This workflow is for family physicians and residents who already have — or are about to buy — NEJM Knowledge+ (now delivered as AMBOSS Knowledge+) and want an implementation, not another product ranking. It addresses the ABFM one-day certification exam. The principal limitation it is built to counter: an adaptive feed optimises your near-term error rate, so left alone it can quietly starve low-volume but examinable domains while your overall percentage climbs.
What NEJM Knowledge+ offers for ABFM right now
| Item | Current state (vendor-reported, checked 19 July 2026) |
|---|---|
| ABFM coverage | Yes — Family Medicine Board Review aligned to the ABFM blueprint; one-day exam and FMCLA |
| Branding | NEJM Knowledge+ acquired by AMBOSS; now delivered as AMBOSS Knowledge+ — check which product you are buying |
| Question count | 2,000+ single-best-answer MCQs (vendor-reported) |
| Adaptive/AI features | Personalised study plans, custom filtered sessions, performance analytics, "AI Mode" copilot (vendor-reported); verify the current adaptive mechanics on the product page |
| Access period | 1-, 6- and 12-month Qbank options (vendor-reported) |
| Price | approx. $298 / $378 / $448 (1-/6-/12-month; vendor-reported, 19 July 2026 — verify) |
| CME | AMA PRA Category 1 Credits (vendor-reported) |
Figures are vendor-reported on 19 July 2026 and change; confirm them before paying.
The ABFM exam anchor
The target is 300 single-best-answer MCQs in four 95-minute sections of 75 questions — about six hours twenty of testing, roughly 100 minutes of pooled breaks, near eight hours at Prometric. A section, once completed, is locked. The 2025 five domains of care are Acute Care and Diagnosis (35%), Chronic Care Management (25%), Emergent and Urgent Care (20%), Preventive Care (15%) and Foundations of Care (5%). The last two — and the images, calculations and ethics scattered through the paper — are exactly what an error-minimising feed tends to under-surface, which is why this workflow builds explicit floors.
A note on pacing: why the section-lock changes your plan
The four-section structure with a locked, non-revisitable section is not a cosmetic detail; it should shape how you practise. Because you cannot return to a section once it closes, banking time by rushing an early section to "come back later" is impossible — every section must be paced to finish inside its own 95 minutes. Practise, therefore, in blocks that mirror a section: aim to complete 75 items in 95 minutes with a deliberate flag-and-move rule for anything that runs past about 90 seconds. NEJM Knowledge+ lets you set session length, so use that to build section-length blocks in the final weeks rather than only short tutor sets. The habit you are training is not raw speed but the discipline of not sinking four minutes into a single item you were always going to guess.
Baseline week: an unseen, blueprint-stratified sample first
Before you let the platform personalise anything, take a small, deliberately stratified unseen sample so you have an honest starting map. Draw roughly 75–100 items spread across the five domains in proportion to their weighting, sit them timed, mixed and without lookups, and record first-attempt accuracy per domain. This is your baseline. It is not a pass prediction; it is the reference the adaptive feed will otherwise overwrite within days. Resist the urge to look anything up while you sit it — the platform's explanations and AI copilot are one click away, but a baseline contaminated by lookups measures what you can find, not what you know. Sit it as if it were the exam, mark it honestly, keep the raw per-domain numbers somewhere outside the app, and only then read around the misses.
First pass: set domain floors
Now translate the blueprint into non-negotiable floors so the algorithm cannot hide an untouched domain behind a rising average. Set a minimum number of first-attempt items per domain for each week — for example, at least 30 Emergent and Urgent Care items and at least 15 Foundations of Care items weekly — and treat those as hard commitments regardless of what the feed recommends. If the personalised session will not deliver them, build a manual custom session that will. A practical way to enforce this inside an adaptive product is to alternate — one personalised session the engine chooses, then one manual session you choose — and never let two engine-chosen sessions run back to back without a floor check. That rhythm keeps the efficiency of adaptation while guaranteeing the low-volume domains get their weekly minimum regardless of what your recent error rate happens to be.
An error taxonomy worth keeping
"Got it wrong" is not a diagnosis. On every miss, tag one cause: knowledge gap (never knew it); misread stem (knew it, read too fast); premature closure (anchored on the first plausible answer); guideline error (out-of-date or wrong threshold); calculation error (arithmetic, units, dosing); or time-pressure error (right instinct, ran out of clock). The distribution of these tags tells you what to change. A wall of misread-stem tags is a pacing-and-attention problem, not a content problem, and no amount of extra questions fixes it. Keep the tally somewhere you will actually look at it — a single running count of the six categories — because the value is in the shape over time: a taxonomy that starts knowledge-gap-heavy and shifts toward misread-stem and time-pressure tags is the signature of a candidate whose content is maturing and whose bottleneck is now exam-craft, which is a cue to move toward timed mixed sections rather than more content.
Review interval: not everything deserves an immediate repeat
Match the response to the error type. A knowledge gap deserves a short source read plus a new, different transfer question a few days later — not an immediate repeat of the same item, which just trains recognition. A misread stem or premature closure deserves a process rule ("read the last line first"; "generate the differential before looking at options") and spaced retesting on fresh stems. A guideline error deserves the current source and a note. Reserve immediate repeats for almost nothing; space the misses instead.
