Dishcount
Turn grocery deals into weekly meals
I'm Bill, a Veteran in Dayton, Ohio with a normal day job and no software background. For the past 6 months I've been building Dishcount (dishcount.co) at night: a free site that pulls weekly grocery ads from 80+ US chains and turns the sale items into dinner recipes with real prices.
The origin is unglamorous. Grocery prices kept climbing and I was spending every Sunday night cross-referencing three store ads to plan dinners my family could afford. I figured software could do that work automatically. I couldn't code, so I built the whole thing by directing AI coding tools, spec by spec, and verifying everything against production because the tools lie to you sometimes. Monthly operating cost is under $40.
Where it actually stands, real numbers:
The product works. Deal extraction runs on cron every Wednesday, recipes generate from live local prices, and Kroger shoppers can push a whole recipe to their cart in one tap.
The business barely exists yet. Week-1 retention has run 1 to 3 percent for most cohorts. My first newsletter went out this week to 59 subscribers. My site is indexed by Google and gets essentially zero non-branded impressions because the domain has about three backlinks. I paused paid ads because retaining users I paid for at 2 percent made no sense.
What I'm doing about it: newsletter as the retention spine, a content sprint aimed at long-tail searches nobody owns ("kroger weekly ad meal plan" type queries), and slowly earning links. A competitor in an adjacent niche went from zero to owning category search in about 90 days of publishing, so the playbook exists.
Happy to share anything: the OCR pipeline for ad extraction, what directing AI tools solo is actually like, the compliance gray zones of deal aggregation, or the retention numbers in detail.
Question for anyone who's been here: when retention is the bottleneck and traffic is too small to measure fixes against, how did you sequence it? Grow traffic first to get measurable cohorts, or keep iterating retention blind?
4 Comments
This could be a very solid idea. One suggestion that I have no clue how to handle. Figure out a way to have users enter their current store saver card info. I did find one issue. I'm near Philly a bit far perhaps its causing issue. ZIP 18914 returned incredients from local GIant market, but "deals" were from Giant Eagle, unfortunately none near me. That being said I really think there is value here, which I could offer any quick fixes or solutions. Have a screen shot if you'd like, but don't know how to post it. either way keep up the good work.
The interesting challenge isn't generating recipes from grocery deals—it's becoming part of how families decide what to cook before they decide where to shop. I'd keep validating whether users return because Dishcount saves them money or because it removes the weekly planning effort. Those are very different habits to build a business around.
Just agreeing. if you could almost build it backwards if I"m catching your drift. Instead of saying well what to you want.... it depend s on what you want, you just click a button, get meals, and if you like them, select find deals at my store. Bill you figure that out an'd I'm in and would be linking to the site I just built. Best wishes to all.
Interesting way to think about the workflow.
The key question I'd keep validating is which moment creates the habit — deciding meals first, or finding savings after the decision is already made.