Why Merito Never Shows an Approval Probability

It would be easy to build a version of Merito that ends every tool result in a single number: "72% likely to be approved." It would probably even be popular — a number like that is simple, shareable, and satisfies the very real anxiety of not knowing where you stand. It's also not something Merito will ever show, on any tool, under any framing, and that's not a marketing position adopted after the fact. It's a hard rule written into the product's own engineering constraints, and it shapes real, specific design decisions all the way down to how a single chart on a public profile page renders its bars.

Understanding why requires separating two different things that often get blurred together: the legal reason (USCIS outcomes genuinely aren't predictable from public information in any way that would make a percentage honest) and the product reason (a readiness review and a probability estimate are different kinds of claims, and conflating them would misrepresent what any of these tools actually do). Both matter, and both show up in concrete places in how the product is actually built — not just in a disclaimer at the bottom of a page.

The legal reason: a percentage would be a claim nobody can actually back

USCIS adjudication of EB-1A, O-1A, and EB-2 NIW petitions runs through a genuinely qualitative standard: a "preponderance of the evidence" review of the whole record, applying published criteria (Volume 6, Part F, Chapter 2) and, for EB-1A, a separate final-merits determination of the totality of the record. Nothing in that process resembles a scoring formula that outside software could reverse-engineer into a reliable percentage. A tool that outputs "72%" isn't summarizing a real probability distribution it computed — it's manufacturing a number that looks precise while resting on no actual predictive model of individual adjudicator judgment. That's true of any extraordinary-ability-style petition; it's also true, for a useful point of comparison, of the UK's structurally similar Global Talent visa, where the official Home Office guidance describes field-specific evidence thresholds and an endorsing body's judgment call, not a scored formula either — case-by-case qualitative review, not an algorithm, is how comparable "exceptional talent" immigration routes actually work on both sides of the Atlantic.

The product reason: readiness and probability are different claims

Separately from the legal risk, a probability score would misdescribe what Merito's tools are actually doing. Merito Score reads a self-reported questionnaire against the published criteria and renders two separate rings — an outer ring for how many individual criteria your evidence plausibly reaches, and an inner ring for whether your record as a whole reads as sustained acclaim. Both are readiness reads: how does this record currently compare to a published standard. Neither is, or claims to be, a forecast of what a specific USCIS officer will decide on a specific day. /help/merito-score states this distinction directly on the results page itself: "a readiness snapshot from Merito — not a law firm, not a probability estimate."

The same discipline runs through Officer Simulator, a tool that, on the surface, comes closest to sounding like a prediction engine — it literally simulates how an officer might read your file. Its per-criterion posture uses three qualitative labels (Likely met, Needs more, Likely not met), and its disposition panel tracks three separate simulated axes — confidence, openness, concern — shown only as named bands (Unsettled through Firm, Guarded through Receptive, Low through High), explicitly labeled "a simulated reading disposition — not a prediction." Even the Decision Preview feature, which generates a simulated approval, RFE, or denial discussion, carries a "SIMULATION — EDUCATIONAL DRAFT" watermark on every output specifically so it can't be mistaken for a real forecast or a real document — a real decision, if a case is ultimately denied and appealed, goes through USCIS's actual Administrative Appeals Office review instead, on a timeline and by a process no simulation can substitute for.

The commitment is load-bearing, not just a phrase

Merito documents this as one of a short list of explicit product commitments, alongside data ownership and citation integrity, on its own commitments page: "Merito is not a law firm and does not predict outcomes. Every tool is framed as a readiness review... never a probability, score-to-odds conversion, or approval likelihood." That page also states plainly that the deterministic audit layer discussed elsewhere runs "in plain code... before any AI touches your file" — the same "code before AI" discipline that underlies why a readiness score can be computed transparently from published criteria, while a genuine probability estimate would require modeling something (an individual officer's discretionary judgment) that isn't observable, auditable, or reducible to code at all.

