Advanced Food Product Development (NPD): Strategy, Reformulation and Global Regulatory Routes

Advanced food product development (NPD, new product development) is the work of deciding which food products a business develops or stops, proving each recipe with designed experiments and sensory statistics, and making it legal in every market before launch. This 24 hour online course does that for a South African food and drinks group, from twelve competing projects to a reduced-sugar still drink for South Africa and the Netherlands. It ends in a board-ready launch dossier built on your own case variant. For NPD managers, heads of technical, R&D leads and exporters.

Twelve projects ask for more than twice the budget, and the one at the top must cut its sugar by at least 30 percent for two markets

You start on a Monday as Group NPD and Technical Director of Thuthuka Foods Group, which owns Thuthuka Fresh Meals (chilled ready meals), Amanzi Beverages (still drinks) and Karoo Meat Processors (processed meat). Your predecessor left twelve projects asking for R16 400 000 against a development budget of R8 000 000, and by Friday the group executive wants to know which you will back and which you will stop. Two of the top seven on score cannot go ahead as briefed, a MAP chilled curry range promised a 21-day life with no controlling factor and a sucralose-sweetened drink for toddlers aged 1 to 3. Karoo’s polony is over its R.214 sodium limit today, at 1 174.9 mg per 100 g against 1 150. The project at the top of the list is Amanzi’s still fruit drink, now at 10.92 g of sugar per 100 ml, which must carry at least 30 percent less sugar in South Africa and the Netherlands. To use sweeteners in the EU it must also be energy-reduced by at least 30 percent. In South Africa the Health Promotion Levy is charged on the sugar an accredited laboratory reports. Whether “reduced sugar” may go on this drink’s South African label at all is an open question under R.146 regulation 53(2), which you flag rather than assume. Thuthuka Foods Group is a training scenario built from real patterns, and the course works each problem on the base case, then has you do the same on your own case variant.

Advanced Food Product Development (NPD): Strategy, Reformulation and Global Regulatory Routes. Course code FS143. Advanced, Level 4. 24 hours online, self paced, on a phone or a laptop. R6 500, once off.

Rated 4.9 out of 5 from more than 2 500 delegate reviews across ASC courses, with 6 700+ course enrolments. A course by the ASC Food Safety Consultants team.

Enrol for R6 500

Who it is for

• NPD managers and heads of NPD who decide which projects are funded and which are stopped
• Heads of technical and technical directors who sign the technical release for a launch
• R&D leads and senior developers who design the trials and defend the results
• Consultants who lead NPD for clients
• Exporters, and the technical and regulatory staff who take South African products into the EU, the UK or the US

The course assumes you can already run a project through gates, read a sensory difference test and plan a factory trial, from the Intermediate course or from leading NPD on a certified site. Every activity runs on the Thuthuka Foods Group case, so you can finish the whole course without a portfolio, a laboratory or an export market of your own. Each lesson also ends with a version of the task for your own business.

What you walk away with

You finish with eight pieces of work, from a portfolio funded inside an R8 000 000 budget to a board memo of 250 to 350 words. You make them on your own case variant in the assessed practical, and on your own products in practice mode.

1. A ranked portfolio, with the legal and food safety gates run before any score, kill decisions a board and an auditor would accept, and the funding line drawn inside the budget
2. A consumer and sensory evidence pack, with JAR penalties put in priority order and a similarity call made on the upper bound of pd, not on a p-value
3. A mixture design for a sweetener blend, with a quadratic Scheffé model, its lack-of-fit check at a replicated centroid and a predicted optimum that a confirmation run still has to confirm
4. A levy and claims position for South Africa and the EU, with the levy worked from an accredited sugar report, the EU sweetener permission tied to the energy-reduced condition and the open South African claim flagged with an owner and a date
5. Preservation checked against the R.965 and EU limits and at the top of the new pH specification, with the benzoate and ascorbic acid pairing broken
6. An EU label corrected line by line, and a route check for each new ingredient
7. A scale-up risk register with owners and dates, and the conditions a co-manufacturer must meet before the transfer
8. A board decision, with a 250 to 350 word memo that carries your own figures, dates, owners and open flags

The practical, a launch dossier in six stages

Module 8 is the capstone, the launch dossier for Amanzi’s reduced-sugar drink, built in your browser in the Launch Dossier, one stage for each lesson from 8.3 to 8.8. Every stage answers a question the board will ask, with a number, its method and a document that carries a reference and a date.

