Any sensible diet can work.
Here is what the evidence says.

I have one quiet advantage over most people who publish diet opinions: I keep score after they stop talking. Personal training is where nutrition arguments meet real life. A new client turns up carrying whatever diet they were most recently convinced by: a keto notebook, a screenshot of a fasting window, a shelf of branded shakes. A few weeks later, we both find out what that diet is actually made of, and whether a real person can live inside it.

Early on, I made the classic mistake. I had one plan and tried to fit every client into it. Some agreed and did well. Others kept their own eating (vegan, keto, plain "normal food") and did just as well, as long as the plan ran on real food and they actually liked living in it. No diet name ever predicted who would thrive. The fit between the plan and the person did, and so did the quality of how the plan was built.

That taught me to ask a better question. Not "which diet is best?" but "what does this diet share with every other one that worked?" So I sat down with the research, expecting it to name a winner. It didn't. Five landmark studies and one population map, all covered below, point the same way: the name of the plan is close to irrelevant. The construction is everything.

What follows is that evidence, the shared foundation every healthy diet stands on, an evidence grade for every pattern SafeTable supports, and the questions I now run every client through before they commit to a plan.

I went looking for a winner: the one study that would crown the right diet and end the argument. Five landmark studies and one population map later, I had finished with the argument instead. Not one of them cares about macro ratios the way the internet does. Together they explain why the diet wars never resolve, and why the people quietly getting results are not the ones arguing.

Each card below lists the study design, who was studied, how long it ran and what it found. The bold figures are the ones worth remembering.

Two people can follow the same named diet and, a year later, be in two very different places. A big part of that gap comes down to how processed the food is. Nutrition researchers use a classification called NOVA, which sorts food by how much it has been processed, not by its nutrients.

The research gives three reasons why processing matters so much.

This is why the diet wars are mostly noise. Almost every named diet (keto, paleo, vegan, Mediterranean, DASH, Whole30) quietly agrees on one instruction: stop eating most ultra-processed food. From an evidence standpoint, that shared agreement does more of the work than the differences between the diets.

If one macro ratio unlocked long life, you would expect to see it in a long-lived population eating that ratio. Instead, the pattern shows up at every ratio, always on the same foundation. Table 1 shows five examples. Read each row as a whole way of eating, not as a macro target.

Not a single healthy long-lived population ran on ultra-processed food. Not one ran on a diet that offered nothing to the body or to the people eating it. If macro ratios were the key, Okinawa and Ikaria could not share a planet.

For years I asked the wrong question whenever a client succeeded: "what made their diet the winner?" The better question is what their plan had in common with every other plan that worked. Strip the brand names off the five healthiest diets, or the five longest-lived towns, and the same five fundamentals are running underneath. You can swap between them freely.

After enough years on the gym floor, I stopped being impressed by which diet people declared and started noticing where it sat. This grid is that observation, drawn out. Pick any cell and you will find evidence-backed eating inside it. The position tells you where a pattern sits on carbohydrate and fat. The colour tells you how much evidence backs it. Colours mark evidence strength, not moral worth. The grade key is explained in How we research, further down.