A grid is not an opinion. A grid is a precondition. We are tired of arguing about this.
Designers love to talk about grids the way founders love to talk about company values. With reverence. With personality. As if choosing a 12-column grid over a 16-column grid is a worldview rather than the price of admission for laying out a webpage. It is not a worldview. It is the thing you do before you start working.
Here is what a grid is. A grid is a set of vertical lines that you don't draw, that nobody sees, that govern where text and images can sit so that human eyes can find the next thing to read without effort. The grid is doing the same job as the rules of grammar. You don't get points for inventing your own grammar. You get points for using grammar so well that nobody notices it and the meaning lands.
When a designer puts up a slide that says “we chose a 12-column grid because it allows for flexibility and modularity” they are not telling you about their design philosophy. They are telling you they have not yet started designing.
The actual design work happens above the grid. It happens in the typography. It happens in the rhythm of section spacing. It happens in the choice of which moments break the grid (briefly, on purpose, for emphasis) versus which moments respect it (almost always). It happens in the contrast between dense information and breathing room. It happens in the editorial decisions about what gets a full row to itself and what gets stacked alongside three siblings.
None of those decisions are easier or harder because the grid has 12 columns instead of 16. The grid is the floor. You don't get credit for building a strong floor.
We've worked with designers who treated the grid as the deliverable. The grid was airtight. The components were neatly inscribed within it. Every spacing value was a multiple of 8. And the result was unbearably boring software, because the grid was doing all the work and the designer was doing none.
We've also worked with designers who treated the grid as a suggestion. The result was incoherent software that gave you a faint headache after thirty seconds of looking at it.
The right answer is the boring one. Pick a grid early. Don't write essays about it. Stop showing it on the title slide of design reviews. Move the conversation to the parts of the work that are actually a design problem.
If you're hiring a designer and they want to talk to you about grids before they show you anything, hire someone else. The grid is the cover sheet. The TPS report is on the next page.