1. Portfolio ranking and kill decisions. Gate results with their reasons, weighted scores, the rank, the cumulative spend and the funding line.
2. Consumer and sensory evidence. JAR means, drops and weighted penalties, the order of the fixes, and the similarity call on the co-manufacturer’s first run.
3. The mixture design and the predicted optimum. The constrained sweetness region, real shares and doses, the Scheffé coefficients worked by hand, the check point, the lack of fit and the confirmation run.
4. Levy, claims, sweetener permission and preservation. The tariff heading, the sugar basis, the levy with and without an accredited report, each claim in both markets, and the active preservative at the top of the pH specification.
5. The EU label and the novel ingredient route check. The seeded errors on a draft EU can label found and corrected, the checks the artwork still needs, and a route for each new ingredient.
6. Scale-up risk, transfer, governance and the board memo. Risks with owners and dates, the conditions for the co-manufacturer, the decision the evidence supports and a memo of 250 to 350 words.

Three working tools

Two free practice tools, and the six-stage Launch Dossier, which marks the practical on your own case variant.

Methods in the lessons, and the certificate

You can see the method before you pay. Open the free Sensory Analyser in advanced mode, load the case study numbers and watch 44 correct out of 120 pass a similarity test, on a result that a difference test alone would have left undecided. Then read your own panel’s result.

Inside the course, 8 modules and 58 lessons

Module 1. Strategy, portfolio and governance

Read an inherited portfolio for the legal, safety and obligation questions that must close before any money moves. Then turn the group’s strategy into scoring weights and gates for NPD portfolio management, rank twelve projects, fund them to budget and see how the ranking moves when the weights change. Choose between weighted scoring, BCG, Kano, jobs to be done and the funnel, knowing what each cannot see, and design gate governance that scales with project risk across three businesses, with agile sprints where they help. Quote launch failure evidence with its scope. Roughly a quarter of new US grocery products stopped selling within a year and about 40 percent within two (Victory and others, 2021, Marketing Letters, 83 719 new US products from 2002 to 2009). The popular 80 to 90 percent failure figure is not supported (Castellion and Markham, 2013). Kill a project the team loves without letting the money already spent decide, and set up a private label relationship in which the specification, the label, change control, audit and liability each have an owner.

Module 2. Consumer and sensory science for decisions

Turn an NPD decision into a research brief, with the method, the population, the action standard and the POPIA steps fixed before fieldwork. Reason about sample size for an acceptance test on the 9-point hedonic scale, and compute a JAR (just-about-right) penalty analysis whose weighted penalties become the next reformulation brief. Plan discrimination and similarity tests from alpha, beta and pd, read a similarity result from an exact upper bound on pd and defend a pass or a fail to a board, with sequential testing under ISO 16820. Then specify a trained descriptive panel to ISO 13299 and ISO 8586:2023, see what preference mapping adds in plain language, and balance serving orders and incomplete blocks to ISO 29842:2024.

Module 3. Designed experiments for formulation

See from a two-factor result why one factor at a time misses interactions and optima. Build the sign table for a 2 x 2 x 2 factorial and calculate every effect by hand, with error from replicates, centre points or a half-normal plot. Then screen with fractional factorials, their aliases and resolution, and find an optimum with a central composite design. Set up a constrained three-component mixture in sucrose-equivalent sweetness shares, convert it to pseudo-components, fit the quadratic Scheffé model on a {3,2} simplex-lattice with a replicated centroid, and see why a seven-run special cubic cannot be trusted. Learn what a spreadsheet can and cannot do, then take Amanzi’s sweetener blend from brief to decision. On the case the model shows no lack of fit (p = 0.670) and predicts a mean liking of 7.013 against an action standard of 6.8. The blend doses 23.4 mg/l of steviol glycosides as steviol equivalents and 40.95 mg/l of sucralose, from illustrative potencies, inside the EU maxima of 80 and 300 mg/l. A replicated confirmation run still has to confirm it.

