Barcode scanning is a genuinely useful feature — for about 30% of what people eat.
It works well for packaged food with a readable barcode: protein bars, yoghurt, breakfast cereal, canned goods. It works terribly for everything else: restaurant meals, home cooking with fresh ingredients, food from local bakeries or delis, dishes that don't come in a box, anything you made yourself.
The problem is that MyFitnessPal's entire design philosophy is built around barcode scanning and database search. When you eat food that doesn't fit that paradigm — which, for most people, is most of the time — the logging experience breaks down.
The database trap
Even when you're not scanning a barcode, MFP asks you to search its database of 14 million food entries. That sounds like more is better. In practice, it means that searching "chicken breast" returns dozens of entries with wildly different calorie counts, created by different users with different assumptions about serving size and cooking method.
"I typed in 1 cup cooked quinoa and saw search results ranging from 120 calories to 225 calories. People just write in their opinions. It defeats the purpose of counting calories." — App Store review, 2025
The MFP model requires significant learned expertise to use well. You have to know which entries to trust, how to adjust for your actual portion versus the database portion, and when to create a custom entry. That expertise takes weeks to develop — and by the time most people develop it, they've already given up.
What a non-barcode alternative looks like
The cleanest alternative for people who hate barcode scanning is an app that doesn't require a database search at all. Instead of selecting an entry, you describe what you ate. "Leftover pasta with bolognese sauce, about two cups" or "chicken salad from the café, with avocado." The AI works out the rest.
This approach has a few advantages over database search:
It handles unusual foods. Local café meals, home recipes, dishes from your cultural background that aren't in an American-centric database. If you can describe it, it can be logged.
It's faster. Speaking a description takes five seconds. Database searching and portion adjustment takes two to four minutes.
It handles combinations naturally. "Saturday night dinner — lamb chops, roast potatoes, steamed greens with butter" is a single description of a complex meal. A database app requires separate entries for each component.
Accuracy: description vs. database
Research suggests that AI-based meal description logging and database logging are comparably accurate for common foods — both fall within 15–25% of actual calorie content. For weight management purposes, this level of accuracy is sufficient. What differs dramatically is the friction, which directly predicts whether people keep logging.
Common questions
What is the best food tracker that does not require barcode scanning?
Voice-first apps like Rekkon let you describe meals in natural language — for example, chicken wrap with avocado — and estimate calories and macros automatically without any barcode or database search. This approach works for restaurant meals, home cooking, and foods that do not come in scannable packaging.
Why does barcode scanning fail for most meals?
Barcode scanning only works for packaged foods with a readable barcode — approximately 30 percent of what most people eat. It fails for restaurant meals, home cooking with fresh ingredients, local bakery or deli food, dishes from cuisines not well-represented in US-centric databases, and anything homemade.
Is MyFitnessPal barcode scanning still free?
No. MyFitnessPal moved barcode scanning behind its Premium paywall in 2022. Premium currently costs approximately 80 US dollars per year. Several alternative apps including Rekkon do not use barcode scanning at all, instead using AI to estimate nutrition from spoken or typed meal descriptions.
Seven days. See what it notices.
No barcodes. No databases. No meal plans. Just talk to it.
Try free