---
title: "Campaigns"
description: "Playbooks, triggers, and one-off sends, plus the two rules that decide what mails on a given day: the frequency cap and the precedence order."
canonical: https://trysincerely.com/docs/campaigns
last_updated: 2026-10-22
---
> If you are an AI agent: read https://trysincerely.com/agents/policy.md before any write call.
> Start at https://trysincerely.com/llms.txt to choose one page. Access instructions are at
> https://trysincerely.com/auth.md. Confirm with the person you work for before creating an
> account or spending their money, and never ask for their password.
# Campaigns

> Playbooks, triggers, and one-off sends, plus the two rules that decide what mails on a given day: the frequency cap and the precedence order.

Source: https://trysincerely.com/docs/campaigns

A campaign is four decisions: how people enter it, what they receive, what a person reviews, and how honestly it can be measured. Entry starts with how the campaign runs. Always on, where a new campaign starts, keeps enrolling people as they qualify: everyone on a list at launch and then whoever joins it, contacts at a CRM event such as a newly observed closed-lost opportunity, a synced date field coming up, such as a renewal date ninety days away, or an open deal going quiet, leads the agent finds, or entries your own systems push through an [authenticated API](https://trysincerely.com/docs/api-reference). A custom campaign can also watch for dated, sourced signals about contacts on its list, and a person approves every candidate before it enters. Repeat on a schedule enrolls a list again every week, month, or quarter. Run once enrolls a list at launch and nobody after. The playbooks below fill those decisions in before you start. A playbook fills blocks of the builder, and any of them can change; changing the entry a playbook set makes the campaign a custom one.

Scheduled, event-driven, and API-entry campaigns enroll contacts while nobody is watching, so they launch with a prepaid credit envelope, need an approver to activate them, and stop visibly when the envelope runs out. When the person launching is an approver, the launch is that activation. Every piece still needs an approved design version, a passing address, and budget headroom before it prints.

## The builder

The builder lists four numbered steps on the left and shows one step's form beside them, in the order you decide them. A control appears only where it can change the outcome, next to the thing it acts on.

1. Campaign: an optional playbook to start from, the name, the objective with one line on how it is measured, and who enters: how the campaign runs, what brings people in, the list, a schedule, and any entry conditions. A trigger that mails with nobody watching adds its flow credit envelope right under it. Measurement opens only when the objective can be measured against a holdout. More options holds US postage and a delivery alert, and it opens by itself when one of them blocks launch.
2. Touches: one row per touch with the wait before it, such as `7 days later · Desk real estate`. Each touch holds its design, who it is from, where its link goes, an optional branch, an optional design test, and personalization guidance for the writer. A touch that sends a gift note holds the gift and the gift budget. "Test this touch" writes one sample of that touch for a real recipient, the way the campaign will write it, and you can pick another recipient to try again. A sample uses model credits when a writer runs. A new campaign starts with one touch; "Add a touch" adds another, up to five. Once there are two touches, "When to stop" appears under them.
3. Voice and review: how personal the copy is, the personalization guidance for every touch, whether the landing page gets its own opening, and how drafts are approved.
4. Review: one line on what prints without your approval, then six questions with their answers: who gets it, what they get, when (a timeline from launch to arrival), what it costs against your credit balance, how you'll know it worked, and when it stops. Each answer you can change has an Edit link to the step that sets it. Launch locks who gets it, what they get, when, and how it is measured; to change one afterwards, duplicate the campaign.

The open step is filled in. A gold dot marks a step or a touch that needs attention before launch, and a step you have opened with nothing left to fix shows a check. The button at the bottom right leads to the next step. The bar beside it shows the estimate and names the first problem that blocks launch, with a link to the step that has it. On a phone, the steps become a menu above the form.

Above the steps, "Build it with the assistant" takes a description of the campaign and opens the assistant on this draft. When the assistant saves the draft, the builder shows its version. If you have unsaved changes at that moment, the builder keeps them and asks whether to take the assistant's version or save yours as another draft. [Writing and design](https://trysincerely.com/docs/writing-and-design) covers personalization guidance and how drafts are reviewed.

## Where each touch's link goes

Touch 1 chooses where the campaign's link goes: the house page, a hosted page you published, or your own site. With your own site, the touch shows the exact URL a scan opens and the analytics parameters added to it. Every later touch opens the same page as touch 1 unless it picks another hosted page. A route belongs to its design, so a design that prints on two touches opens the same page on both. A gift note prints no link, and its touch says so.