Module 4. Reformulation under South African regulation

Open a reformulation as a controlled change, knowing that a new recipe means a new analysis under R.146 regulation 50(9). Work the Health Promotion Levy from the laboratory report, converting g per 100 g to g per 100 ml with each recipe’s own density, and see what the deemed 20 g does to the business case. Test sweetener permission and label wording under R.733 of 2012. Bring Karoo’s polony and boerewors inside the R.214 phase 2 sodium limits of 1 150 and 600 mg per 100 g, with every sodium source counted, and check a fat change against R.127 of 2011 on the right basis. Rework preservation after a sugar, acid or salt cut. Check benzoic and sorbic acid against the R.965 soft drink limits, work their undissociated share at the top of the pH specification, and deal with benzene, aspartame stability and salt on the water phase. Plan the challenge test to ISO 20976-1:2019 that keeps the shelf life. Then test each claim against R.146 regulation 53 and R.1283 regulation 14(6), and decide where regulation 53(2) blocks a claim or leaves it open. Handle the R.3337 front-of-pack draft without labelling to it, and place each meat or meat-replacing product under R.1283, R.2410 or R.6436 of 2025.

Module 5. Novel ingredients and regulatory routes

Say what makes an ingredient novel and why South Africa has no novel food approval route, and state the South African position on hemp, CBD, insect protein and cultivated meat without overstating it. Take an ingredient through the EU novel food regulation and test whether an African ingredient can use the EU traditional food notification. Learn how baobab and Bambara groundnut reached the Union list by different routes, and run a status check for rooibos, honeybush, marula and moringa that states only what has been confirmed. Then US GRAS, self-affirmed or notified, with mandatory notification treated as the proposal it is, the GB route since 1 April 2025, what Codex can and cannot do, and the novel ingredient dossier that turns a route check into a gate decision.

Module 6. Designing for export markets

Choose an export market on evidence and read its rules in order, from Codex and the WTO SPS Agreement to the African trade agreements. Build the EU label for the reformulated drink, with the Article 9(1) particulars, the 14 allergen groups, the x-height, the nutrition declaration, QUID, the Annex III sweetener statement and the Article 15 language check. Then work the energy cut with the factors in Annex XIV to Regulation 1169/2011, and show why the drink must be energy-reduced by at least 30 percent to use sweeteners in the EU. Read row 1.2 of Regulation 2073/2005 for Listeria monocytogenes as amended from 1 July 2026, and how growth potential from a challenge test is read against it. Then set out what a US shipment needs, from the nine major allergens and the Nutrition Facts label to facility registration, prior notice and the importer’s supplier verification. Finish with Great Britain’s assimilated law and its novel food, promotion and advertising rules, including the restrictions on advertising less healthy food from 5 January 2026. Then use Codex as the common design baseline, with one allergen matrix across Codex, the EU, the US and South Africa.

Module 7. Scale-up risk, validation, transfer and substantiation

Build a scale-up risk register for a move to a co-manufacturer, keeping the heat transfer and mixing effects that engineering evidence supports apart from what experienced people see on a line. Show a board or a certification body that a change was validated at the worst case and before it went live. Plan confirmation runs with true replicates, a random order and a sampling plan written in advance. Write the technical agreement that carries the product to a co-manufacturer, with change control, liability, the levy, trade marks, designs and know-how. Then run one claims register for the portfolio under FSSC 22000 2.5.2(c), and design shelf-life verification at scale under 2.5.13(e), which Version 6 applies to every category in 2.5.13. Set launch decision rights across the group and the co-manufacturer, run the first post-launch review and prune by a rule written in advance.

Module 8. Capstone, the launch dossier

What a launch dossier must prove to a board and to a certification body, and how the server marks it, criterion by criterion. Then the six-stage practical in the Launch Dossier on your own case variant, ending in the board memo, and the final.

