Benchmarking a Product You Admire: What R&D Can Match
By the Silk Foods Ceylon Editorial Team
Buyer’s snapshot
- A benchmark product gives R&D three things to work back from: taste, texture and the label. It does not give the recipe, because a label shows ingredient order, not quantities.
- Silk Foods Ceylon (SFC) runs R&D in 3 to 4 weeks at most from signed brief to approved formulation, typically in 2 to 4 sample iterations.
- Taste, texture and ingredient order can be matched closely. Exact percentages, a competitor’s process and its shelf-life data cannot be read from a jar.
- For a brand owner who can hand over the physical product and name one approver. Not for anyone hoping to copy a protected recipe.
“Make it like that one, but a bit less sweet.” It is the most common first sentence of an R&D brief, and a good one, because it hands the R&D team a physical target. It is also a sentence that hides three different jobs: matching a taste, matching a texture and reproducing a label.
Some of that can be done well. Some of it can only be approximated, and the difference decides whether a brief lands in two sample rounds or five. This post sets out what a benchmark product tells R&D, what it cannot tell, and how to write the brief so the 3 to 4 week window holds.
What does a benchmark product actually tell R&D?
A benchmark tells R&D what the finished product tastes, feels and declares, not how it was made. The team can measure and score the first two and read the third off the pack. Everything about quantities, sourcing and process is inference.
That inference is useful. A team that has made several coconut spreads can taste a benchmark and say, with reasonable confidence, that it is sweeter than the target, set with a particular body, and cooked longer than a fresh-tasting version would be. What it cannot do is name the percentage of coconut, the supplier of the sugar or the batch temperature. Those are the maker’s, and a brief should not pretend otherwise.
Think of the benchmark as a ruler, not a blueprint. It tells you how far a sample is from the target. It does not tell you how to get there.
Taste, texture, label: three matches, three levels of precision
Each of the three attributes can be matched to a different degree, and the brief works better when it says which degree it wants.
| Attribute | What R&D can read from the benchmark | How closely it can be matched | What stays approximate |
|---|---|---|---|
| Taste | Sweetness, salt, acidity, heat, bitterness, aroma, finish | Close, scored against agreed notes | Exact balance across batches and seasons |
| Texture | Spreadability, set, viscosity, mouthfeel, separation | Close, checked by spoon, fridge and time | Behaviour at the end of the benchmark’s shelf life |
| Label | Ingredient order, declared nutrition, claims, pack size | Close on order and pack | Exact percentages, which the label does not give |
| Process | Almost nothing directly | Inferred from texture and flavour | Cook time, temperature, equipment, order of addition |
| Shelf life | The printed date only | Tested on the new product, not assumed | The benchmark’s own stability data |
The pattern is that the further an attribute sits from the consumer’s spoon, the less a jar reveals. That is why the brief should spend its words on the top three rows.
How does R&D work back from the label?
R&D reads the ingredient list as a ranking. Under the Codex General Standard for the Labelling of Prepackaged Foods, ingredients are listed in descending order of ingoing weight at the time of manufacture (Codex Alimentarius, CXS 1-1985, 2024 amendment). So the first ingredient on a benchmark weighed most, and the last weighed least.
That ranking narrows the formulation without fixing it. If sugar is listed ahead of coconut, the formulation starts from sugar-heavy territory. If a stabiliser appears last, the team knows it is a small addition and can test levels. But the standard also lets a compound ingredient with less than 5% share of the food omit its own sub-ingredients, with certain allergen exceptions (Codex Alimentarius, CXS 1-1985), so some detail never reaches the label at all.
Sri Lanka’s current rules sit on the same principle. The Food (Labelling and Advertising) Regulations 2026 were published in Gazette Extraordinary No. 2494/47 on 26 June 2026 and have been in operation since 1 July 2026 (Government of Sri Lanka, 2026). The new product’s label has to meet them in its own right. Copying a benchmark’s label wording is not a safe route, and a benchmark made before the new rules may not comply with them. How the 2026 labelling regulations affect a private-label launch covers what changed.