Where the commitment shows up in the code, not just the copy

The clearest evidence that this is a real engineering constraint rather than a phrase in a style guide sits in the smallest, least visible corner of the product: the public profile page. A Merito user can choose to share a public profile at /u/{slug}, and can opt to include a score-trend visualization alongside it — a small sparkline showing whether their readiness has been trending up over time. The implementation is explicit about what it will and won't expose: the serializer that builds this public view computes "a 1-5 relative bucket per data point," with the code comment stating outright that this is "'shape only,' never the underlying score" (apps/web/src/lib/profile/publicProfileSerializer.ts). Each bar's height on the public page is that bucket value, not the real number — so a viewer can see a general upward or downward trend without ever learning the specific score behind it, even when the profile's owner has actively opted into sharing.

This is worth contrasting directly with the private trend chart a user sees on their own dashboard, which does show a rounded, specific number in its tooltip and axis labels when the user is looking at their own data (apps/web/src/components/dashboard/CasePulse.tsx) — because showing someone their own readiness score, on their own private dashboard, isn't the same claim as broadcasting a comparable, precise figure publicly. The distinction is deliberate: your own number, in your own private workspace, is useful information about your own case. The same number, stripped to shape-only bars the moment it's shared publicly, avoids turning individual scores into something that reads — to a stranger scrolling past a public profile — like a comparable, quasi-official rating. Neither version is a probability of any USCIS outcome; the public version goes one step further and refuses to even let the number itself travel outside the account it belongs to.

What this means in practice, for you

None of this is a limitation to work around — treating a Merito readiness score as if it secretly encodes a probability, or trying to reverse-engineer an implied percentage from a Strong/Moderate/Weak badge, would be reading more precision into the tool than it's designed to provide or than the underlying legal process would support. The more useful frame: a criterion marked Weak, an officer persona reading a section as Vulnerable, or a downward-trending score, are all telling you the same kind of thing — here is a specific, current gap between your record and a published standard, worth addressing with real evidence — not a countdown to a verdict. Case Canvas reinforces this same framing on the evidence side: strength badges are described explicitly as "a plain-language reading of how well-supported a criterion is, not a probability of approval."

If you're deciding between visa categories or trying to understand why "approved" itself doesn't fully resolve a case's practical status, it's worth reading beyond the approval decision itself — a favorable I-140 doesn't guarantee an immediate green card, since priority-date backlogs and visa-bulletin retrogression are a real, separate mechanism entirely outside any petition's own merits; the priority dates and visa bulletin guide and the Final Merits Determination guide both cover ground directly relevant to why a single number could never honestly capture "where you stand." eb1mentor.com's own writing makes a closely related point from a different angle in Priority Dates and Retrogression: What an Approved Petition Does Not Guarantee — worth reading as a reminder that "approval" itself isn't the single, clean endpoint a percentage-style tool would imply anyway. Petitioners discussing how they've each interpreted a readiness result without over-reading it are a regular presence in Merito's community, and the pricing page covers what the Pro toolset adds once your own readiness picture from the free Score assessment is in hand.

FAQ

If Merito won't give a probability, how do I know if my case is "good enough"?

Merito is deliberately built to answer a different, more honest question: how does your record currently compare to the published criteria and to the totality-of-the-record standard, and specifically where are the gaps. That's information you can act on directly. Whether it's "good enough" for your specific case is a legal judgment outside what any automated tool — this one included — can honestly answer, which is exactly why Merito doesn't attempt to answer it with a number.

Does the public profile page hide my score entirely, or just the exact number?

Just the exact number, and only on the public version. The code computes a relative 1-to-5 bucket per data point for public display — enough to show whether a trend is generally moving up or down — while your own private dashboard shows your actual, specific score to you.

Are the Strong/Moderate/Weak and Likely-met/Needs-more/Likely-not-met labels just percentages in disguise?

No — they're independently computed qualitative reads (rubric-based for Merito Score, AI-generated reasoning tied to cited evidence for Officer Simulator), not a percentage translated into words. There's no hidden number behind a badge that a user could back out by decoding the label.

Does this "no probability" rule apply to every tool, including Pro-tier ones?

Yes, without exception — it's listed as one of Merito's core product commitments, not a free-tier-only restriction. RFE Studio's drafts, Officer Simulator's decision previews, and every other AI-generated result carry the same standard disclaimer and the same absence of any probability or likelihood language.

This article explains a specific product design commitment and the reasoning behind it — it is not legal advice, and it does not evaluate or comment on the merits of any individual case. Merito is not a law firm and does not predict USCIS outcomes.