How you are assessed, and the certificate

• Two scored quizzes in every module. A Checkpoint in the middle of the module draws 10 questions at random from 15, at 45 seconds each, including one-step calculations. A Knowledge Check at the end draws 15 at random from 25, at 30 seconds each. 70 percent to pass and three attempts, with the questions drawn at random on every attempt.
• Fourteen games, two in each of Modules 2 to 7 and one in Modules 1 and 8. Modules 1 to 4 bring “Which tool answers this question?”, “Which test?”, “Read the result”, “Effect or noise?”, “Which design?”, “Levy or not?” and “Is this reformulation safe and compliant?”. Modules 5 to 8 bring “Novel or not, and where?”, “Which route?”, “EU label check”, “Which market says so?”, “Validated, verified or neither?”, “Enough evidence?” and “Defend it or drop it”. Ten quick decisions drawn from 14, at 45 seconds each, with 70 percent to pass and three attempts, like every other scored assessment.
• A six-stage online practical in the Launch Dossier, on one case variant generated for your account, so your prices, volumes, sugar, pH readings, sensory counts, liking scores and seeded label errors are your own. You enter your own figure or decision before the dossier shows its own, and the marks go to your entries. A figure passes within 1 percent of your variant’s key, or within half a unit of the last decimal place the field asks for where that is wider, and counts, ranks, the cumulative spend and every choice must be exact. Filling every field earns at most 20 percent of a stage, and an evidence field needs a reference and a date. Each stage is marked automatically the moment you submit, needs 70 percent, allows three attempts and opens the next stage. Progress is saved with your account, so you can start on a laptop at work and finish on a phone at home. Nothing is uploaded and nobody waits for a marker.
• Your own words in the stage 6 memo. The board memo of 250 to 350 words has to carry your own rand figure, a date and a named role who owns each open item, and it has to address every open flag in your variant. A memo with no figure of yours and no detail that could only come from your variant is checked for other signs that you did not write it. With two or more such signs it is returned unmarked under ASC’s rule on machine-written submissions, and the attempt is used. A second memo returned this way locks the practical and ends the enrolment without a certificate and without a refund. A weak memo in your own words with your own numbers is marked, and the feedback tells you what to add.
• A completion code. When all six stages show Passed, the dossier gives you a completion code. You paste it into the practical result assignment, which marks itself, before the final.
• A final assessment of 40 questions drawn at random from 88, across all eight modules, at 30 seconds each. 70 percent to pass and three attempts, with the questions drawn at random on every attempt, video proctored, with copying blocked.
• Feedback that teaches. A failed attempt tells you which lesson to reread, never the answer. After a third failed attempt on a quiz, a game, a practical stage or the final, a R250.00 reopening fee and a single-use unlock code apply.
• The certificate is issued once you have passed every Checkpoint, game and Knowledge Check, all six practical stages and the final, pasted your completion code into the practical result, and submitted your course review.

Twenty-four hours, and what R6 500 is measured against

The course is 58 lessons of 15 to 25 minutes, plus the quizzes, games, practical and final, about 24 hours in all. It works well as a module a week for eight weeks beside a working week, on a phone or a laptop. Set R6 500 beside two things a group NPD programme often gets wrong.

• A project funded on its score. In the course case, the MAP chilled curry range on a 21-day life with no controlling factor scores 74.0 and ranks fifth of twelve. Funded on its score, it takes R1 100 000 of the R8 000 000 budget, after R650 000 already spent on the line study. As briefed, with no controlling factor, the UK FSA and FSS guidance caps its life at 10 days from first packing. That guidance is not South African law, and shelf-life counts cannot support 21 days. The food safety gate stops it before the score is read. R1 100 000 is about 169 times the course fee.
• A sugar report that lapses. Without an accredited laboratory report the levy is charged on a deemed 20 g of sugar per 100 ml, whatever the drink holds. In the course case the reformulated drink then pays 116.69 cents a can instead of 20.60 cents. At the 26-week review a five-week lapse in the report put 480 000 cans on the deemed basis and cost R461 209.32 of levy the drink did not need to pay, about 71 times the course fee. The accredited analysis costs R60 000 a year.

Enrol for R6 500

What this course is not

It is not a statistics course, a regulatory affairs course or a sensory science course. It teaches the designed experiments, the sensory statistics and the law an NPD lead needs to make a decision and defend it, and points you to the ASC courses that teach labelling and shelf life in full. It does not replace your laboratory, a written answer from SARS, a regulatory adviser in the market you export to, your certification body or your own licensed copy of your scheme. The ASC certificate does not approve a product, a claim, a label, a process or a site. It does not teach gate scorecards, factory trial protocols or choosing and reading a sensory test again. Those belong to the Intermediate course, and this course starts where it stops, with statistical power, optimisation and defending the decision to a board.