A practical reading. The label gives R&D an order of ingredients and a nutrition panel to aim near. The new label is then built from the new formulation, not copied from the old one.
What goes in the brief, and what stays approximate?
The brief should state the benchmark, the direction of difference, the scoring and the limits. It works because each line can be checked on a bench.
- The physical benchmark. Courier two or three units from the same lot, with the date code visible to R&D. One jar is an anecdote. Three from a single lot show how much the benchmark itself varies.
- The direction of difference. “Match it, but 20% less sweet” is workable. “Make it better” is not. Name each attribute that should differ and by roughly how much.
- The scoring sheet. Agree it before the first sample: which attributes, what scale, and what counts as a pass. The R&D brief guide lists the target fields, and sensory panel screening explains how to check a panel before you rely on its verdicts.
- The cost ceiling. A benchmark may use an ingredient the new product cannot afford. R&D needs the ceiling in LKR per unit to decide what to trade.
- What you do not want matched. If the benchmark has a long ingredient list the new product should shorten, say so. A clean-label reformulation is a different brief with different trade-offs.
- The line you will not cross. A brand owner should say up front that the aim is a product that competes on taste and texture, not a copy of a protected recipe or pack.
Service snapshot: R&D / NPD (Co-Development on the SFC brochure) at Silk Foods Ceylon
Service: in-house formulation against a physical benchmark, from a signed brief at the Matale facility
Window: 3 to 4 weeks maximum, signed brief to approved formulation
Iterations: typically 2 to 4 samples, with a verdict requested within 3 working days of each
After approval: 2 to 3 weeks from purchase order to dispatch on a locked recipe, with small runs welcome and quoted on request
Audit coverage: BRCGS- and FSSC 22000 V6-audited, with SLSI submission support inside the engagement
How do you score a match without arguing about it?
Scoring a match works when the question is fixed before the tasting. A triangle test, standardised in ISO 4120:2021, gives each assessor three samples, two identical and one different, and asks them to pick the odd one out (International Organization for Standardization, 2021). It is a forced-choice method for checking whether two products can be told apart when the nature of any difference is unknown.
For a benchmark brief, that makes it a useful final gate, not a first-round tool. In early rounds, a scored sheet with named attributes tells R&D which way to move. Once a sample is close, a triangle test asks the harder question: can a panel tell it from the benchmark at all? The standard notes limits, including products with strong carryover or lingering flavours, so a hot chilli sauce is a poor fit for the method while a plain spread suits it well. Agree in the brief which question each round answers, and who counts as the panel.
Match does not have to mean identical. Many brand owners want a product that is no worse than the benchmark on the attributes buyers care about and clearly better on one. Write that as a pass mark: “no more than one notch sweeter, spreads straight from the fridge, no separation at day 10.” The 3 to 4 week R&D clock explains why pass marks written before sample one are what let the clock stop.
A worked example: a benchmark spread, two brief styles
Take a composite case, not a named client: a brand owner wants a 300 g coconut spread benchmarked against a leading jar they admire. The figures are illustrative.
In the first brief, the owner sends a photo and the line “like this one, less sweet”. R&D buys the jar, tastes it and builds a first sample. The verdict comes back as “too thin and not quite the same”. With no named attributes, the second sample changes sweetness, body and colour at once, and misses on two. Two rounds are spent and the team has learned little.
In the second brief, the owner couriers three jars from one lot, with the scoring sheet attached: sweetness one notch lower, a body that holds a spoon upright for 3 seconds, colour within the benchmark’s range, no separation at day 10, cost under a stated ceiling per jar. The first sample is scored line by line, the second changes two named variables, and a triangle test on the third closes the question.
The recipe is the same in both cases. What differs is how much of each round is spent finding out what “right” means. Why one R&D trial rarely locks a recipe explains why the iterations still matter even with a good benchmark.
What stays approximate even with a good benchmark?
Four things remain approximate however well the brief is written: exact percentages, the benchmark’s process, its shelf life and its supply.
- Exact percentages. The label gives order, not weights, and inference gets close without ever being certain.
