What barcode scanning can and cannot do

Barcode scanning is the feature most prominently advertised by calorie apps and the one that solves the smallest share of the actual problem.

What it does well

For a packaged product, scanning is the best logging method available.

It removes search error — you get the exact product rather than something with a similar name. It removes the per-serving ambiguity, because the entry carries the manufacturer's own figures. And it takes two seconds.

For anyone whose diet includes packaged snacks, breakfast cereals, protein bars, yoghurts and drinks, it is genuinely useful and worth having.

What it does not do

It cannot scan food without a barcode. Home-cooked dishes, restaurant meals, food from a halwai, loose atta and dal from a kirana shop, vegetables, fruit, anything cooked fresh.

For most people in India, that is the large majority of what they eat. A feature that handles a minority of your meals is a convenience, not a solution.

It does not know how much you ate. The scan identifies the product and its per-100g values; you still enter the quantity. With multi-serving packets that is where most of the remaining error lives, which is the trap described in how to read a nutrition label in ten seconds.

It does not remove the label tolerance. Energy values on packaged food are permitted to differ from the measured value by roughly 20 per cent in most jurisdictions. A scan gives you the declared number, not the true one.

Coverage is uneven

Barcode databases are built where their users are. Products sold widely in North America and Europe are covered thoroughly; regional Indian brands considerably less so.

Crowd-sourced databases such as Open Food Facts fill some of the gap, with the quality variation that implies: photographs of labels transcribed by volunteers, sometimes correctly.

If you scan five Indian products and three are not found, that is the coverage you will live with, and it is worth testing before choosing an app on this feature — part of the evaluation in how to judge a calorie app's food database.

Where the same barcode gives different answers

You may notice two apps returning different values for the same scan. Usually one of three reasons:

Different databases. One reading manufacturer data, another reading a user-photographed label.

Regional formulations. The same brand and product name can be manufactured to different recipes in different countries.

Label revisions. A reformulated product keeps its barcode and changes its numbers, and databases update at different rates.

The general phenomenon is covered in why two apps give the same roti different calories.

What fills the gap

For the food that has no barcode, three methods, in order of accuracy.

Calculate your own recipes once. The pot method: count raw ingredients including all fat, divide by servings, save the number. This is the highest-accuracy option for home cooking and it is done once per dish — how to calculate the calories in a recipe you cooked.

Use composition table values for ingredients. A lentil is a lentil; published reference data for raw ingredients is genuinely reliable.

Use a default estimate for meals you did not cook. Fast, approximate and far better than a blank — how to log a restaurant meal you didn't cook.

One place it is genuinely underused

Scanning the things you buy rather than the things you eat.

Scan the oil bottle, the atta packet, the dal, the sugar, the milk carton — once — and save each as a custom ingredient with its label values. Those are the inputs to every home-cooked dish you will ever calculate, and having accurate per-100g figures for them makes the pot method faster and better.

Most people scan finished snacks and never scan the ingredients that make up the majority of their intake, which is the wrong way round.

The honest place for scanning

Use it for what it is good at. Scan the biscuits, the yoghurt, the protein bar, the bread, the packet of namkeen.

Do not choose an app primarily on this feature, and do not assume that a scan makes an entry authoritative. It makes it convenient, which is worth something, and it leaves the harder part of the problem exactly where it was.