Adifo BESTMIX Alternatives for Feed Mills and Formulators: Where Custom Software Actually Helps

Why formulators start looking for a BESTMIX alternative

The complaint is almost never about the mathematics. Ask a nutritionist why the company is evaluating alternatives and you will hear about everything that surrounds formulation rather than formulation itself. Ingredient prices arrive by email and get typed in, so the optimiser is solving against last week's market. A formula change approved on Tuesday reaches the batching system on Friday because somebody has to transcribe it. Quality results live in the lab system and nutrient matrices live in the formulation system, and reconciling them is a manual ritual. Customers ask for a specification sheet and the specification sheet is assembled by hand.

The second complaint is version control and approvals. Formulas are living documents with commercial and regulatory consequences, and the question of who approved which version, against which constraint set, at which ingredient prices, tends to be answerable only by the person who did it. When that person retires, the company discovers how much of its formulation practice was undocumented.

Neither of these is a reason to replace a formulation engine. They are reasons to build the operating system around it, which is a different and much cheaper project.

What BESTMIX genuinely does well

Least cost formulation is a harder problem than it looks. You are optimising against nutrient minimums and maximums, ingredient availability, physical constraints, price, and often multiple blends and plants at once, with results that have to be nutritionally sound rather than merely mathematically optimal. Adifo has been doing this for decades, and the value sits in the accumulated handling of constraint types, nutrient models, and the practical edge cases that make a solver useful to a nutritionist rather than interesting to a mathematician.

The surrounding recipe and specification management matters too. Ingredient and nutrient databases, formula versioning, and the declarations required for labelling and customer documentation are unglamorous and expensive to build correctly. If you formulate for animal feed, petfood, or food and your constraint sets are conventional, this is a category where buying beats building by a wide margin.

There is also a quiet benefit in holding a defensible formulation record. When a customer, an auditor, or a veterinarian asks why a diet contained what it contained, an answer that points to a versioned formula, a named constraint set, and a recorded approval is a far better answer than a spreadsheet and somebody's memory. Formulation packages carry that provenance as a by product of doing their job, and reconstructing provenance after the fact is considerably harder than capturing it as you go.

Where it actually strains

The first strain is that a formulation system is not a plant system, and most companies eventually want it to behave like one. Connecting formulation to batching and process control, to purchasing, to inventory, and to the lab is integration work, and every one of those connections is specific to the equipment and systems you happen to own. That work does not come in the box, from any vendor.

The second is data discipline. An optimiser is only as good as its nutrient matrices and its prices, and both decay. Matrices drift as suppliers change, and prices go stale the moment the market moves. Software cannot fix a maintenance problem, but the absence of automated price feeds and matrix review workflows turns a maintenance problem into a confident wrong answer, which is worse than an obvious one.

The third is the ordinary economics of specialist software. Licensing is per user and per site rather than published, so the cost of giving a purchasing manager or a plant supervisor read access is a commercial conversation rather than a click. That is how organisations end up with one licensed user emailing screenshots, which is exactly the workflow you were trying to remove.

Your real options

Staying is the first, and for the formulation engine itself it is usually correct. The second is another commercial formulation package. Format Solutions is the other name you will hear most often, and there are regional and species specific tools beyond it. Evaluate on constraint handling, multi blend and multi plant capability, nutrient model depth, and how the vendor supports integration, not on the interface. Switching optimisers is disruptive because nutritionists carry deep familiarity with how their current tool behaves, and that familiarity is a real asset.

The third option is building on a commercial optimisation solver. This is genuinely viable for a large integrator with nutritionists and developers on staff, and some large protein and petfood companies do exactly this so the formulation logic can be embedded in their wider planning systems. It is not viable for a single mill. You are taking on responsibility for the model, the constraint library, and every nutritional consequence of a bug.

The fourth, and the one most companies should look at first, is keeping the optimiser and building everything that touches it.

When a custom build pays back

The custom cases that pay back consistently in feed and food formulation are unglamorous:

• An ingredient price pipeline that pulls contracted and spot prices automatically, so formulas are optimised against current cost rather than last month's.
• A formula approval workflow with versioning, electronic sign off, and an audit trail that answers who approved what, when, and against which constraints.
• A handover to batching that pushes an approved formula to the plant control system without transcription, with confirmation back.
• A cost variance view comparing formulated cost against actual batched cost, which is where real money leaks and where almost nobody looks.
• A customer specification and declaration portal that generates documents from live formula data instead of a shared drive.

Each of these sits next to the formulation system rather than replacing it, and each one removes a manual step that costs both time and accuracy.

What you should not rebuild

Do not rebuild the optimiser unless formulation is a core competency you intend to staff permanently. The solver is not the hard part; the constraint library, the nutrient models, and thirty years of accumulated handling of awkward cases are. A custom optimiser that is ninety percent right produces formulas that look fine and cost you money or performance quietly. That is the worst failure mode in this category, because nobody notices until an animal performance trend or a margin trend goes the wrong way for two quarters.

Migration and integration reality

If you do switch formulation packages, the migration is mostly about nutrient data rather than recipes. Ingredient definitions, nutrient matrices, constraint sets, and the assumptions embedded in them have to move with their meaning intact, and meaning is where migrations fail. Run both systems against the same set of formulas and compare the outputs before you trust the new one. If a formula solves differently, understand exactly why before you accept it, because a difference is either an improvement or a defect and you cannot tell which from the price alone.

For integration projects the reality is simpler but still needs planning. Batching and process control systems vary by vendor and vintage, some speak modern interfaces and some do not, and the discovery phase should establish exactly what is possible before anyone commits to a delivery date. Where a system genuinely cannot be integrated, a supervised handover with confirmation is still a large improvement over retyping.

Cost bands

Adifo licensing is quoted per user and per site rather than published, so compare on the total including additional users you will inevitably want. On the build side, using Digital Heroes delivery experience as the reference: a workflow and integration layer covering price feeds, approvals, and plant handover typically runs $40k to $95k over 8 to 14 weeks. A broader formulation to plant platform with cost variance reporting and a customer specification portal runs $130k to $280k. A genuinely custom optimisation capability built on a commercial solver starts around $150k and should be treated as a permanent product commitment rather than a project.

The honest recommendation

Keep BESTMIX if your formulation needs are conventional and it is producing sound formulas, because the engine and the nutrient model are the expensive part and they work. Switch packages only for a specific capability your nutritionists have tested, and expect the familiarity cost to be real. Build the layer around formulation almost regardless of which package you keep, because price freshness, approval traceability, plant handover, and cost variance are where the money and the risk actually live. And treat any proposal to rebuild the optimiser with suspicion unless you are prepared to fund nutrition and engineering together, permanently.