure is ongoing benefits communication, not a one-time announcement.
Mention the plan 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. That habit is what separates a plan people use from a plan people forget.
Frequently Asked Questions
How much does it cost to set up a 401(k) for a startup?
Among providers that publish their prices (October 2026), a new plan carries a one-time setup fee of nothing to about $500, then a yearly administration fee of about $1,000 to $2,000 plus around $20 to $100 for each participant. Some providers also charge a fee set as a percentage of plan assets, and that one usually comes out of employee balances rather than the company budget. But the administrative fee is the small part of the number. If you match employee contributions, the match will typically be ten to twenty times your administrative cost. Model the match before you compare provider quotes, because that is where the real budget decision sits.
When should a startup offer a 401(k)?
No law sets the moment for you, so in practice the time is when you start losing candidates over it or when a state mandate forces the question. Retirement access is close to universal at large employers and much less common at small ones, so offering a plan closes a gap with bigger competitors and puts you ahead of many small ones. Three signals suggest it is time: candidates asking about retirement during interviews, your first employees over forty joining the team, and a state auto-enrollment mandate applying to your headcount. The tax credits also front-load the incentive, since they are largest in the first three years of the plan.
What are the SECURE 2.0 tax credits for a new 401(k)?
Three credits apply to a new plan. Two of them owe their current shape to SECURE 2.0, which added the employer contribution credit and enlarged the startup cost credit; the auto-enrollment credit came from the first SECURE Act in 2019. The startup cost credit covers setup and administration expenses. If no more than 50 of your employees earned $5,000 or more the year before, it pays 100 percent of qualified costs for three years, and the credit for each year is capped at $250 for every non-highly-compensated employee eligible for the plan, never under $500 and never over $5,000. The employer contribution credit pays up to $1,000 a year for each employee paid $110,000 or less in 2026, for five years, with the percentage stepping down after year two. The auto-enrollment credit is a flat $500 per year for three years. You claim all three on IRS Form 8881. If 51 to 100 of your employees cleared that $5,000 mark, the startup cost credit covers 50 percent of costs instead, and the contribution credit loses two percentage points for each employee above 50.
Can an unprofitable startup use the 401(k) tax credits?
Not immediately, and this catches a lot of venture-backed companies. The plan credits are nonrefundable general business credits, so they can only cut income tax you actually owe. A company with no taxable income has no income tax to cut, which leaves the credit with nothing to offset that year. It is not lost, though. Unused general business credits can generally be carried back one year and forward up to twenty, so a startup that turns profitable later can still use them. Treat the credit as a deferred benefit rather than a first-year discount, and have your tax advisor confirm how it applies to you.
Does a new 401(k) plan have to include automatic enrollment?
Generally yes, with important exceptions. A plan set up on or after December 29, 2022 has to sign eligible employees up automatically, at a starting contribution rate somewhere from 3 to 10 percent of pay. That rate then climbs one percentage point a year to at least 10 percent, with 15 percent as the upper limit. Any employee can opt out or pick a different rate. The rule has exemptions that matter for small companies: a business that normally has 10 or fewer employees, a business less than three years old, governmental and church plans, and SIMPLE 401(k) plans. Many startups qualify for one of the first two exemptions at launch and grow into the rule later.
What is the difference between a traditional and a safe harbor 401(k)?
A traditional 401(k) must pass annual nondiscrimination testing, which compares what owners and highly compensated employees defer against what everyone else defers. If rank-and-file participation is low, the plan can fail, and the company must then refund contributions to its highest earners, often the founders, or make an extra contribution for everyone else. A safe harbor 401(k) avoids those tests by requiring a specific employer contribution that is immediately vested, typically a match worth up to 4 percent of pay or a nonelective contribution of 3 percent to everyone. You are trading a guaranteed contribution cost for the certainty that founders can defer the full amount without a refund.
Do I have to match employee contributions?
Not in a traditional 401(k). The plan can run as a pure deferral vehicle with no employer money in it at all, and that is a legitimate starting point for a company that cannot afford a match. But two things push toward matching. A traditional plan with no match tends to see weak participation among lower earners, which is exactly what makes nondiscrimination testing fail, and a failed test can force refunds to the founders. And the employer contribution credit pays back as much as $1,000 per employee a year for five years, so the effective cost of a match in the early years is far lower than the headline number.
How long does it take to set up a startup 401(k)?
With a modern bundled provider, a straightforward plan typically takes a few weeks from signed paperwork to first payroll deferral. Most of that stretch goes to drafting the plan document, connecting payroll, and waiting out the required employee notice periods, so very little of it is work you do yourself. The timing rules themselves are not flexible: a newly established safe harbor plan generally needs an effective date no later than October 1 because its first plan year must run at least three months, and the annual notice for a safe harbor match plan belongs in the window 30 to 90 days ahead of each plan year. If you want a safe harbor plan running in a given year, start the conversation in the summer, not in December.
What is a 3(16) fiduciary and do I need one?
A 3(16) fiduciary is the plan administrator: the party responsible for the operational work of running the plan, including participant notices, distributions, testing coordination, and signing and filing Form 5500. If nobody is formally hired for that role, the employer holds it by default. For a company with no benefits staff, hiring a provider that accepts 3(16) responsibility in writing is the single most useful thing you can do, because it moves the recurring administrative burden and much of the associated liability off your desk. It does not remove your duty to select and monitor providers prudently, which never transfers.
Is a 401(k) required by law for small businesses?
No federal law makes any employer offer a retirement plan. State law is a different matter. A growing number of states now require employers above a set headcount to either sponsor a qualified plan or sign their employees up for a state-run automatic IRA program, and each program sets its own registration deadline and penalties for missing it. Those thresholds, deadlines, and penalty amounts vary widely from one state to the next and change often. With staff in more than one state, review the rules of each state on their own. Running your own 401(k) generally counts as compliance with the state requirement, so you would not also need to join the state program.
Make HR the easy part of running your business
Onboarding, e-signatures, training and core HR in one place. Up and running in 5 minutes.
- 7-day free trial
- No credit card required
- Cancel anytime