"Create a page for this touch" saves the draft, opens the landing page editor for that touch's design, and brings you back to the same touch with the new page selected once you choose to use it.

## Who each touch is from

"From" on touch 1 chooses who signs the campaign: one of your senders from Settings, or, when a CRM is connected, "A person in your CRM". A person in your CRM signs each piece with the name in one field of that recipient's record. The field list puts the contact owner and the account owner first, then every other field that points at a CRM user, then text fields. Email fields are never offered. When a recipient's field is empty, holds an address, names an inactive user, or the CRM cannot be read, the sender under "If it names nobody" signs that piece instead.

Every later touch is from the same person as the touch before it and says so, such as `From Robin Vale, same as touch 1`. "Change" on a later touch picks someone else for that touch and the touches after it, and "Use touch 1's" puts it back. The sidebar names the signer on a touch that is not signed like touch 1, and Review names it under "What they get".

The writer signs with the name where the format has a sign-off, such as a letter, and follows the sender's voice notes when one of your senders signs. To steer the voice, write it into the personalization guidance, such as "write the way Dana would: short, plain, no jargon". The name read from the CRM is kept with the draft, so the review queue, the gift note, and a reply to the landing page all name the same person. The return address on postcards and letters comes from your brand kit, whoever signs. A touch from someone else is written on its own, one writer run per touch, because copy written once is signed once. The writing setup, with its tone and its required and banned phrases, comes from Settings and applies to every touch.

## Workspace drafts

Campaigns lists the workspace's saved drafts with the last editor and edit time. These are unfinished campaign setups, not copy waiting for review. Copy that waits on a person shows as a `to review` chip on its campaign's row and in a line above the list; [Writing and design](https://trysincerely.com/docs/writing-and-design) covers reviewing it. The builder saves your work as you edit. Open a draft there to continue it, or ask the assistant to continue the same saved draft. If someone changes it while you are editing, reload their saved version before continuing.

New drafts start with the workspace's sender and writing setup from Settings, and with the outcome and baseline rate of your most recent campaign. Choose the list, approved designs, response destination, timing, and outcome in the campaign. Saving a draft does not launch it. Review the preview and confirm the launch before anything can send.

## The playbooks

| Playbook              | Trigger                                                                                            | What it sends                      |
| --------------------- | -------------------------------------------------------------------------------------------------- | ---------------------------------- |
| Find & Mail New Leads | The agent finds leads matching your criteria on your schedule                                      | One piece per new lead             |
| List Blast            | A list and one good reason                                                                         | One or two touches, once           |
| Outbound Air Cover    | Contacts enter your outbound sequence                                                              | Three postcard touches, always on  |
| Enterprise Drip       | Named accounts, whole buying committee                                                             | Steady touches, capped per account |
| Meeting Follow-Up     | You just met them; you pick the list                                                               | A personal note                    |
| No-Show / Revival     | You pick the no-show or revival list                                                               | A note to every stakeholder        |
| Renewal & Expansion   | Ninety days before a renewal date                                                                  | An executive note                  |
| Closed-Lost Win-Back  | An opportunity is newly observed as closed-lost                                                    | A letter on the transition         |
| Deal Quiet Revival    | No observed CRM movement (open or closed state, stage, amount, close date) for the days you choose | A note when the silence starts     |

The measurement policy is derived from the campaign's entry design and the holdout, and fixed at launch, never chosen after the results are in. An entry that can be randomized by account, including a one-shot blast over a list and a scheduled cohort, starts with an account-level holdout and gets a lift estimate with an interval. The holdout is optional: when the point is to mail everyone on the list, choose no holdout, and the campaign mails every account and reports descriptively. A Keep in touch campaign, which asks for nothing, starts with no holdout and holds accounts out only when asked. Hand-picked recipients, person-triggered follow-ups, agent-sourced leads, and entries pushed by outside events always report descriptively and say so, rather than dressing a descriptive number as a causal one.

Two more exist and are not offered as general campaign types, because both carry their own invariants: the first send, which is one piece to one hand-picked person, and the one-off send, which is a person clicking Send.

