Nearly 40% of restaurant specials fail to meet sales expectations, creating costly waste. Most establishments jump straight into bulk ordering for new dishes, but poor sales leave you with spoiled inventory. A calculated testing approach lets you validate demand first, then scale smartly.
Why testing small is smart
A special can be a goldmine or a cost center. If you order ingredients for 100 portions upfront and only 20 sell, you lose money on two fronts:
- Wasted ingredients
- Missed opportunity for other dishes
- Storage of perishable goods
⚠️ Watch out:
Most restaurants overestimate demand. Always start with a small test to avoid disappointments.
The 20-portion test strategy
Start with ingredients for a maximum of 20 portions. This gives you enough data to see how guests react, without major financial risk.
💡 Example:
You want to test a salmon special for €28.50 per portion:
- Salmon for 20 portions: €120
- Vegetables and garnish: €40
- Total test investment: €160
At 15 portions sold: €427.50 revenue - €160 cost of goods = €267.50 gross profit
Measure the right data during your test
During your test period of 3-5 days, collect important information. One of the most common blind spots in kitchen management is failing to track actual prep time versus estimates.
- Sales figures: How many portions per day?
- Guest feedback: Compliments or complaints?
- Kitchen logistics: How long does preparation take?
- Actual food cost: Does your calculation match?
💡 Example test results:
Salmon special over 4 days:
- Day 1: 6 portions sold
- Day 2: 4 portions sold
- Day 3: 5 portions sold
- Day 4: 3 portions sold
Average: 4.5 portions per day = 31 portions per week
From test results to purchasing strategy
Based on your test results, determine your purchasing strategy for the actual special:
- Strong result (75%+ sold): Scale up to desired period
- Moderate result (50-75%): Adjust recipe or price, test again
- Weak result (<50%): Stop or change concept drastically
💡 Example scaling:
Test shows 31 portions per week. For 3-week special:
- Expected sales: 93 portions
- Order with 10% buffer: 102 portions
- Order salmon: 6.8 kg (102 × 67g per portion)
Much lower risk than ordering 150 portions upfront
Timing your test
Test your special on representative days. Not during extremely busy or quiet periods, as that gives a skewed picture.
- Ideal: Normal weekdays + 1 weekend
- Avoid: Holidays, school vacations, extreme weather
- Duration: Minimum 3 days, maximum 1 week
⚠️ Watch out:
Don't test during Valentine's Day or Mother's Day - those days aren't representative of normal sales.
Adjust cost price after test
During your test you often discover that your cost price differs from your calculation:
- Trimming loss higher or lower than expected
- Portion sizes different than planned
- Extra garnish or sauce needed
Recalculate your actual food cost after the test and adjust your selling price for the actual special.
How do you test a new dish step by step?
Calculate cost price and order for 20 portions
Calculate what all ingredients cost per portion. Order ingredients for a maximum of 20 portions to limit your risk. Include garnish, sauces and oil in your cost price.
Test for 3-5 days on normal days
Put the dish on your menu for at least 3 days. Choose normal weekdays, no holidays or extreme busy periods. Keep track of how many portions you sell per day.
Measure results and scale up
Calculate your average sales per day. If you've sold more than 75%, you can scale up. With less than 50% sales, adjust the recipe or stop.
✨ Pro tip
Run your test for exactly 4 days including one weekend day. This captures both weekday and weekend demand patterns, giving you the most accurate scaling data for your 3-week special.
Calculate this yourself?
In the KitchenNmbrs app you can do this in just a few clicks. 7 days free, no credit card.
Was this article helpful?
Frequently asked questions
How many portions should I test at minimum?
Test a minimum of 15-20 portions over 3-4 days. This gives you enough data without major financial risk. Less than 15 portions provides too little information for reliable scaling decisions.
What if my test special doesn't sell well?
Stop after your test period and analyze why. Maybe the price is too high, the taste isn't right, or it doesn't fit your concept. Better to lose a small test investment than large inventory costs.
Should I tell my team it's a test?
Yes, tell your kitchen team it's a test. They can provide valuable feedback on preparation, timing and guest reactions. Their input helps you decide whether to scale up and often reveals prep issues you'd miss otherwise.
📚 Sources consulted
- EU Verordening 852/2004 — Levensmiddelenhygiëne (2004) — Official source
- EU Verordening 853/2004 — Hygiënevoorschriften voor levensmiddelen van dierlijke oorsprong (2004) — Official source
- EU Verordening 1169/2011 — Voedselinformatie aan consumenten (2011) — Official source
- NVWA — Hygiënecode voor de horeca (2024) — Official source
- NVWA — Allergenen in voedsel (2024) — Official source
- Codex Alimentarius — International Food Standards (2024) — Official source
- FSA — Safer food, better business (HACCP) (2024) — Official source
- BVL — Lebensmittelhygiene (HACCP) (2024) — Official source
- Warenwetbesluit Bereiding en behandeling van levensmiddelen (2024) — Official source
- WHO — Foodborne diseases estimates (2024) — Official source
Food Standards Agency (FSA) — https://www.food.gov.uk
The HACCP standards shown in this application are for informational purposes only. KitchenNmbrs does not guarantee that displayed values are current or complete. Always consult the FSA or your local authority for the latest regulations.
Written by
Jeffrey Smit
Founder & CEO of KitchenNmbrs
Jeffrey Smit built KitchenNmbrs from 8 years of hands-on experience as kitchen manager at 1NUL8 Group in Rotterdam. His mission: give every restaurant owner control over food cost.
Purchase smarter with real-time insights
Seasonal prices fluctuate — so do your recipe costs. KitchenNmbrs automatically recalculates your margins when purchase prices change. Never get surprised again. Start free.
Start free trial →