# Evidence-to-email worksheet

Original Cerrito examples, October 2, 2026. Hypothetical and untested. These are message bodies, not complete sending/compliance instructions. Replace placeholders only with facts you can verify. Do not send until identity, audience, opt-out and sending requirements are checked. No outcome claims or recommended sending volumes.

## 1. A hiring signal with an operational question

Observed fact: a current company-owned job description assigns manual customer-report preparation to an operations hire. Save URL, exact relevant text and observation date.

Unverified inference: report preparation consumes significant time. A job listing alone does not establish hours lost, urgency or available budget.

Useful deliverable: an example report mapped to the advertised workflow, made with synthetic data.

Subject: the reporting workflow in your ops role

> Your operations listing includes preparing customer reports. I build software that assembles those reports from [supported source]. I made a small example using synthetic data against the workflow in the listing. Would the sample be useful to whoever owns reporting?

Send only if you actually built the sample and support the named source. Skip if the listing is closed, reporting is incidental, or the sample cannot demonstrate relevant work. Never replace the question with an invented time-saving figure.

## 2. A public integration request

Observed fact: a company publishes a current request for an integration in its own public roadmap or forum. Record source/date and whether the request is still open.

Unverified inference: the company would adopt an outside solution. It may prefer to build internally.

Useful deliverable: a working connector demonstration with clear scope and constraints.

Subject: [system A] to [system B] request

> I found your public request to connect [A] with [B]. We have a working connector for [specific operation]; it does not yet handle [important limitation]. I can send a short demonstration with sample data if the request is still active. Would that help?

Skip if you cannot perform the operation, the request is already resolved, or the evidence came from a private community you are not authorized to use. Do not call a prototype production-ready.

## 3. A real implementation decision

Observed fact: the company has publicly announced a migration that touches a workflow your product supports. Record the actual announcement, not an enrichment guess.

Unverified inference: the migration creates a gap your product should fill. A change of tool is not evidence of dissatisfaction.

Useful deliverable: a migration checklist for the specific integration, with known failure modes and test steps.

Subject: [workflow] during the [system] migration

> Your announcement mentions moving [workflow] to [system]. We maintain an integration for that setup and have a checklist covering [concrete check] and [concrete check]. Would you like the checklist for your implementation team? It is usable without our product.

Skip if the checklist is a generic brochure, it depends on buying your product, or you cannot verify the migration. Do not promise an unbiased audit while concealing that you sell the solution.

## 4. Recruiting a design partner honestly

Observed fact: a company publicly describes a workflow problem relevant to a prototype you are building. A technical blog describing an old problem is not proof that it is still unresolved.

Unverified inference: the team has time to help develop a new product. Research participation and purchase intent are different.

Useful offer: a narrowly scoped design-partner session, with the time commitment, unfinished parts and what the participant receives stated upfront.

Subject: a prototype for [specific workflow]

> Your post describes [specific workflow constraint]. I'm building an early prototype for that problem. It currently handles [working capability]; [missing capability] is unfinished. I'm looking for feedback in one 20-minute session using sample data. Would that be relevant, or is this already solved for your team?

Skip if you need production access, cannot demonstrate the stated capability, or intend to turn an advertised feedback session into an undisclosed sales demo. The 20-minute limit is a promise to honor, not a proven conversion tactic.

## 5. A distribution partnership with a real contribution

Observed fact: the prospective partner publishes resources for an audience your company can help with a distinct task. Audience overlap does not establish permission to contact their subscribers.

Useful offer: an original, usable asset that complements one named resource and can be assessed without a meeting.

Subject: a worksheet for your [resource]

> Your guide covers [specific decision]. We made a worksheet that helps readers do [adjacent task], including [limitation]. Would you like to review it as a possible companion resource? There is no signup required to use it; we would agree attribution and placement before anything goes live.

Skip if the asset is gated contrary to that promise, merely repeats their guide, or the proposed partnership depends on sharing personal contact data without permission. Do not offer a backlink swap disguised as reader research.

## 6. A useful follow-up after an unanswered offer

Observed fact: you sent a specific offer and have not received a reply. Silence tells you nothing reliable about whether the person read it or why they did not respond.

Useful deliverable: one concrete clarification or new piece of evidence. If nothing useful changed, another message may not be justified.

Subject: Re: [original subject]

> One clarification on the [asset] I mentioned: it works with [relevant constraint], and the example uses sample data. Here is [a permitted public preview]. If the problem is already handled, no action needed.

Skip after refusal or opt-out, when your previous claim was inaccurate, or when the only addition is “bumping this.” Do not manufacture urgency or claim the recipient opened the first email. This is an example of a substantive follow-up, not a recommendation for a fixed sequence length.

## Before sending any example (all six situations)

Write down the evidence separately from the inference. Deliver the promised asset without adding an undisclosed meeting requirement. Use your actual role and product capability. Stop on a refusal or opt-out. A funding announcement by itself does not establish a buying need.

## Measurement worksheet

Keep comparable audience and observation windows. Record delivered messages, distinct human replies, positive replies, meetings booked and opt-outs separately. Define each before a test. Deduplicate people; exclude automated replies from human replies. Use counts as well as rates. Tiny samples are diagnostic notes, not winning-template evidence. No results exist for these examples.


## Blank evidence card

| Field | Your evidence |
| --- | --- |
| Source URL and observed date | |
| Exact observed fact | |
| Unverified inference | |
| Useful asset ready to deliver | |
| Who plausibly owns the problem? | |
| Condition that means skip | |
| Draft message | |
| Decision and reason | |

## Results log

| Audience / dates | Delivered | Human replies | Positive replies | Meetings | Opt-outs | Definitions / exclusions |
| --- | --- | --- | --- | --- | --- | --- |
| | | | | | | |
