## Overview The Follow-ups **Settings** layer holds the controls that govern **when** and at **what pace** your cadences send: timezone, quiet hours, per-contact limit, send throttle (anti-ban), business days and the defaults for each new sequence. Instead of repeating these in every sequence, you set them once and all cadences obey. Each value is resolved by an **order of precedence**, from the most general to the most specific. The right-most one wins: > **code defaults** ← **account default** ← **per-inbox override** ← **sequence** ← **step** In other words: the platform ships safe defaults; the **account default** adjusts everything at once; a **per-inbox override** changes just that channel; and, when it makes sense, a **sequence** or a **step** can refine further. This screen handles the two adjustable layers: the account default and per-inbox overrides. ## Prerequisites - The **Follow-ups** module enabled for your account. - An **administrator** user (only administrators can save settings). - For a per-inbox override: the inbox already created in the account. ## Step by step 1. In the **Follow-ups** module, open **Settings**. 2. Choose the scope: the **account default** (applies to every inbox) or a **per-inbox override** (pick the inbox in the scope selector). 3. Adjust the fields you need (timezone, quiet hours, limits, business days, defaults). 4. Click **Save**. The screen always shows the **resolved configuration** with precedence already applied, so you see the value that actually takes effect for that scope. 5. To send a field back to the account default, simply clear it in the override. ## Settings & options ### Timezone Sets the timezone used to compute waits, business days and quiet hours. There are two modes: - **Automatic (recommended)**: stores no fixed timezone — the platform resolves it in order: the **contact's** timezone (when the toggle below is on and the contact has one) → the **account** timezone (the same one reports use) → the **inbox** timezone → **UTC** as a last resort. The screen shows which zone "Automatic" currently resolves to (e.g. "Automatic — account timezone: America/Sao_Paulo"). - **Specific timezone**: you pin an IANA zone (e.g. `America/Sao_Paulo`) that applies to that scope, ignoring the automatic resolution. **Use the contact's timezone when available** (default **on**, account scope only): when a contact has their own timezone, it outranks the account default **for that contact's schedule only** — handy for audiences across zones. Turn it off to treat everyone in the account timezone. Sequences and steps can **refine further**: a step's **Scheduling** section lets it inherit the resolved zone or set a **timezone override** for that step alone, and pick a **fixed local time** (e.g. "1 day later, at 09:00" in the resolved zone). Without the right timezone, "21:00", "business days" and the fixed time are interpreted at the wrong moment. > **Schedule preview**: the sequence builder shows, per step, a preview of the next send already in the > resolved zone (e.g. "~Mon 09:00 · America/Sao_Paulo (account timezone)"), so you can check the effect > before publishing. ### Quiet hours A window in which **nothing is sent** — for example, **21:00 to 08:00**. It has three parts: on/off, **start** and **end**. When a send would land **inside** the quiet window, it isn't dropped: it's **pushed to the end of the quiet hours** (the time the window reopens). The window can **cross midnight** (start later than end, like 21:00–08:00) and the platform handles that correctly. Quiet hours are **off** by default. ### Per-contact daily limit A safety cap on **how many cadence messages the same contact can receive per day** (`rate_limit` → per contact/day). The default is **3**. It keeps several cadences combined from ever turning into spam for one person. ### Anti-ban send throttle Controls **how fast** a mass cadence dispatches, **per inbox/connection** (`send_rate`). You can set a ceiling **per second** and/or **per hour**. When volume exceeds the ceiling, the enrollments above the limit are **deferred and re-queued — never dropped** — and sent as soon as there's headroom. Blank fields use **safe provider-derived defaults** (for example, the WhatsApp Cloud per-second tier or the practical WhatsApp Web limit, around 240/h); `0` means **unlimited**. This is your main protection against a block on large sends. ### Hold while an agent is attending (Maestro) Available when the account uses **Maestro**. With **Hold the cadence while an agent is attending** on, if Maestro pauses because a human took over the conversation, the follow-up neither **enrolls** that conversation nor **sends** the next step of an existing enrollment. **Holding is not exiting.