The rejection came, and I won't lie. I was disappointed. I'd spent many hours tweaking, building, learning in the hopes that I'd get a chance to take the giveready.org platform more seriously through a foundation grant. Their rejection notice was polite enough. There were many high-quality proposals. Priority given to those most closely aligned with the criteria. No per-proposal feedback. Nobody teaches you what to do after a grant rejection, so here is the version I am planning, one week in.
This is Day 138 and the last post was Day 126. I took twelve days off writing. The systems kept running without me, which is rather the point of building them, and it turns out one of the things they kept doing was quietly tracking the learning loop I had built but was not checking for the platform.
The application in question was for GiveReady, the donation engine I build for my son's memorial fund, the Finn Wardman World Explorer Fund. Back in April we applied to a foundation challenge on AI and charitable giving. The submission described a prototype that was two weeks old and proposed to spend a year measuring whether AI-assisted giving could reach small charities.
The decline is not data
The first move is to be precise about what a decline actually tells you. A form rejection from an oversubscribed pool carries almost no information about the work. The funder's email said as much: no feedback given, standard wording, an invitation to apply to future rounds. It's now early August and I've put in a lot of work on the platform. That work hasn't really paid off though, and product market fit hasn't shown up. Yet, and I'm planning to continue until it does.
The temptation is to rewrite your strategy, especially after a rejection. I have done this before with lost sales deals and regretted it every time. The only thing that changed on the day of the decline was my attitude. Disappointed, but not surprised. I knew that the competition would be tough, I'm one person doing this in my spare time.
Verify what you promised against what you can now prove
So what now? The first thing is to pull up the application you submitted and read it next to your current numbers. The April version of GiveReady was a promise: give us a year and we will measure. The August version is a result: one operator made unrated niche charities citable by a general answer engine, zero of ten frozen test prompts to nine of ten in 71 days, on roughly 50 dollars a year of infrastructure. The prompt list was written before the content existed and frozen since, so we cannot accuse ourselves of changing the rules. Anyone can re-run the prompts and check, which is something I do regularly for testing.
I wrote about the mechanics of the citation share in Allowed Is Not Crawled, and about the pivot that produced the metric in retiring a theory in public. Downstream is still zero. The guides drew 314 views in the last 30 days and not one arrived from an AI assistant. The only settled donation is a single dollar I sent myself to prove the rails work. I publish that next to the citation number because a learning loop that doesn't divulge the bad numbers isn't educating anyone.
File the next one while we can to not lose momentum
The third move is speed. Grant rejection has a half-life. Wait a month and the project quietly files itself under "we tried grants, they said no." So the week of the decline is the week to line up the next applications, while the material is fresh and the drive is still high.
Three are in the pipeline. An accelerator built for early-stage nonprofit tech, whose two hard requirements, a nonprofit model and a working MVP, we meet exactly; applications close in early September, so that deadline is now on the todo list. Builder grants from the ecosystem of the payment protocol GiveReady runs on; those fund the build rather than the organisation, which keeps the charity paperwork out of it entirely. And retroactive public-goods funding rounds, where you apply after shipping, with evidence of use instead of a written proposal. For a project whose whole approach is measuring in public, applying with a results table instead of a promises document is a natural fit.
There is also a pivot check built into this. The grant that declined us will eventually publish the projects it funded. I intend to study that funded list line by line. What did they actually pay for? Which categories, which geographies, which stage of maturity? If every winner sits in a category we are not in, that is real information, the kind the decline letter itself never told us about, and it feeds either the next application or an honest pivot. We have retired a theory in public before.
The learning loop is still measurement
If you are eyeing a nonprofit project, or any project that runs on other people's ideas of what has merit: a grant application is a hypothesis about what a funder values. One decline is one data point, no error bars, no feedback to go off. You cannot learn anything from a sample size of one except that sample sizes of one feel terrible.
A pipeline of applications is a measurement. File several, log the outcomes, read the funded lists, adjust. It is the same loop that took the citation number from zero to nine, pointed at money instead of prompts. That loop has been the whole campaign since Day 0: try the thing, write down what happened, let the table argue with your feelings.
The tracker writes its next row at 06:55 tomorrow morning.
Day 138 of 365.