How it works
Most “upgrade chance” tools score how good your profile looks. This one doesn’t. It asks the only question that decides it: will Delta clear enough seats to reach as far down the list as you are sitting?
This is a steady-state estimate. It can’t see irregular operations — weather, outages, cancellations, rebooking cascades — which override every assumption here. The numbers are informed estimates, not built from actual upgrade-outcome data.
You clear an upgrade when the seats that drop to the upgrade list reach down at least as far as your spot in line. So the model estimates two things and compares them:
Supply — how many premium seats will actually clear to the complimentary list (not how many look open today; the two are very different).
Demand ahead of you — how many people sit ahead of you in line, built from the field of higher-status flyers plus where you fall inside your own status tier.
If supply reaches your position, you clear. The output is the probability that it does.
Both sides are estimated as full distributions, not single numbers — because real flights are uncertain. That’s why the result is an honest probability, not a false-precision point.
Everyone above your tier is ahead of you. Within your tier, Delta breaks ties in a strict order — cabin/fare purchased, then Million Miler, then fare experience, then Delta Reserve card, then corporate-coded ticket. The model follows that exact order, so a lower tiebreaker can never leapfrog a higher one. Things you can’t know relative to the rest of the plane (your MQD standing, exactly when you requested) can’t shift the headline either way.
Route is the biggest supply lever. A hub-to-hub business run sells its front cabin hard and clears very little for free; a thin off-season leisure route clears more. Type two airports and the tool infers the route type; or pick it directly. Premium cabins (Delta One, Premium Select) are modeled separately because they’re heavily sold on every route that flies them, so complimentary clears there are rare regardless of how “easy” the route otherwise looks.
Day of week and departure time set how business-heavy — and therefore how contested — the flight is. A Tuesday mid-morning and a Friday evening are different worlds. Days-to-departure also sets your clearance window by tier and cabin (those windows are exact Delta policy, not estimates).
If you can see open premium seats, enter them — it’s the single most informative input. The model discounts the count, because seat maps overstate true open inventory (unassigned and Basic Economy seats stay hidden until the gate), and discounts more the further out you are. It trusts the Delta app’s availability screen most and a pre-booking shopping map least.
Holidays crush supply — the cabin sells for cash and almost nothing clears free. A leisure season lever raises competition on vacation routes in-season. And traveling with someone changes the math: on one reservation you need two adjacent seats; on linked separate bookings you both clear only when the lower of the two statuses clears.
The headline % is the median probability — the honest middle estimate — with a plain-language read beneath it (from Next to Impossible up to Locked In) so the number lands at a glance.
Data confidence grades how much to trust that estimate from your inputs — how good your seat source is and how close to departure you are. It’s your signal for how settled the number is: Low means treat it as a rough read; High means it’s on firmer footing.
The limiting factor tells you the wall in plain language: sometimes it’s seat supply, sometimes it’s the crowd ahead of you, sometimes buying in is simply the realistic path.
The seat strip up top is a picture of the same thing: gold seats clear, the pin is you, and the gold reaches your pin exactly when the headline says you’ll clear.
It is a policy-constrained probabilistic prior. The rules — eligibility windows, Basic Economy exclusion, auto Comfort+ for Diamond and Platinum, day-of Delta One, same-reservation vs linked-booking priority — are exact. The clearance rates are reasoned estimates anchored to what Delta has disclosed publicly (notably its 2024 Investor Day figures), not coefficients fit to real upgrade-list outcomes, which Delta does not publish.
So treat it as a well-reasoned read, not a guarantee. It’s built to be honest about its own uncertainty rather than to sound confident.
Want the full mathematical write-up? The complete framework — every distribution, parameter, and design choice — is documented in the v4 framework note.