by Carmen Sook, Director of Customer Solutions, BESTMIX Software, the USA
A premium dry petfood formula does not travel from spreadsheet to shelf in a single step. Before a kibble reaches the bag, it passes through a mixer, an extruder, a dryer and a coater. Each of those stages does something to the product. Moisture evaporates. Starch gelatinises. Fat is applied after the base formula has already been fixed. Nutrient ratios shift as the dry matter composition changes. By the time the finished product is weighed and labelled, it may look quite different from the formula that was approved.
For much of the industry’s history, formulation software handled this complexity with workarounds. Formulators built single-step recipes, applied manual correction factors for moisture loss and estimated post-coating nutrient levels using rules of thumb. The tools were not designed to reflect what actually happens on the production line. And in many segments, the margin for error was wide enough that this approach held.
That margin is narrowing. As the petfood category continues to premiumise, as label transparency demands increase and as recipe complexity rises, the gap between a formula calculated on paper and a product measured off the line becomes a real business problem. The industry needs formulation software that models the process not just the recipe.
Why pet food production defies single-step formulation
To understand why multi-step formulation matters, it helps to walk through what actually happens during production of a standard extruded kibble.
The process begins with dry and wet ingredient blending. Meat meals, grains, vitamins and minerals are combined in defined proportions. This blended mix then enters the extruder, where heat, pressure and mechanical shear cook the starch matrix and shape the kibble. At this point, significant moisture is driven off; typically 20 to 30 percentage points depending on the recipe and process parameters, which immediately changes the dry matter composition of everything in the formula.
After extrusion, the kibble enters a dryer, where additional moisture is removed to reach the target water activity for shelf stability. Each percentage point of moisture removed at this stage changes the relative concentration of every other nutrient. A formula that was calculated at 25 percent moisture content before drying will show different protein, fat and carbohydrate percentages once it reaches 10 percent moisture; even if no ingredient quantities changed.
The final step, coating, adds palatants, fats and oils to the surface of the dried kibble. These coatings are applied after drying, which means they are not part of the base formula that was calculated prior to extrusion. The fat content in the finished product, which directly affects the guaranteed analysis on the label, may be substantially different from the fat content in the uncoated extrudate.
Each of these transitions represents a genuine change in the composition of the product. If the formulation system does not account for each transition, the numbers it produces are not a model of the finished product. They are a model of what goes into the blender.
Why the stakes are higher today
The case for more precise formulation has always existed but several converging trends have made it urgent.
Premiumisation is the most visible driver. The premium and super-premium petfood segments have grown consistently over the past decade and with that growth has come a significant increase in recipe complexity. Shorter ingredient lists, higher-quality protein sources, functional ingredient inclusions and ‘clean label’ positioning all reduce the buffer that used to absorb small formulation errors. When a recipe has 30 ingredients, a 0.5 percent moisture estimation error has limited impact. When a recipe has 12, it matters more.
Functional claims are adding another layer of precision requirements. Formulas designed to support joint health, digestive function or immune response are often built around specific nutrient inclusion levels that need to be maintained throughout the production process, not just at the blending stage. A fish oil inclusion that supports an omega-3 claim must account for what happens to that fat fraction during extrusion and drying.
Label transparency is a third driver. Regulatory requirements and consumer expectations around nutritional accuracy continue to tighten. Guaranteed analyses are increasingly scrutinised, both by regulators and by informed consumers comparing products. A label that reflects the blended formula rather than the finished product is a compliance risk not just a formulation oversight.
Finally, portfolio complexity is expanding the surface area of the problem. A manufacturer running 40 or 50 SKUs, each with its own production routing and process parameters, cannot rely on manual correction factors applied consistently across every formula. The error surface is simply too large.
What multi-step formulation looks like in practice
A multi-step formulation system addresses these challenges by structuring the formula to reflect the production process directly, rather than treating it as a single calculation followed by manual adjustments.
In practice, this means that a formula is built as a series of linked sub-recipes, each corresponding to a production step. The blended mix is one level. The extruded and dried product is another. The coated finished product is a third. Each level has its own ingredient inputs, its own moisture accounting and its own nutritional targets or constraints.
When a formulator sets a target protein level in the finished product, the system works backwards through the production steps to determine what inclusion level is required in the blend, accounting for the moisture loss that will occur during extrusion and drying. When a coating addition is specified, it is applied at the correct step, so its contribution to fat content is calculated based on the dry kibble mass after drying, not the wet blend mass at the start of production.
Critically, when an ingredient changes, whether due to a supply disruption, a seasonal variation in raw material composition or a deliberate reformulation, the system recalculates across all levels simultaneously. The formulator sees the impact on the finished product composition, not just on the blend composition and can adjust targets at each step independently.
This architecture also enables more meaningful constraint setting. Rather than applying a single set of minimum and maximum nutrient constraints to a single formula, the formulator can set step-specific constraints; for example, defining moisture targets at each stage that correspond to actual dryer performance data or setting coating inclusion ranges that reflect equipment capabilities.
Closing the gap between formula and finished product
There is a narrative among a portion of the petfood formulation community that the software tools available cannot handle multi-step processes. This perception appears to be based on the state of the market some years ago when single-step formulation was indeed the norm across most commercial systems. That assessment is no longer accurate.
The capability to model production as a sequence of linked steps, each with its own composition changes, moisture accounting and constraint logic, exists in current formulation platforms like BESTMIX Software. The question is not whether the technology is available but whether manufacturers are applying it and whether their formulation workflows have been designed to take advantage of it.
For manufacturers that have not yet made this transition, the cost of remaining with single-step approaches is increasing. As portfolios grow more complex and product claims become more specific, the accumulation of manual correction factors and estimation-based adjustments introduces compounding error.
For those evaluating their formulation infrastructure, the key question to ask of any system is not simply whether it can optimise a recipe against a cost target. It is whether the system models the full production process from blend through to finished coated product and whether that model is updated automatically when any input changes.
As the pefood category matures and the expectations placed on both product performance and label accuracy increase, the formulation system must reflect what actually happens during production. This is not a software question in isolation — it is a question about whether the process of developing a formula and the process of manufacturing a product are genuinely aligned.
When they are, the guaranteed analysis on the label, the nutritional targets in the formula and the results of finished product testing should tell the same story. Multi-step formulation is the methodology that makes this alignment possible.