Three levels, and where to go next

If you have not yet run projects through gates on a certified site, take Intermediate Food Product Development (NPD): Running Projects in a Food Factory, 16 hours, R3 950, first. It teaches choosing and reading, which test, which trial and what the result means, so this course can start where it stops.

If you are new to product development, start with Introduction to Food Product Development (NPD): From Recipe to Retail Shelf, 8 hours, R1 950. It takes one recipe through every stage and gate.

For the full route a product developer follows, from food science to allergen labelling, see the product developer courses pathway.

Related ASC courses for the questions this one raises.

• Advanced Food and Beverage Labelling for sugar declaration, sweetener systems and the label in depth
• Labelling Claims and Regulatory Framework for claims and how each is substantiated
• Overview of Conducting a Shelf-Life Study for the study behind a date
• Food Technology for Industry Professionals for scale-up and plant equipment
• Allergen Management in Food Supply Chain for allergen controls when a recipe or a supplier changes

Buying for a team

If you are training an NPD leadership team, the technical heads of several sites or the people who prepare your board papers, ASC enrols teams in bulk, invoices the company and gives the manager a dashboard to track completion. Get a team quote, or WhatsApp ASC on +27 61 483 0381.

For the board members and commercial managers who read those papers, Food Product Development (NPD) for Management works the same Thuthuka Foods Group case from the executive committee’s chair, in 4 hours for R1 495. It teaches them to run the gates before the scores, keep money already spent out of the decision and send back a memo that asks for “Go with conditions”.

Questions people ask before they enrol

Who is this advanced food product development course for?

NPD managers and heads of technical, R&D leads and senior developers, technical directors, consultants who lead NPD for clients, and exporters taking South African products into the EU, the UK or the US. It is written for people who decide what gets developed and defend those decisions to a board. The course is 24 hours online, self paced, and costs R6 500.

Do I need to do the Intermediate course first?

Not if you have led NPD projects on a certified site. The prerequisite is Intermediate Food Product Development (NPD) or that experience. This course assumes you can already run a project through gates, choose and read a sensory difference test and plan a factory trial. It builds from there into statistical power, optimisation, reformulation law, export rules and defending the decision. If you have not run gated projects on a certified site, take the 16 hour Intermediate course first.

Do I need my own portfolio, a laboratory or an export market?

No. Every lesson activity and the assessed practical run on a provided case, Thuthuka Foods Group, a training scenario with chilled ready meals, still drinks and processed meat businesses and twelve projects to rank. So you can finish the course without data of your own. If you do have your own products, the DOE Planner and Analyser, the Sensory Analyser and the Launch Dossier also run in practice mode on your own figures, and nothing about your own products is sent to ASC.

Will it work in my plant?

For the method, yes. The gates, the sensory statistics, the mixture design, the claims screen, the route check, the transfer agreement and the board memo work for any food or drink. Every lesson has a version for your own business, and practice mode takes your own figures. Some rules belong to certain products. The levy applies to the drinks listed in Part 7A, and the R.214 sodium limits to the foods they list, processed meat among them. For a market the course does not work through, Lesson 6.1 shows how to read its rules in order, from Codex and the WTO SPS Agreement to the importing country’s own law.

How is the Health Promotion Levy, South Africa’s sugar tax on sugary drinks, worked out?

The levy is 2.21 cents per gram of sugar above the first 4 g per 100 ml, under Schedule No. 1 Part 7A to the Customs and Excise Act at the rate set by GN R.506 of 2019, checked on 31 October 2026. The sugar per 100 ml comes from a test report by a SANAS or ILAC accredited laboratory, and without one the drink is deemed to hold 20 g per 100 ml whatever its recipe. Laboratories often report per 100 g, so multiply by the drink’s density first. In the course case 10.5 g per 100 g at 1.040 g/ml is 10.92 g per 100 ml, which is 50.47 cents on a 330 ml can, while the deemed basis charges 116.69 cents. Juices of heading 20.09 are outside the levy, and exported drinks carry no South African levy once the export is proven. Lesson 4.2 works it through, with the question to put to SARS on whether non-nutritive sweeteners count as sugar content.

