No company decides to have an incoherent portfolio. It arrives one SKU at a time, and every individual step is defensible.

A buyer asks for a smaller format for a discounter. A new variety comes into production. A market requires a different claim on front of pack. Someone in sales spots an opportunity in food service. Each addition is a real commercial decision, made in a real week, usually with a deadline attached. Each one is handed to whoever is available — the in-house designer, the printer's studio, an agency used once two years ago, a freelancer found the day the deadline landed. Each output is fine on its own.

Then one day the company assembles the whole portfolio for a trade show or a catalogue, and the effect is unmistakable: this is not a range. It is a collection of products that happen to share an owner. Different logo sizes, three greens, two typefaces, a hierarchy that reverses between formats, a premium tier that reads cheaper than the standard tier because it was designed eighteen months later by someone else.

The moment this becomes visible is usually somewhere between twenty and forty SKUs, but the number is not the point. The point is that the portfolio crossed a threshold where nobody could hold it in their head anymore, and no rule existed to hold it instead.

What is lost is specific and measurable in the negotiation.

Shelf presence disappears first. Four coherent facings read as a block and occupy more visual space than they physically own. Four incoherent facings read as four unrelated products and occupy exactly their own width. The company paid the same listing fees for a fraction of the effect.

Then transfer stops working. A customer who liked the standard product should find the premium one without help, and should recognise the new format as the same company. When the range does not signal its own relationships, every SKU has to earn its trial independently. That is the difference between a portfolio that compounds and one where each launch starts at zero.

And unit costs quietly rise. Where nothing is standardised, every new product is a design project rather than an application of an existing system: a new brief, a new round of approvals, new artwork from scratch, new print setup, and a fresh opportunity for something to arrive at the plant wrong.

The correction is not a redesign of forty packs. Very few companies can fund that, and doing it does not prevent a repeat, because the mechanism that produced the drift is still in place.

What prevents a repeat is deciding three things and writing them down.

What is constant across everything the company sells — normally the mark, its position on the pack, one typeface, and one structural rule about how the front is organised. This is what makes the range recognisable, and it should be short enough that nobody has to consult a document to remember it.

What varies, and what the variation means. Colour that signals variety rather than being chosen per project. A tier system where the difference between standard and premium is a defined, repeatable move, not a subjective attempt at looking more expensive each time. A format logic where a small pack is visibly the small version of the large one.

And who decides. Almost every portfolio drift traces back to an artwork approved by someone who had no basis for saying no, because no basis had ever been established.

With those three settled, a new SKU stops being a design question and becomes an application taking days, not weeks, and arriving correct because the decisions were made once rather than forty times. Cost per launch drops. Time to market drops. The shelf starts working as a block.

The reason this rarely gets addressed in time is that the damage is invisible from inside the company. Each product is reviewed alone, in a meeting about that product, against the brief for that product. Nobody sees the portfolio the way a buyer sees it — all at once, in one aisle, next to a competitor whose range was built as a system.

It costs nothing to look. Print every pack the company sells, put them on one table, and stand back three metres. Most companies have never done it, and most of the ones who do it come back with the same sentence: it looks like six suppliers.