We analyzed 50 steakhouse by-the-glass programs. Here's what most get wrong.
Sommetry sampled fifty US steakhouse BTG lists from publicly posted menus and scored list breadth, category balance, and bottle-to-glass price logic. Most programs over-list slow slots and under-specify steak-ready reds - patterns you can audit on your own list without guest data.
Why we ran this
Steakhouses live on beverage margin, but BTG is often designed like a mini bottle list - too many slots, weak food bridges, and glass prices that don’t match bottle math. We wanted a baseline of what public menus actually show, not what consultants wish they showed.
This is not a consumer ranking. We don’t name venues in this draft, and we don’t claim sell-through or pour-cost outcomes we can’t see from menus alone.
Methodology
- Sample: 50 independent and small-chain US steakhouses with posted wine/BTG menus online (PDF or HTML) as of July 2026.
- Fields captured: BTG count, varietal/category tags, published glass price, matching bottle price when listed, pour size when disclosed.
- Scoring (menu-only): breadth bucket (≤8, 9–12, 13+ slots), red/white/bubbles mix, implied bottle÷glass ratio, presence of a clear “steak red” anchor (cab, malbec, syrah, or blend marketed as such).
- Planned enrichment: Sommetry gap-analysis tooling will add pairing coverage vs. standard steakhouse menus, duplicate-SKU detection, and estimated pour-cost bands when wholesale hints exist.
Honest limitations
- Public menus may lag what’s actually on the floor tonight.
- We cannot see depletions, spoilage, or server attach rates from menus.
- Glass pour sizes are often unstated - we assume 5–6 oz when not listed.
- Chains with identical BTG across locations are counted once in the sample.
- Findings below marked PLACEHOLDER require the automated gap-analysis pass - do not cite them as final until populated.
Findings snapshot
btg_breadth_histogram.csv.
Pattern 1: Too many BTG slots
Operator theory caps BTG at 6–8 movers; many public lists show 12–16 lines. Extra slots without preservation infrastructure imply spoilage risk - we can’t measure waste from menus, but we can count slots.
Pattern 2: Weak steak-red anchoring
A BTG program should make the ribeye pairing obvious in two words. Lists that bury cab/malbec under five “interesting” reds force servers to improvise.
steak_anchor_score.
Pattern 3: Bottle–glass arbitrage
When bottle price ÷ glass price < 3.5, guests rationally buy bottles and BTG becomes a decoy. When ratio > 5.5, BTG feels expensive vs. bottle and attach suffers. Healthy band for four-pour math is roughly 4.0–5.0.
| Glass price | Bottle list | Ratio | Signal |
|---|---|---|---|
| $18 | $72 | 4.0 | Balanced |
| $22 | $88 | 4.0 | Balanced |
| $16 | $96 | 6.0 | BTG feels steep |
| $28 | $84 | 3.0 | Bottle arbitrage |
btg_bottle_ratio_outliers.json.
Pattern 4: Category gaps vs. steakhouse menus
Standard steakhouse menus still sell shellfish towers, creamed spinach, and shared seafood platters. BTG lists heavy on cab alone miss easy white/rosé attach unless those slots exist.
menu_pairing_coverage_pct per list.
What operators can fix without waiting on our data
- Count BTG slots; rotate anything you wouldn’t finish a bottle on in 36 hours.
- Put one obvious steak red in the first two BTG lines servers read.
- Check bottle÷glass on your top three movers - adjust glass or bottle before raising list prices.
- Match BTG pushes to pre-shift scripts (see server training note).
What this means for your program
Your public BTG menu is a margin document whether you treat it like one or not. Run the four checks above on your list this week. When our gap-analysis batch completes, we’ll replace placeholder blocks with distributional stats - still menu-derived, still no invented sell-through claims. If you want the same scoring on your private list with pour sizes and costs, send the PDF through the analysis form below.