Can a fruit drink say “reduced sugar” in South Africa?

Not until an open question is answered. R.146 regulation 53 needs at least 25 percent less sugar than a similar food, with the percentage and the comparator stated. The claimed food carries its nutrition table, with the nutritional information of both foods. Regulation 53(2) bars comparative claims on foods with Agricultural Product Standards Act compositional standards unless the standard allows them. Whether one covers a still fruit drink is not settled, so the course flags the claim with an owner and a date and keeps it off the artwork. Lesson 4.7 works it through.

Can a reduced-sugar drink use sweeteners in the EU?

Only if it is energy-reduced or has no added sugars. In category 14.1.4, flavoured drinks, Regulation 1333/2008 permits sweeteners only in drinks with energy reduced by at least 30 percent against the original or a similar product, or with no added sugars. The maxima include 80 mg/l for steviol glycosides as steviol equivalents and 300 mg/l for sucralose. Under Regulation 1924/2006, as amended by Regulation 1047/2012, the claim “reduced sugars” also needs at least 30 percent less sugar than a similar product and energy no higher than it. The course case keeps some added sugar, so it must be energy-reduced. Worked with the energy factors in Annex XIV to Regulation 1169/2011, its design falls from 189.0 to 118.2 kJ per 100 ml, 37.5 percent less, a design estimate to confirm on analysed values. In South Africa the comparative threshold is 25 percent, and whether “reduced sugar” may go on this drink at all is an open question under R.146 regulation 53(2). Lessons 4.7 and 6.3 cover both.

What are the R.214 sodium limits for polony and boerewors?

R.214, as amended, limits total sodium per 100 g. In phase 2 that is 1 150 mg for cured processed meat, 650 mg for uncured processed meat and 600 mg for raw-processed meat sausages such as boerewors. The row follows the product’s SANS 885 class, not its name, and every sodium source counts, from salt and phosphate to nitrite and the meat itself. In the course case the polony carries 1 174.9 mg, 24.9 mg over its limit today, and the new recipe reaches 870.2 mg, 25.9 percent less. Lesson 4.4 works it through.

Does a non-significant triangle test show that two products are the same?

No. A difference test controls alpha, the risk of finding a difference that is not there, and says little about beta, the risk of missing one that is. To show similarity you fix beta, the largest acceptable proportion of distinguishers (pd) and the number of assessors before testing, and the products pass only if the exact upper bound on pd is below that maximum. In the course case a co-manufacturer’s first run gives 44 correct of 120 in a triangle test. The upper bound on pd is 0.1677 against a maximum of 0.20, so similarity passes, while the difference reading of p = 0.2472 could not have shown it on its own. Lesson 2.4 and the Sensory Analyser’s advanced mode work it through.

How is the course assessed, and what happens if I fail?

Each module has two scored quizzes, a Checkpoint and a Knowledge Check, plus a game, and Modules 2 to 7 have two games each. The course ends with a six-stage online practical on your own case variant and a final assessment of 40 questions. Quizzes, games, practical stages and the final all need 70 percent, with three attempts each. The quiz and final questions are drawn at random from a larger bank on every attempt, and the practical runs on the one case variant made for your account. Everything is marked automatically, and feedback on a failed attempt names the lesson to reread. After a third failed attempt a R250.00 reopening fee and a single-use unlock code apply. A board memo with nothing from your own case and two or more other signs that you did not write it is returned unmarked under ASC’s rule on machine-written submissions, and the attempt is used. A second memo returned this way locks the practical and ends the enrolment without a certificate and without a refund.

Is the course accredited, and does the certificate approve my product?

It is a non-credit-bearing short course. You receive a QR-verified ASC certificate with your name, the course title, the code FS143 and the date, which your employer can use in its competence records and an auditor can check online. It is issued once you have passed every assessment, pasted your practical completion code and submitted your course review. It shows that you completed and passed this course. It does not approve a product, a claim, a label, a process or a site.

Enrol

R6 500, once off. 24 hours, on your phone, at your own pace. You finish with a launch dossier a board and an auditor can both read, and a clear list of what still needs an analysis, a confirmation run or a written answer from SARS.

Your learner account is created from the order, and the course opens once payment clears.

Enrol for R6 500