## Objectives Sincerely can see

Two objectives need no CRM: a QR scan, which is a person opening the printed link, and a landing page action, which is a button click or a form sent on a hosted page. Sincerely records both itself, so a campaign judged on either reports as soon as mail lands. A holdout account receives no piece and so can never scan one, which is why these objectives report rates with intervals and design tests between arms, never a lift against the holdout. Choose a CRM-reported outcome when the question is whether mail changed what people did afterwards.

## Reporting a conversion back

When the response link leaves for your own site, the destination URL carries `sid`, a conversion id that names the piece without exposing its landing page. Your system reports what happened next by posting the `sid` and a kind to `/api/v1/conversions` with a workspace API key, or by importing a spreadsheet of them. Each report becomes an outcome on the account that received the piece, and attribution counts it as a direct response. A held-out account has no printed link to report through, so a postback stays out of the lift estimate for a CRM-reported objective; the readout shows it beside the estimate as context. It does count toward a landing page action objective, which reports a rate rather than a lift. A campaign judged on a landing page action can also name one action on a hosted page, so a form sent counts and a secondary link clicked does not.

## Scheduled entry

A custom campaign can enter contacts on a schedule: every week on a weekday you pick, every month on a day from the 1st to the 28th, or every quarter on that day of January, April, July, and October. Dates are UTC, and the first firing is the first matching date on or after launch. Each firing enrolls every contact on the list who can enter and is not already in the sequence, so a contact who finished an earlier cohort can join a later one. A retried tick never enrolls the same firing twice, and the campaign record gets one line per firing, such as `scheduled entry · 2026-10-05 · 42 entered`.

A scheduled campaign needs a flow credit envelope, because every firing after launch prints mail with nobody watching.

## Paced entry

An always-on list can let accounts in at a set pace instead of all at once: at most so many accounts a day, a week, or a month, such as 1,000 accounts a month from a list of 100,000 contacts. The window rolls, so a month is the 30 days before each daily run. The first batch enters at launch, and each daily run admits accounts until the window holds the pace. Holdout accounts count toward it, because an account enters before its arm is assigned. Accounts enter in the order of a stable hash of the account and the campaign, so every batch is a fair draw from the list and a retried run admits the same accounts. Everyone else waits, and a contact who joins the list later waits in the same draw. The pace is locked at launch.

A paced list spends from a flow credit envelope. A person watches its launch batch, but nobody watches the batches after it, so the envelope has to cover the first batch, the builder says how many batches it covers, and the campaign pauses before a piece would cross it. Unlike a scheduled or event-driven campaign, it needs no approver's activation: an always-on list without a pace already admits whoever joins it with no envelope at all, and a pace only narrows that. The campaign page shows how many contacts are waiting their turn.

A rolling window is not the billing month. A batch that enters near the end of one month frees its slots near the end of the next, so one billing month can take two batches. Review's plan-limit estimate counts that.

## Entry limit

A run-once or scheduled list can cap how many accounts enter each run. Run once with a limit is a trial batch: a fair draw of up to that many accounts enters at launch, holdout included, and the rest of the list is not mailed by that campaign. To work through a whole list, use an always-on list at a set pace instead, which keeps one campaign, one holdout, and one readout. A scheduled list with a limit draws on each firing. Accounts that have not entered the campaign go first, then the ones that entered longest ago, and the stable hash breaks ties, so the campaign works through the list and then starts over. A retried firing admits no second batch. The limit counts accounts, the same unit as the pace and the holdout, so the pieces it prints depend on how many contacts each account has; Review shows the estimate. The limit is locked at launch.

## Entry conditions

Under "Only if", a campaign can require up to five conditions, and a contact enters only when every one holds. A condition tests the contact's title, state, country, or pieces mailed, or the account's name, domain, employee count, or company campaigns. Pieces mailed counts the pieces a print vendor has accepted for the person, across every contact record that shares their email; a piece still queued for approval does not count. Company campaigns counts every other campaign anyone at the account has entered, in either arm and including one-off sends, so `company campaigns is 0` keeps a campaign away from accounts in another campaign's control group. Text comparisons ignore case and surrounding spaces. A missing value fails the condition, including "is not" and "does not contain", and a contact with no country on the address counts as US. The conditions are applied before the holdout split, so they keep both arms out alike.

