This page explains why a single average return misleads, what the 4% rule actually rests on, and what a "success rate" really measures. It exists because "the pot lasts 22 years" is an answer to a simpler question than the one retirement asks.
Our pot-longevity table says a £500,000 pot supports £30,000 a year for 22 years at 2.5% real growth. That number is honest as far as it goes, and it says so on the page: it assumes the same smooth return every single year.
Real markets deliver their long-run average as a jumble: a strong year, two flat ones, a bad one, a recovery. For someone still saving, the order barely matters, because nothing is being taken out along the way. For someone drawing an income, the order can matter as much as the average itself. That effect has a name, sequence of returns risk, and it is easiest to see with two small worked scenarios.
Both scenarios start with £500,000, withdraw £30,000 at the start of each year, and experience exactly the same three annual returns: +25%, 0% and −20%. The only difference is the order they arrive in.
| Year | Good years first (+25%, 0%, −20%) | Bad year first (−20%, 0%, +25%) |
|---|---|---|
| Start | £500,000 | £500,000 |
| After year 1 | £587,500 | £376,000 |
| After year 2 | £557,500 | £346,000 |
| After year 3 | £422,000 | £395,000 |
Two things are worth sitting with. First: with no withdrawals, both orders end at exactly £500,000, because multiplication does not care about order. The gap exists only because money is being taken out along the way, so a bad early year means selling more units at low prices, and those units are not there for the recovery. Second: this was only three years and one bad year. Across a 30-year retirement with a deep bear market in the first five years, the same mechanism is the difference between a plan that coasts and a plan that needs surgery.
Information and guidance only. Not personal advice. The scenarios above use illustrative returns chosen to make the mechanism visible, not to represent any market. Investment values can fall as well as rise and you may get back less than you put in. Verify decisions with a qualified FCA-authorised adviser before acting.
The famous rule of thumb says: withdraw 4% of the pot in year one, raise it with inflation each year, and a balanced portfolio historically lasted at least 30 years. The Sonuswealth engine uses 4% as its default starting rate too. It is a reasonable opening move. It is not a law of nature.
The rule comes from 1990s research on historical United States market data starting in 1926. Three honest caveats follow from that. The United States was one of the best-performing markets of that century, so the same exercise run on other countries' histories generally supports lower starting rates. The research ignored fees, and a typical 0.7% a year of platform and fund charges works directly against the withdrawal. And "it worked in every historical 30-year window" is a statement about the past: the future is not obliged to stay inside the range the past happened to explore.
None of that makes 4% useless. It makes it what it always was: a first sentence in a conversation, to be adjusted for your age, your other income, your fees, and how much flexibility your spending has. Which raises the question of what "it worked" even means.
Planning tools often express an answer as a success rate: the share of simulated or historical market paths in which the plan's spending was sustained to the end. The phrase sounds binary, succeed or fail. The reality it describes is gentler than that.
Take a plan quoted at a 90% success rate. That does not mean that in 10% of futures the money vanishes and nothing can be done. It means that in roughly 1 path in 10, somewhere along the way, the plan as originally written needed changing: spending trimmed for a few years, a purchase deferred, part-time income added. Real households notice a bad decade as it happens and adjust; the simulation's "failure" is usually the point where a real person would have made a modest course correction years earlier.
This matters in both directions. A nervous saver can over-save for decades chasing a 99% figure, paying for that certainty in years of working life. A fixed-cost household, where most spending is contractual and cannot be trimmed, genuinely does need a higher bar. The success rate is not the answer: the right reading of it depends on how adjustable your life is.
The single biggest input into "will it last" is not the growth rate you cannot control. It is how much of your spending could flex if the first decade turned out badly. Two households with identical pots sit at opposite ends of a spectrum.
| Fixed spender | Flexible spender | |
|---|---|---|
| Spending shape | Mostly contractual: rent or mortgage, support for family, fixed commitments | Large discretionary share: travel, hobbies, gifts that could pause in a bad stretch |
| Response to a bad decade | Little room to cut, so the pot absorbs the whole shock | Trims spending 10 to 25% for a few years, so the pot sells fewer units at low prices |
| Sensible success bar | High: the plan has to work in the bad paths because spending cannot bend | Lower is defensible: the "failed" paths are the ones where flexing happens |
| Sequence risk exposure | Full | Partly self-insured by the ability to adjust |
Flexibility also shows up in the plain arithmetic. On the pot-longevity table's middle case, a £500,000 pot at £30,000 a year lasts 22 years. Model the well-documented tendency of spending to fall later in retirement (here, 25% lower from age 75 for someone retiring at 60) and the same pot lasts 24 years. Modest, but it points the honest direction: the question is not only "how much do I have" but "how adjustable is my life if the market is unkind early".
One more honest caveat belongs here: spending does not only fall with age. Late retirement can bring the steepest costs of all, which is why the care-fee cliff has its own page.
Methodology. Sequence scenarios: £500,000 starting pot, £30,000 withdrawn at the start of each year, annual returns applied to the remainder (order A: +25%, 0%, −20%; order B reversed); both orders end at £500,000 when nothing is withdrawn. Longevity figures use rules bundle UK-2026.1.1 defaults: 5% nominal growth, 2.5% inflation, so 2.5% real, withdrawals at the start of each year in today's money. Success-rate language on this page is definitional; no backtest percentage is quoted as a result. All arithmetic machine-checked before publication. Published 7 August 2026 · figures last verified 7 August 2026.
Sonuswealth is pre-launch. The app models your pot, your state pension and your spending flexibility together, kept current as rules change.