- Process. Cook time, temperature and order of addition change texture, and R&D can only infer them from the result.
- Shelf life. The date on the pack is a claim. The new product needs its own testing before a date goes on its label.
- Supply. The benchmark may use an ingredient grade or origin that is not available at the same price. R&D substitutes, and the sample shows what that costs in taste.
Where SFC walks away. A benchmark brief suits a brand owner who has validated the concept and wants to compete on the same shelf. For a brief that asks for an exact copy of a protected recipe or pack, or for a team that cannot name who approves samples, the better first step is a decision on what the product stands for, before any sample is made.
Frequently asked questions
Can Silk Foods Ceylon develop a product to match a benchmark I admire?
Yes. The in-house R&D team works from a physical benchmark and a signed brief, and takes it to an approved formulation in 3 to 4 weeks at most, typically in 2 to 4 sample iterations. It matches taste, texture and ingredient order closely, and builds the new label from the new formulation.
Can R&D work out a benchmark’s exact recipe from its label?
No. Ingredients are listed in descending order of ingoing weight (Codex Alimentarius, CXS 1-1985, 2024 amendment), so the label shows order but not percentages, and a compound ingredient under 5% of the food may omit its own sub-ingredients. R&D infers a close formulation, not the original.
Can I copy a benchmark product’s label wording?
Not safely. The new label has to meet the Food (Labelling and Advertising) Regulations 2026 in its own right (Government of Sri Lanka, Gazette 2494/47, 2026), and a benchmark made earlier may not comply. Build the label from the new formulation and check it against the regulations before artwork is approved.
How many benchmark units should I send to R&D?
Two or three from the same lot, with the date code visible. One unit cannot show how much the benchmark itself varies between jars. A single lot keeps the comparison fair, and the date code lets the team judge how fresh the benchmark was when tasted.
How Silk Foods Ceylon can help
For a brand owner with a product they admire and no formulation of their own, Silk Foods Ceylon runs R&D / NPD and Contract Manufacturing on one site at Silk AgTech Park in Hapugasyaya, Matale. R&D takes a signed brief and a physical benchmark to an approved formulation in 3 to 4 weeks at most, and the first commercial run follows 2 to 3 weeks after the purchase order. Small runs welcome, quote on request. The facility is BRCGS- and FSSC 22000 V6-audited, with SLSI submission support inside the engagement. To book a start date, send a signed brief to b2b@esilkroute.com.lk or call +94 77 157 7294.
Sources
Codex Alimentarius Commission, General Standard for the Labelling of Prepackaged Foods (CXS 1-1985, as amended to 2024), clause 4.2.1 on the list of ingredients in descending order of ingoing weight and the 5% rule for compound ingredients. fao.org. Retrieved 15 November 2026.
Government of Sri Lanka, Food (Labelling and Advertising) Regulations 2026, Gazette Extraordinary No. 2494/47 of 26 June 2026, in operation from 1 July 2026. documents.gov.lk. Retrieved 15 November 2026.
International Organization for Standardization, ISO 4120:2021, Sensory analysis, Methodology, Triangle test, on the forced-choice design and its limits with strong carryover products. iso.org. Retrieved 15 November 2026.
Silk Foods Ceylon, internal capability data for the Matale facility: R&D 3 to 4 weeks maximum from signed brief to approved formulation; typically 2 to 4 sample iterations; a buyer verdict requested within 3 working days of each sample; small runs quoted on request; 2 to 3 weeks from purchase order to dispatch on a locked recipe. The worked example, the pass marks and the table’s precision levels are illustrative planning examples, not SFC recipe or schedule data.
Written by the Silk Foods Ceylon Team. Silk Foods Ceylon (Pvt) Ltd. is a BRCGS- and FSSC 22000 V6-audited contract manufacturer in Matale, Sri Lanka, offering contract manufacturing, private labelling, co-packing, and in-house R&D for local Sri Lankan brand owners, FMCG companies, hotel and restaurant groups, and distributors. To brief a project: b2b@esilkroute.com.lk or +94 77 157 7294.