** The enrollment stays active: the platform only defers the next step and re-checks periodically — the **Re-check every (minutes)** field sets that interval (default: 15). As soon as the agent lets go, the cadence resumes where it stopped. What actually ends a cadence is still a reply, a resolution or a goal. The setting is **on by default** — holding is the safe default. A held cadence loses nothing: it resumes on its own the moment the agent leaves the conversation. Ignoring the pause, on the other hand, sends the automated message **on top of the reply the agent just wrote** — and that mistake cannot be undone. If your operation wants the cadence running even with an agent on the conversation, turn the setting off at the scope you want, or pick **Never hold** on the sequence. Precedence is the same as the other options: account → inbox → sequence. Each sequence can **Inherit**, **Always hold** or **Never hold** in its own settings. While an enrollment is held, the timeline shows the reason **"On hold (an agent is attending)"**. ### Business days Which days count as **business days** (default: **Monday to Friday**, i.e. 1 to 5). Steps whose wait is in **"business days"** skip weekends using exactly this list, preserving the time of day. ### Cadence defaults - **Default compose mode**: the initial mode of new steps (default: **static**, i.e. you write the text). It can be overridden per sequence/step. - **Max steps**: a ceiling on the steps actually sent per enrollment (default: **10**), a guard against overly long cadences. ## Use cases - **Respect business hours**: turn quiet hours on (for example 20:00–08:00) and set the timezone so you never message a customer in the middle of the night. - **Safe WhatsApp cadence (anti-ban)**: tune the per-second/per-hour throttle on the inbox so a large send drains at the connection's pace, with no risk of a block. - **Per-inbox default**: a billing channel with more conservative limits and a welcome channel with looser ones — each with its own override. ## Tips, limits & best practices - Get the **timezone** right first: it anchors quiet hours, business days and waits. - Keep the **throttle on** for WhatsApp inboxes and prefer to start conservative — it's better to drain slower than to risk a block. - Remember that blank throttle fields do **not** mean "unlimited": they fall back to the provider's safe defaults. Use `0` only if you really want to remove the limit. - A per-inbox override only changes the fields you fill in; the rest keep inheriting the account default. ## Troubleshooting - **A message went out at dawn**: quiet hours are **off** or the resolved **timezone** isn't your audience's ("Automatic" may be falling back to UTC because neither the account nor the inbox has a timezone). Turn quiet hours on and set the timezone — in Automatic mode, fix the **account timezone**; or pick a **specific timezone** for the right scope. - **A step didn't send and I don't know why**: open the enrollment **timeline** — each run shows the skip/failure **reason** (e.g. "Sequence not published", "No inbox for this channel", "WhatsApp 24h window closed", "Throttled (rate limit)", "On hold (an agent is attending)"), and the enrollment shows the failure reason at the top, never silently. - **The cadence stopped sending but the enrollment is still active**: most likely the Maestro hold — there is an agent on the conversation. It resumes on its own once they let go; if that isn't what you want, turn the setting off in the right scope or set **Never hold** on that sequence. - **Sends are "lagging"**: that's the **throttle** holding the pace. Enrollments above the ceiling are deferred and re-queued (never lost); raise the per-second/per-hour limit if you want to go faster. - **Business days weren't respected**: check the **business days** list and the **timezone** of the scope — only steps with a "business days" wait use this rule. - **I changed the account default but one inbox didn't change**: it probably has an **override** with that field filled in — the override beats the account default. ## See also - [Follow-ups and cadences overview](/hc/ajuda/articles/follow-ups-overview-en) - [Sequences, steps, channels and composition](/hc/ajuda/articles/follow-ups-sequencias-passos-canais-composicao-en) - [Cadence funnel and exit rules](/hc/ajuda/articles/follow-ups-funil-e-saida-en)