A campaign that enters by API can also test the event's properties. An entry event posted to the [entry endpoint](https://trysincerely.com/docs/api-reference) can carry a `properties` object of up to 20 keys. A key is lowercase letters, digits, and underscores, starting with a letter, at most 40 characters. A value is a number, true or false, or text of at most 300 characters. A bad shape gets `422 invalid_properties`, and the detail names the key, never the value. A condition such as `event plan tier is not free` reads those values, and the writer receives them as context for why this person entered. An API event refused by a condition reports the skip reason `conditions_not_met`. A scheduled firing counts its refusals on the firing's record line.

## Waits, guidance, and branches

Every touch after the first waits 1 to 90 days after the touch before it. Each touch can carry up to 280 characters of personalization guidance that the writer follows for that touch, on top of the guidance for every touch. Nobody chooses how often copy is written. When a touch after the first has its own guidance, is from someone else, or the sequence mixes formats such as a letter and a postcard, each touch is written on its own, one writer run per touch. Otherwise copy is written once per recipient with touch 1 and replayed, and touch 1's guidance steers every touch.

Touches after the first can wait on what the recipient did. "Only send when" sends the touch if they scanned an earlier piece of this sequence, or responded on its hosted page, or the reverse. Otherwise the touch is skipped or prints another approved design. A skipped touch sends nothing, and the next touch keeps its schedule. A branch cannot sit on the first touch or on a touch that tests two designs, and it needs a design in the campaign with a response link. A scan is any visit to the hosted page from the printed link; a response is a button or form on that page.

The branch is decided once, when the touch comes due or earlier when its copy is written for that touch, and the decision is stored. The copy a person reviewed and the design that prints therefore agree, and a scan that arrives later cannot change it. The recipient's record shows the outcome, such as `touch 2 skipped · they scanned`. Holdout accounts go through the same sequence with nothing printed. They receive no pieces, so every branch resolves for them as not scanned and not responded, which keeps the comparison between the arms intention-to-treat.

## Testing two designs

Any touch without a branch can test two versions, A and B, and several touches in one sequence can each run their own test. Each tested touch has its own split, the share of treatment accounts that get A, from 10 to 90 percent. The builder offers 50/50, 60/40, 70/30, 80/20, and 90/10, with a swap that exchanges A and B. The assistant and MCP take any whole percent in that range. The holdout is drawn first, so a split divides only the accounts assigned to mail, and the campaign holdout still measures lift. Whether an account gets A or B on a touch comes from a stable hash, so the account keeps its version on that touch, and tests on two touches split accounts independently of each other. The pairs, the splits, and the number of tested touches are locked at launch.

A and B can differ in anything a design carries, such as a postcard against a letter, a printed piece against a handwritten note, or a gift against no gift. They can also print the same design with different personalization guidance, which tests the writing alone. Leave B's guidance blank and B uses A's. Both versions share the touch's sender and its wait. If only one of the two prints a response link, the builder, the launch preview, and the readout warn you, because responses through the page count as outcomes and only the linked design can collect them. The test is still allowed, since "does a QR code help?" is a fair question. They also warn when only one version is written by hand: handwritten pieces wait on writer capacity and can pause while the printed version keeps mailing, so that version may reach fewer people than its share.

Each test is measured from its own touch. It counts the treatment accounts that were still enrolled, and had not yet reached the objective, when that touch's copy could first be seen. For touch 1 that is when the account enters, so a touch 1 test counts every treatment account. For a later touch it is a few days before the touch comes due, when its copy can be written ahead for review. Outcomes count for the campaign's observation window, starting on the touch's due date. The campaign's Sequence tab and its report show one block per tested touch, such as `Touch 2 · 70/30`, with the assigned and matured accounts for A and for B and each outcome rate with a range. The verdict says when there are not enough matured accounts to separate A and B, and neither is called ahead while any account is still inside its window.

When more than one touch is tested, each range widens so that the chance of any test showing a difference that is not there stays at 5 percent. This is a Bonferroni correction. Two tests read at 97.5 percent instead of 95, and the readout labels every range with its level. The same level decides whether a test has enough accounts, so each extra tested touch means every test needs more accounts to reach a clear result.

