You had a great dish. You meant to remember it. Months later the restaurant is on the calendar again and all you have is a dim photo of a plate and a vague feeling that it was "the pasta thing" or "the spicy one." Friends ask what to order. You shrug. Someone picks wrong. Dinner is fine, not great, and you leave wishing Future You had been kinder to Present You.
Remembering restaurant dishes is harder than remembering grocery products. The name might be in another language. The plate looks different at night under restaurant lighting. The same dish name at two places is not the same food. And the memory you need is usually a bundle: which restaurant, which visit, which dish, and whether you would order it again.
Purchase History from StoicSoft treats that bundle as a first-class workflow: log the place, attach what you ordered, snap the plate if you want recognition help, and keep a rating in your own words. This guide covers why camera rolls fail for dining memory, how a visit-centered approach works, and how to use it on a normal night out without turning dinner into admin.
Why camera-roll food photos are not a memory system
Food photos are easy to take and hard to reuse. They sit sorted by date between selfies and screenshots. They rarely include the restaurant name in a searchable way. They almost never include your honest rating — especially when the dish was only okay and you did not want to be rude in the group chat.
Maps apps remember that you visited a place. They do not remember that the duck was perfect and the salad was a miss. Notes apps can hold a list if you type it before dessert arrives. Most people do not. Review sites remember the crowd's opinion, not yours.
What you need on the next visit is narrower: your dishes, your scores, your "order again" or "skip" notes, tied to that restaurant.
A visit-first way to remember dishes
Purchase History's Places tab is built around visits rather than orphaned plate photos.
- When you arrive, allow approximate location if you want nearby suggestions (about 150 meters), or search the restaurant by name.
- Start a visit. You can rate the place and add notes. The visit stays current for about three hours so orders and photos attach while you are there.
- Add what you ordered: type a dish name for free, or photograph the dish for AI recognition (one analysis credit on a successful description).
- Optional: scan the menu (photo or PDF) to pull dish suggestions for that place — one analysis per scan; the menu file is not kept after extraction.
- Add a rating and a short review while the taste is still obvious.
Menu items stay matched within their place, so the same dish name at two restaurants never merges, and a restaurant dish never collapses into a grocery product. That separation is what makes dining memory trustworthy.
Place lookups use approximate coordinates only for the nearby request — coordinates are not stored as a tracking trail. You choose the venue; the visit is your record.
Scenario: the neighborhood place you half-remember
Three months ago you loved a dish at a neighborhood restaurant. Tonight you are back with friends. As you sit down, you open Purchase History on your phone — the installable app you open in your browser, or the home-screen icon you added from Safari or Chrome — and pull up the last visit. Your note says the braised short rib was an 9 and the side was too salty. You order the rib again, skip the side, and suggest a different plate for the friend who wants "whatever you got last time."
Without that visit record, you would be scrolling a camera roll for a brown plate in bad lighting and arguing about whether it was short rib or lamb. The photo might still exist. The judgment would not.
What to log on a normal night out
Keep the bar low so you actually do it:
- Start the visit when you sit down or when the first plate arrives — whichever you will remember to do.
- Log the dishes you care about, not every shared appetizer by committee.
- Prefer a short honest note over a long review ("order again," "too rich," "ask for spice on the side").
- Photograph plates when names are unclear or when you want recognition help; type names when you already know them.
- Rate the place once if the overall night mattered — service, noise, and "would return" are useful next time someone asks.
If you are with people who share food decisions, Household sharing helps. Up to five members can share one purchase and visit memory, so the person who remembers the dish and the person who books the table are looking at the same history.
Menu scans without turning dinner into data entry
Menu scan is optional. Use it when the menu is long, the names are unfamiliar, or you want faster dish suggestions for that place. Photograph the menu or pick a PDF, let the service read dishes (section and price when available), then pick what you ordered. Each scan costs one analysis. Typed dish names stay free.
You do not need a perfect digital menu for every restaurant. A single confirmed dish with a rating already answers the question that matters later: what should we order when we come back?
Travel, takeout, and repeat favorites
Restaurant memory is not only for sit-down nights. Travel weeks produce the worst "what was that place called?" problems because you are out of your usual map. Logging a visit while you are still nearby — even with one dish and a rating — beats trying to reconstruct the trip from a camera roll when you get home.
Takeout and delivery benefit too. The "place" might be the restaurant brand you reorder from, and the dish notes still matter when the app suggests the same items again. If you share ordering with a household, one person's "too spicy for the kids" note saves another person's accidental rebuy.
Repeat favorites are where the system compounds. The third visit to a place is when vague memory usually fails. By then a short visit history with two or three rated dishes is enough to order confidently and introduce friends without the shrug.
Common mistakes to avoid
- Waiting until the next morning — ratings fade fast; capture while the taste is obvious.
- Logging every shared plate by committee — focus on dishes that will drive a future decision.
- Relying only on photos without notes — a plate image without "order again" still leaves you guessing.
- Merging restaurants that share a name in different cities — keep places distinct so dishes stay local to the venue you meant.
How this ties to the rest of Purchase History
Restaurant dishes sit beside retail products in the same product. Snap recognition, ratings, and lookup work for plates the way they work for bottles. Visits add the place layer grocery aisles do not need as often. Optional Codex MCP access is there if you want AI tools to query the same history under your Google-authenticated account — most diners will never need it, and that is fine.
Think of it as one memory for "things I tried" whether the context was a shelf or a table.
Plans for solo diners and households
- Personal — $1.99/mo: 50 analyses per month, 5 GB storage, 1 member.
- Household — $4.99/mo: 250 analyses per month, 25 GB storage, up to 5 members.
Photographing a dish during a current visit uses the usual analysis pipeline. Typing a dish name does not. Attaching a photo to a typed name does not add an extra recognition credit. Billing runs through Stripe; cancel or change plans in the Customer Portal.
Try it on your next reservation
Before the next meal out, skim the Purchase History landing page for places-and-visits docs, then open the live app and sign in with Google. At the restaurant, start a visit, add one dish you care about, and leave a one-line note. Next time that place comes up, you will not be guessing from an old photo — you will be reading your own dining memory.
Remembering restaurant dishes you loved should not depend on luck and lighting. Capture the visit, keep the judgment, order better next time.