Mixed-block switch: when to reduce topic filtering
Topic-filtered blocks are for building; mixed blocks are for proving. Shift the balance toward timed, random, mixed blocks when three things are true: every domain is at or above its floor, first-attempt accuracy has stopped rising on filtered sets, and your per-item time is near 76 seconds. Until then, keep filtering to attack weak domains; after then, filtering mostly re-teaches what you know, and the binding constraint becomes cross-domain switching and stamina. If two of the three criteria are met but pacing is still slow, do not delay the switch on content grounds — slow pacing is itself trained by timed mixed blocks, not by more filtered practice.
Exit criteria that are not "I finished the bank"
Completing 2,000 items is not a finish line. You are ready to taper when: coverage floors are met in all five domains; first-attempt accuracy is stable across two unseen, timed, mixed blocks; pacing holds near exam budget; retention is holding (spaced misses are staying corrected); and you have calibrated against official material — the ABFM blueprint and any official sample items. Bank completion alone satisfies none of these on its own.
Worked example: a seven-day plan
Give NEJM Knowledge+ / AMBOSS the content job and iatroX the unseen-measurement job — the two-Q-bank rule, with no claim to anyone's algorithm.
- Monday — NEJM Knowledge+: timed 40-item block on the weakest domain from baseline; tag every miss.
- Tuesday — NEJM Knowledge+: 40-item block on the second-weakest domain; short source reads on knowledge-gap tags only.
- Wednesday — iatroX: fresh, unseen, timed 50-item mixed ABFM block; no lookups; record first-attempt accuracy per domain.
- Thursday — NEJM Knowledge+: floor session — force Emergent and Urgent Care, images, calculations and Foundations of Care.
- Friday — NEJM Knowledge+: spaced retest of last week's misses on new items, not repeats.
- Saturday — iatroX Socratic Tutor: rework 8–10 misses until you can state the discriminating feature and the correct rule.
- Sunday — rest, or one timed mixed block; update your blueprint-coverage matrix and your error-tag tally.
Continue, supplement, switch or stop
Continue (learn) while filtered blocks still raise first-attempt accuracy and domains sit below floor. Supplement (retest on unseen) when the in-app average looks strong but no independent unseen block confirms it. Switch (simulate) to mostly mixed, timed, full-section mocks once floors are met and pacing is stable. Stop only when remaining items duplicate proven skills — measured on unseen blocks, not felt as boredom.
Three mistakes this workflow is designed to stop
First, letting the feed set your priorities so that low-volume domains never surface until the last fortnight. Second, immediate-repeating missed items until your repeat accuracy looks impressive and your unseen accuracy has not moved. Third, treating "I finished the bank" as readiness when you have never sat an unseen, timed, mixed section under exam pacing.
Frequently asked questions
Is NEJM Knowledge+ enough for ABFM on its own? It can be a sufficient content spine — a blueprint-aligned, adaptive, 2,000-plus item bank does most of the teaching. It is not sufficient as a readiness check, because its analytics run on a curated, partly repeated feed. Add an unseen, timed source to close that gap.
Which ABFM component does NEJM Knowledge+ not reproduce well? The full four-section, section-locked exam day. Adaptive tutor blocks do not train the stamina and no-going-back discipline of the real assessment, and the low-volume domains are precisely where an error-minimising feed is thinnest unless you force floors.
How many NEJM Knowledge+ questions should I complete per day for ABFM? There is no mandated figure. For most, 40–60 timed items reviewed properly on a working day is realistic, rising in the final weeks. Your access tier and price (vendor-reported, 19 July 2026) set the runway; plan daily volume against the window you bought rather than chasing a round number.
When should I stop using NEJM Knowledge+ and move to mixed mocks? When all five domains are at floor, first-attempt accuracy has plateaued on fresh items, and pacing is near 76 seconds. Filtered drilling then re-teaches the known; mixed, timed sections train the switching the exam demands.
How should I combine NEJM Knowledge+ with iatroX without duplicating practice? Separate the jobs: NEJM Knowledge+ for content and weak-area drilling, iatroX for unseen, timed measurement and Socratic rework. Never re-attempt a seen item in the other bank — measure only on fresh questions, so the signal stays clean.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026. Vendor-reported figures were accurate to the product pages on that date and change without notice — verify the current figure on the product page. Disclosure: iatroX operates a competing question bank; here it is confined to unseen readiness measurement and Socratic rework, jobs a single adaptive bank cannot perform on itself, and is not offered as a replacement for a primary content bank. Corrections via the feedback route on iatrox.com.
References: American Board of Family Medicine — One-Day Exam and 2025 Exam Blueprint (theabfm.org); AMBOSS / NEJM Knowledge+ Family Medicine Board Review (amboss.com/us/board-review/family-medicine); iatroX ABFM bank (https://www.iatrox.com/abfm-family-medicine); "Your Q-Bank Percentage Is Not Your Exam Score" (https://www.iatrox.com/blog/qbank-percentage-not-your-exam-score); and the blueprint-coverage matrix method (https://www.iatrox.com/blog/question-bank-completion-is-not-coverage-how-to-build-a-blueprint-coverage-matrix-for-any-medical-exam).