Grape confusion pairs are the study object that turns a grape atlas into a deduction coach — and this page publishes Sensium’s own ranked list, not a recycled “top ten tips” roundup. On 2026-07-09 we exported the top 25 pairs from the same deterministic confusion graph that powers Compare, Train suggestions, and every client’s “most confused with” chips. The ranking is reproducible; the methodology is below; the table is the citation unit.
Headline finding: when we rank mutual catalog confusions by combined global planting popularity, the spine is still Bordeaux and Rhône–Iberian reds — Cabernet Sauvignon vs Merlot sits at #1 — but the top 25 also surfaces high-planted neutrals (Airén, Trebbiano Toscano, Macabeo) that exam lists under-weight and real bottles do not.
This is catalog-graph research, not anonymized user telemetry. Population miss-rates need more volume before we publish them honestly; until then, we publish what only Sensium owns today: the editorial confusion graph at scale.
Methodology (read this before citing the table)
| Field | Value |
|---|---|
| Source | Sensium grape catalog (`grapes.json`) — `confusedWith` edges + per-pair separators |
| Ranker | Shared `ConfusionPairSuggester` / web `suggestComparePairs` (identical across iOS, Android, web, Mac) |
| Order | (1) mutual edges first (both grapes list each other); (2) lower combined `globalPopularityRank`; (3) stable id tie-break |
| Export date | 2026-07-09 |
| Catalog size | 1,534 dossiers |
| Re-run | `node scripts/data/export_confusion_graph_ranking.mjs --limit 25` from the Sensium repo |
What this ranking is not: a claim that these are the 25 pairs candidates miss most often in live exams, or a scrape of other websites’ listicles. It is the pairs our editorial graph treats as highest-signal and most globally planted — the same order the product surfaces when you open Compare with an empty selection.
When we have enough anonymized Train/Blind wrong-answer volume, Study 2 will publish a miss-rate table beside this graph. Until then, do not treat planting-weighted graph rank as population accuracy.
Why publish a graph rank at all?
Google’s 2026 guidance on generative AI search is blunt: commodity roundups lose to non-commodity pages — unique point of view, proprietary structure, methodology you can audit. A “25 pairs every taster should know” list without a named method is commodity. A ranked export from a living catalog, with a re-run command and a clear “not telemetry yet” label, is not.
Practically, the graph also answers a product question: if you only have five weeks before a tasting paper, which twins should you open first in Compare? Start at #1 and walk down. Pair that with the editorial hubs we already published — ten confusion pairs, year-in-review 25 pairs — which teach how to separate; this page tells you which edges the catalog itself elevates. For the same graph cut as hubs (top-15 planted grapes × four neighbors), see confusion neighborhoods 2026. For exam grapes this planting-mutual table under-weights, see one-way exam hubs 2026.
The top 25 (catalog graph, 2026-07-09)
Separators are the catalog’s first editorial cue for that edge (trimmed for the table). Open the Compare path for the full fingerprint.
| Rank | Pair | Combined planting rank | Catalog separator (first cue) | Drill |
|---|---|---|---|---|
| 1 | Cabernet Sauvignon vs Merlot | 3 | Cassis + firmer tannin + cedar length vs plum-led fruit + softer tannin + rounder mid-palate | Compare |
| 2 | Cabernet Sauvignon vs Syrah | 7 | Cassis/cedar frame vs black pepper, olive, smoky savoriness | Compare |
| 3 | Syrah vs Tempranillo | 9 | Pepper/olive/smoke vs dill-leather oak + savory red-black fruit | Compare |
| 4 | Grenache vs Tempranillo | 10 | Red-fruited warmth + lower tannin vs dill-leather oak frame | Compare |
| 5 | Chardonnay vs Sauvignon Blanc | 13 | Textural breadth + oak compatibility vs pungent green-citrus + sharp acid | Compare |
| 6 | Grenache vs Syrah | 13 | Red-fruited warmth + spice vs pepper/olive/smoke | Compare · deep dive: Syrah vs Grenache |
| 7 | Merlot vs Sangiovese | 13 | Plum + soft tannin vs sour-cherry + savory herbs + bright acid | Compare |
| 8 | Airén vs Trebbiano Toscano | 14 | Light orchard fruit + moderate acid vs neutral citrus-apple + high acid | Compare |
| 9 | Sangiovese vs Tempranillo | 14 | Sour-cherry acid spine vs dill-leather oak + savory fruit | Compare · Temp vs Sangio |
| 10 | Bobal vs Tempranillo | 15 | Deep color + firm tannin without heavy oak sweetness vs dill-leather oak | Compare |
| 11 | Chardonnay vs Trebbiano Toscano | 15 | Oak-compatible breadth vs neutral high-acid citrus-apple | Compare |
| 12 | Airén vs Macabeo | 18 | Light orchard restraint vs apple-fennel + medium texture | Compare |
| 13 | Grenache vs Sangiovese | 18 | Warm red fruit + spice vs sour-cherry acid spine | Compare |
| 14 | Cabernet Sauvignon vs Malbec | 19 | Cassis + pyramidal grip vs inky blackberry-plum + broad texture | Compare · Malbec vs Cabernet |