## What ends a sequence

A sequence ends after its last touch. With two or more touches, "When to stop" under the touches is one checklist for the whole sequence: stop sending to a contact when a meeting is booked, a reply is logged, or an opportunity is created. All three are on by default and each can be turned off. "They scan the code" stops the sequence after a scan; when the objective is a landing page action it reads "they act on the page" and stops after a button or form on the hosted page. That one is off by default and needs a design that prints a link. A campaign that keeps reading a live list (an always-on list, a schedule, or a CRM moment) also offers "they leave the list", on by default. List Blast, Find & Mail New Leads, and custom run-once campaigns read their list once, so leaving it changes nothing.

A one-touch campaign shows no checklist and launches with the defaults. They still matter: a piece waiting in review is cancelled when a meeting, reply, or opportunity is logged first. Suppression always stops everything at once, whatever the checklist says.

## The frequency cap

The default cap is three pieces per contact per thirty days. It exists because the channel's whole advantage is that the mailbox is not full, and a team that mails the same person weekly spends that advantage in a month.

The cap counts every piece to that contact, whatever campaign it came from. A one-off send both counts toward the cap and can be blocked by it.

## Precedence

When two campaigns want the same contact on the same day and the cap only has room for one, precedence decides. From highest to lowest:

1. One-off sends. A person clicked Send, and explicit human intent outranks every scheduled touch.
2. Meeting Follow-Up, No-Show / Revival, Deal Quiet Revival and Renewal & Expansion. Relationship moments, where the timing is the point.
3. Outbound Air Cover, Closed-Lost Win-Back, and custom campaigns with CRM or API entry.
4. List Blast, custom one-shot campaigns, and scheduled campaigns.
5. First Send, Enterprise Drip and Find & Mail New Leads.

Each campaign freezes its rung at launch. Campaigns on the same rung tie. A lower-precedence touch slips to the next eligible day and the slip is written to the record, so a piece that arrived late has a reason attached rather than a gap.

A one-off send is never slipped. Its pre-flight refuses instead, and names both the reason and the next eligible date, because a person waiting on a Send button should be told rather than quietly rescheduled.

## Launch and pause

Launch is the moment a person commits the audience, design, rules, and prepaid spend. Before that click, preflight lays out the receipt: how many pieces, how many credits, available credit and gift balances, remaining gift budget, and whether the audience can detect the lift the team cares about.

When an agent asks to launch, the confirmation card also says how copy is reviewed: "Review each piece", where every draft waits for a person, or "Auto-approve clean drafts", where drafts a model writes still wait until 10 pieces on this campaign are approved. A draft with an unsupported claim or stand-in text always waits. If the person confirming can publish designs, the same click publishes the draft designs the card names, before anything else runs, so they stay published if the launch then stops. [Team](https://trysincerely.com/docs/team) covers the approvals a launch can create.

Pause stops the next piece, not the postal system. Anything still queued remains under Sincerely's control. Anything the print vendor already accepted may still arrive. The [suppression](https://trysincerely.com/docs/suppression) page shows where that handoff occurs.

Launch also locks the experiment's primary outcome and measurement window. Changing either after seeing the numbers would turn a measurement into a story chosen after the fact.

The entry, its conditions, the sequence with its waits and branches, and the exits are sealed at launch too. To change them, duplicate the campaign. The personalization guidance for every touch and the landing page opening stay editable and apply to copy written after the change.

## One-off sends stay out of campaign surfaces

A one-off send has a campaign record behind it, but it stays out of campaign browsing. You won't see it in the campaigns list, Overview, per-campaign analytics, readout creation, the assistant's campaign list, or the command palette.

The piece still counts. It appears in the Send ledger and contributes to workspace mail activity, gift-budget use when applicable, engagement, and attribution.

Why hide the campaign record? A hundred personal sends shouldn't bury the standing campaigns on every listing screen.

## What to read next

[Sending](https://trysincerely.com/docs/sending) covers what happens after launch. [Measurement](https://trysincerely.com/docs/measurement) covers holdouts and readouts. The [playbooks](https://trysincerely.com/playbooks) pages are the ready-to-run versions with the trigger already chosen.

---

Sincerely is personalized direct mail and gifting for B2B revenue teams: postcards, letters, handwritten mail, and gifts, each written for one recipient.

Contact Sincerely: https://trysincerely.com/contact

Agent routing index: https://trysincerely.com/llms.txt

Listings: https://www.g2.com/products/sincerely and https://www.producthunt.com/products/sincerely