How do sponsors actually calculate the IRR they show in the deck when the cash flows are uneven

Your XIRR setup is almost certainly correct, and the gap is real. The most common source of that spread is that sponsors calculate IRR on equity cash flows only, then add the pref layer back in at the summary level without running it through the same time-value calculation. So their 17.4 percent may be the unlevered or equity-only return on a compressed timeline, while your XIRR is correctly penalizing the early quarters when distributions are thin and capital is fully deployed.

The other frequent culprit is how the exit proceeds get split. If the sponsor is crediting the full projected sale price to IRR before subtracting fees, promote, or a GP co-invest that sits ahead of LP equity at exit, your model will come out lower because you are presumably modeling what actually lands in the LP account.

What is the assumed exit price they are discounting back, and does the waterfall show a catch-up provision before LP gets the residual? That single number, combined with whether there is a catch-up, will tell you whether the gap is methodology or something more structural.

Your XIRR setup is probably fine. The gap is almost certainly coming from the sponsor's timing assumptions, not a more sophisticated calculation on their end.

The most likely culprits, in order of how often I see this:

Day-zero cash flow. If you modeled your $50k outflow at day zero and the sponsor backdated or offset it differently, XIRR is sensitive to that. A one-month shift in your initial outflow date moves IRR more than most people expect on a 38-month hold.

Exit date precision. XIRR is date-sensitive. "Month 38" in your model might be landing on a different date than what the sponsor plugged in. Try moving your assumed exit date two to four weeks in either direction and watch what happens to your output.

Pref accrual vs. pref distribution. This is the assumption doing the most work in your gap. If the sponsor is showing accrued but undistributed pref as a lump at exit rather than as the quarterly cash flows you modeled, their IRR calculation is treating that capital as returned later, which inflates the rate. The sponsor may be running IRR on the total-proceeds-at-exit model while you are correctly discounting each quarterly payment as it arrives. Those two approaches produce different numbers on the same deal.

A simple annualized return dressed as IRR. Some sponsors calculate equity multiple over hold period, then annualize with a simple formula rather than a true discounted cash flow. A 1.8x over 3.17 years annualizes to roughly 17.5 percent that way. That matches their number almost exactly and would explain the whole gap.

That last one is worth raising directly. Ask the sponsor to provide the actual cash flow table behind the IRR figure, with dates and amounts, so you can reconcile. Any sponsor worth committing capital to will hand that over without friction.

The risk you have not mentioned: even if you reconcile the IRR figure, the projected exit proceeds are doing significant work in a 38-month deal. What is the sponsor assuming on exit cap rate, and how does that compare to where the asset is trading today?