Planning
It returns a percentage, usually in the eighties, and the client nods. The question they are actually asking is harder.
The dominant design answers: given these assumptions, what is the probability this plan succeeds? It returns a percentage, usually in the eighties, and the client nods.
The question a client is actually asking is different and harder: what would have to go wrong for me to be in trouble, and would I see it coming?
Across a book, the planning functions that earn their keep are narrower than the feature lists suggest:
| Function | Why it matters |
|---|---|
| Cash flow through time | Almost every real question reduces to this |
| Withdrawal sequencing | Where decades of tax outcome is actually decided |
| Social Security timing | One of the few genuinely large, genuinely reversible decisions |
| Sensitivity, not certainty | Which assumption is carrying the plan |
The rest — and there is a great deal of rest — is usually built to win software evaluations rather than to be opened twice a year.
Planning software is typically the most isolated system in a firm. It holds assumptions about a household that exist nowhere else, and it is fed by hand from data that lives in the portfolio system. Which means the plan is accurate on the day it is built and decays from then on.
Firms rarely notice, because plans are revisited annually and the drift is invisible in between. The cost surfaces at the review, when the numbers on the plan and the numbers on the statement do not reconcile and the conversation becomes about the discrepancy rather than the client.
Planning on the same records as the portfolio, so the plan reflects what is actually held rather than what was held when someone last typed it in — and output framed as which assumptions matter rather than a single probability. Income planning, tax sequencing and estate considerations sit in the same place as the holdings they depend on.
We are not going to describe the modelling approach in detail. What we will do is run it against a real household of yours on a call and let the output make the argument.
Questions this did not answer? Ask them directly — that is what the twenty minutes is for.
Book 20 minutes with Kyle