Open source · for Engaging Networks

< Crescendo

Every donation page has a signal. Crescendo turns it up.

The open-source fundraising page optimizer for Engaging Networks. It syncs your donation pages, layers in GA4 and NetDonor revenue data, and asks the AI model you choose — cloud or local — what to improve.

View on GitHub How it works $ docker compose up

pp the quiet problems

If you run fundraising pages on Engaging Networks, you already know these.

They're quiet. They cost you anyway.

The revenue your analytics never saw

GA4 says one number. Engaging Networks says another. The gap is real donations your tracking missed — and nobody is measuring how big it is.

Your story is split across three tools

Conversion data in EN reports, traffic in GA4, the actual appeal on the page itself. Assembling one picture of one page is an afternoon. You have two hundred pages.

Nobody has time to audit every page

The pages that quietly underperform stay quiet, because reviewing each one by hand never makes it to the top of the list.

The tools that could help are priced for enterprises

Conversion optimization platforms assume enterprise budgets. Nonprofit digital teams rarely have one.

mp monitor

It watches your pages so you don't have to.

Crescendo connects to your Engaging Networks account and syncs every donation page. Then it layers in the numbers: traffic and conversions from Google Analytics 4, revenue from EN's NetDonor data. Every page, one dashboard — on the schedule you choose: on demand, hourly, daily, or weekly.

Crescendo dashboard listing fundraising pages with conversion rate, revenue, page views, and donation counts
All pages, one dashboard — conversion rate, revenue, page views, donation counts.

mf measure

It sees the page the way a donor does.

On deep scans, Crescendo scrapes each page's live content — appeal text, buttons, donation amounts — and captures desktop and mobile screenshots. It detects your payment gateway and digital wallets (Apple Pay, Google Pay, PayPal, Venmo), and flags console errors a donor would never report. Cloudflare-protected pages get a headless-browser fallback.

And it measures the gap: GA4 conversions against EN's donation records over the last 30 days, with an estimate of the revenue your tracking never saw.

Page detail view with campaign metrics, trend charts, and tracking accuracy indicator
Page detail — campaign metrics, trends, and the tracking-accuracy gap.

f optimize

Then it asks the model you trust what to improve.

Recommendations are grounded in your actual page content and conversion data — not generic best practice. Three surfaces, one goal:

01 · generate Recommendations

Per-page optimization suggestions, shaped by context profiles — an emergency appeal is not a mid-level campaign. Each arrives with a confidence score; dismiss it, or mark it implemented once you've made the change.

02 · explore Explorations

Curated queries you run against any page — traffic sources, mobile vs desktop, what changed since the last scan. Results are cached per page, with history, so you can compare runs over time.

03 · chat Chat

A free-form assistant that can see the page's content, metrics, and fundraising history. Conversations persist per page, so the thread picks up where you left off.

Under the hood, the AI reaches your data through tools — GA4 queries, page performance, site-wide comparisons, snapshot diffs, content audits, and web search — and shows its work in a tool activity panel you can expand, call by call.

Make it yours.

Every system prompt is editable — recommendations, explorations, and chat each expose theirs in AI Config. Write your own explorations in a full-page editor (reorder, toggle, delete them too), and set tool guardrails like a web-search domain allowlist.

AnthropicClaude
OpenAIGPT
GoogleGemini
Ollamalocal · your hardware

Assigned per surface — recommendations, explorations, chat — and switched any time.

AI Explorations panel with configurable query templates and a streaming result window
The Explore surface running a traffic-source breakdown, tool calls visible on the right.

sotto voce security

Quiet by design.

Donor data deserves a tool that shows its work. Every claim below links to the code that makes it true.

ff open source

Turn it up.

Free and open source. Running in your infrastructure in two commands.

git clone https://github.com/orkestre-ai/crescendo.git
cd crescendo && docker compose up

AGPL-3.0 · free forever · your data stays yours

Built by a 20-year nonprofit-sector veteran at Orkestre AI.