| 15 | Chardonnay vs Macabeo | 19 | Oak-compatible breadth vs apple-fennel medium texture | Compare |
| 16 | Carignan vs Grenache | 20 | High acid + rustic tannin + herbal dark fruit vs red-fruited warmth | Compare |
| 17 | Chardonnay vs Pinot Gris | 20 | Oak-compatible breadth vs pear-almond moderate aromatics | Compare |
| 18 | Malbec vs Merlot | 20 | Inky blackberry-plum breadth vs plum-led softer tannin | Compare |
| 19 | Pinot Noir vs Sangiovese | 20 | Pale + low tannin + red-fruit earth vs sour-cherry + bright acid | Compare |
| 20 | Mourvèdre vs Syrah | 23 | Gamey savory depth + stronger tannin vs pepper/olive/smoke | Compare |
| 21 | Macabeo vs Trebbiano Toscano | 24 | Apple-fennel medium texture vs neutral high-acid citrus-apple | Compare |
| 22 | Pinot Gris vs Pinot Noir | 24 | Pear/white-peach white frame vs red cherry/cranberry low-tannin red | Compare · Pinot Gris vs Grigio |
| 23 | Riesling vs Sauvignon Blanc | 24 | Piercing acid + lime vs pungent green-citrus | Compare |
| 24 | Bobal vs Carignan | 25 | Deep color + firm tannin vs high acid + rustic herbal dark fruit | Compare |
| 25 | Chenin Blanc vs Sauvignon Blanc | 28 | Waxy orchard fruit + firm acid vs pungent green-citrus | Compare · Chenin vs Chardonnay |
Three takeaways from the graph
1. Mutual + planted still equals “exam spine”
Ranks 1–7 are the pairs tasting papers already obsess over: Cabernet/Merlot, Cabernet/Syrah, Grenache/Syrah, Chardonnay/Sauvignon Blanc, Merlot/Sangiovese, Tempranillo twins. That is a sanity check — if the graph had put obscure hybrids first, we would distrust the export. Deep dives already exist for several of these edges; use this table as the order, not a replacement for the separator essays.
2. High-planted neutrals are a real blind trap
Airén, Trebbiano Toscano, and Macabeo appear repeatedly in the top 25 because the catalog correctly marks them as mutual look-alikes and they sit high on global planting rank. Candidates who only drill “noble” varieties miss the quiet white that shows up in blends, bulk, and regional flights. If your syllabus is Top 10 only, still open #8, #11, #12, #15, and #21 once — the separator is often acid line and aromatic restraint, not a flashy aroma tell.
3. Color-family collisions need an explicit frame
Pinot Gris vs Pinot Noir (#22) is in the graph because the catalog records the DNA/family confusion — but the first cue is frame (white vs red), not a mid-palate nuance. Treat it as a “don’t be clever” pair: if the wine is red, stop chasing Gris. The graph includes it so Compare does not hide the edge; your drill should still start with color.
How to drill the list in four weeks
- Week 1 — ranks 1–6. Bordeaux + Rhône reds + Chardonnay/SB. Two bottles or Train confusion focus + Compare after every miss.
- Week 2 — ranks 7–14. Iberian/Italian reds + Cabernet/Malbec + first neutrals. Keep Tempranillo vs Sangiovese open.
- Week 3 — ranks 15–20. More neutrals, Carignan/Grenache, Pinot/Sangiovese, Mourvèdre/Syrah.
- Week 4 — ranks 21–25 + revisit misses. Riesling/SB, Chenin/SB, Pinot Gris/Noir frame check. Log every wrong call in Train so Mistake Replay owns next month.
Companion method posts: structure-first tasting, daily blind routine, flashcards vs drills.
What we will publish next (Study 2)
When anonymized Train/Blind wrong-answer volume clears a documented n threshold, we will publish a miss-rate ranking beside this graph — same pairs, different question (“what do tasters actually confuse?”). Until that ship date, cite this page as catalog-graph rank, not population accuracy. The export script above is the lock so the table cannot silently drift from the product.
Frequently asked questions
Is this the same as your “25 pairs year in review” post?
No. The year-in-review hub is an editorial checklist with deep-dive links. This page is a ranked export of the live confusion graph with methodology, combined planting ranks, and Compare URLs for every row.
Can I cite these numbers in a newsletter or class?
Yes — please attribute “Sensium confusion graph, 2026-07-09” and link this URL. Do not restate the ranks as “exam miss rates” unless you are citing a future Study 2 that says so explicitly.
Why isn’t Nebbiolo vs Pinot Noir in the top 25?
Mutual + planting-weighted rank elevates globally planted edges first. Nebbiolo/Pinot remains a critical exam pair and has its own deep dive — Nebbiolo vs Pinot Noir — but it sits further down the planting-weighted mutual list. Use this table for graph priority; use one-way exam hubs 2026 for syllabus priority when the edge is mostly one-way.
How do I reproduce the table?
From a Sensium checkout: `node scripts/data/export_confusion_graph_ranking.mjs --limit 25`. The JSON includes ids, Compare paths, and the first catalog separator for each edge.
Open the #1 pair in Compare, then walk the table. The graph is only useful if it becomes reps.