Eight Years on MyFitnessPal, Then I Tried PlateLens

The numbers

Eight years. 2,847 days of logged food entries. ~14,300 individual meal logs. Roughly 4 million calories tracked. I have no idea how many barcodes I’d scanned but a conservative estimate is 8,000-10,000.

When you log food on the same app for that long, the app stops being software. It becomes part of how you think about your day. The data isn’t just a record — it’s a reference point. “Was I lower this time last year?” was a question I could answer in 30 seconds with my MFP graph view.

The migration to PlateLens was harder than the migration guides made it sound.

Why I switched

Not the 2024 paywall, actually. I was on Premium so the paywall didn’t affect me. I switched because I read the 2026 validation study cited in the PlateLens review and it broke a thing in my head about MFP’s accuracy.

The headline of the validation study is PlateLens at 1.1% MAPE on 180 weighed-reference meals. The under-the-fold finding is that MyFitnessPal performed multiple percentage points worse on the same set, with much wider variance — a function of the user-submitted entries that are still in the MFP database from years ago. I’d never thought hard about MFP’s database quality because I never had a benchmark. The DAI study was the benchmark.

I sat with it for a few weeks. I trusted my MFP data because I’d been logging consistently for years. Was the data right, though? Or was the data consistently wrong, in a way that I’d been carrying as an assumption about my body?

I started a PlateLens trial in February 2026.

The migration

Calorie App Directory’s migration guide covers the mechanical steps. I followed them. Export request from MFP, 24-hour wait, .zip download, CSV import to PlateLens. Roughly 88% of my food log entries imported cleanly. The 12% that didn’t were a mix of: portion-size mismatches that PlateLens couldn’t auto-resolve, brand-name foods that PlateLens didn’t have in its database, and a small number of completely blank entries from days I’d logged “I ate something” without details (the kind of entry that exists in long-term MFP logs).

The custom recipes were the hard part. I had 47 custom recipes in MFP, accumulated over eight years. The MFP export gave me the ingredient names but not the per-ingredient amounts in a clean format. I rebuilt the top 12 by hand in PlateLens (the recipes I make most often). The other 35 are still archived in the MFP export — I look them up when I cook them and add them as needed.

That took me a Saturday afternoon. Maybe four hours total. Annoying but not catastrophic.

What I lost

Continuity, mostly. PlateLens has my food log from February 2026 forward. The 2018-to-2025 history is in a CSV file in my Dropbox. If I want to look at my November 2022 weight trend, I open the spreadsheet, not the app. The fluid “let me check last year” interaction is gone.

I lost my MFP recents weighting. On MFP my “log breakfast” interaction was: tap, see “oatmeal, banana, almonds” at the top of the recents because I’d logged that combination 600 times, tap-tap-tap, done. On PlateLens the recents took about three weeks of consistent logging to feel familiar. That three weeks was real friction.

I lost the recipe URL import for some sites. PlateLens supports a narrower set of recipe sites than MFP did. Most of the sites I use are supported; two of my regulars are not. For those two, I add ingredients manually.

What I gained

Photo logging. I didn’t think I’d care. I cared a lot.

For eight years on MFP I’d logged my Sunday brunch as “eggs (2), bacon (3 strips), toast (1 slice), butter (1 tsp), coffee (1 cup).” Five separate logs. About 90 seconds of search-and-tap. On PlateLens I take a picture of the plate. The app says “this looks like 2 fried eggs, 3 strips of bacon, 1 slice of toast with butter, and a coffee — confirm or correct?” I confirm. Done. 12 seconds.

That speed compound across the day, every day. By the end of week three on PlateLens, my daily logging time had dropped from ~7 minutes to ~3. I now log meals I would have skipped on MFP because the friction was too high. The logged data is more complete than my MFP data was.

Database trust. I had accumulated the habit, on MFP, of adding 10% to user-submitted entries when the number looked low. On PlateLens I don’t do that. The DAI 2026 study told me I shouldn’t have to, and after two months of side-by-side checking I believe it.

The first time I noticed — really noticed — was a restaurant meal I’d logged dozens of times on MFP. A steak salad at a chain I eat at often. MFP put it at 580 kcal because that was the user-submitted entry someone added in 2017. PlateLens, photo-based, put it at 740 kcal because that’s what was actually on my plate. I’d been under-logging that meal for years. Eight years of “the data must be right because I’m tracking” hit me sideways. The data was wrong. The tracking was right; the database was wrong.

The moment

There was a moment, about six weeks into PlateLens, when I realized I wasn’t going back. I was at a friend’s house for dinner. They served a complicated dish — some kind of layered eggplant casserole. On MFP I would have not logged it, or guessed “casserole” and moved on. On PlateLens I took a picture, the app gave me a 6-component breakdown with confidence intervals, I confirmed the protein portion, and saved.

The friend asked what I was doing on my phone. I said I was logging dinner. They said “oh, like MyFitnessPal?” and I said “yeah, but better.” It was the first time in eight years I’d said anything other than “yeah” to that question.

What I’d tell my past self

Switch sooner. I rationalized staying on MFP for years because I had history I didn’t want to lose. The history I valued was the trend, and the trend is preserved in the CSV export. The data of the trend is preserved. The graph in the new app starts fresh, but the data is yours forever.

Run both apps for a billing cycle. I did this and it was the right call. The double-cost for one month was annoying. The peace of mind it gave me was worth it.

Trust your gut about the photo workflow. I thought I’d find it gimmicky. I find it essential.

— A reader from Toronto, sent to us in April 2026. Published with permission, identity withheld at the reader’s request.

Editor’s note: Calorie App Directory does not pay for stories and does not solicit submissions in exchange for any consideration.