Your portal says one number. Your Excel says another. Now what?
Every sponsor with an investor portal knows the ritual: run the distribution in your waterfall model, key the results into the portal, and check — line by line — that the portal's numbers match yours. When they don't, the afternoon is gone.
This page is for the moment you searched some version of "why doesn't my investor portal match my Excel."
Why they disagree
Your portal doesn't run your waterfall. It runs its waterfall — a configured approximation of yours, built from dropdowns and settings during onboarding. Two engines, two sets of assumptions:
- Day-count and timing conventions. Your model accrues pref Actual/365 from the capital call date; the portal defaults to Actual/Actual monthly. Small drift, compounding quarterly, surfacing as an IRR question from your most detail-oriented LP.
- Tier-logic edge cases. Your catch-up is 50/50 until the GP reaches 20% of profits above the pref; the portal's catch-up toggle means something subtly different. Your side letters and specific-LP terms don't fit the configuration at all — so they live in your Excel anyway.
- The re-keying itself. Even when both engines would agree, the numbers travel by hand. Fat-fingering LP 23's distribution into LP 24's row is not a hypothetical.
And the portals know it. Service terms disclaim responsibility for variances against your model. Support's answer to "your number is wrong" is "our number follows your configuration."
The standard advice is backwards
The vendors who rank for this problem all propose the same fix: migrate off Excel. Rebuild your waterfall in their engine, retire the model, adopt theirs as the system of record — six to twelve months, and now your LPA lives in someone else's configuration language, verified against... the Excel you kept anyway.
Notice what the industry's own workflow admits: when the platform and the model disagree, the model wins. Your Excel already is the system of record. The problem was never the model — it's that your counterparties can't see into it.
Keep the model. Publish from it.
Formualizer takes the opposite bet. Your workbook — formulas, named ranges, tier logic, side-letter carve-outs — runs as-is on our engine. Each LP gets a permissioned live statement rendered directly from the formula graph:
- Nothing to reconcile. The statement is the calculation. There is no second engine to drift from your model, and no re-keying step to fat-finger.
- Every number defends itself. An LP clicks their distribution and traces it through pref, catch-up, and promote to your actual cells — with a plain-English explanation grounded in the trace.
- Republishing shows its work. Fix the model, upload new data, republish: LPs see exactly what changed and why, versioned.
- $18,400Preferred Return (8%)Waterfall!D14= 8% × capital account × period
- $2,500,000Capital AccountCapAccts!C7
›Preferred return of $18,400 = 8% annual on the $2,500,000 capital account, accrued for the quarter (Waterfall!D14).
›LPs see what changed and why — the promote moved because Q2 proceeds cleared Hurdle 2.
"But we already pay for a portal"
Keep it — portals are good at documents, K-1s, and capital-call workflows. Formualizer replaces the part where you hand-compute distributions and re-key them: the calculation layer. Sponsors run us alongside their portal, or link Formualizer statements from it.
FAQ
Why don't my portal's distribution numbers match my waterfall model?
Because the portal recomputes distributions with its own engine from configuration you set at onboarding, while your model is the negotiated logic. Day-count conventions, tier edge cases, side letters, and manual re-keying each introduce drift.
Do I have to rebuild my waterfall to fix this?
No. Rebuilding is the standard vendor answer and it inverts the trust hierarchy — platforms get verified against the model, not the other way around. Formualizer executes the model itself.
What if my model has an error?
Then it's visible and traceable — which is the point. A trace that ends at a wrong cell is a five-minute fix and a transparent republish, instead of a quarter-end fire drill discovered by an LP.
How fast can this be live?
48 hours from a sanitized workbook: paid concierge pilot ($2–3k setup), then a monthly per-LP fee. If the pilot doesn't earn the subscription, walk away.
48 hours from workbook to live statements.
Concierge pilot: send a sanitized workbook, we map it and stand up your counterparty portal in 48 hours. Pilots are paid — $2–3k setup, then a monthly fee per recipient. If the pilot doesn't earn the subscription, walk away.