Skip to content

Drafting replies with rules

Writing rules that draft a reply for a matching inbox item, approving the draft, and which networks Social Studio can actually post the approved reply to.

Updated

A reply rule watches for inbox items that match a keyword, a kind or a sentiment, and drafts a reply from a template you wrote. Every draft is reviewed by a person before it counts as anything.

The template is substitution, not generation. Write {handle} and {platform} in it and those are filled in. No AI provider is involved and no credits are spent.

Sending depends on the network. Social Studio posts an approved reply on a network whose terms have been determined to permit it, through an account you have connected — one network qualifies today. On the other ten, nobody at easyMarketing has read their terms for this question yet, so the send is refused with the reason and whose decision it was, and the approved text stays on the page for you to copy and post yourself.

Nothing posts silently, on any network. A reply goes out only after a person has approved the wording and pressed Send, and the action gate records the decision.

BEFORE YOU START

  • The "reply" or "moderate" permission to create, pause, delete or approve, and to ask for a draft.
  • Access to the workspace.
  • Inbox items to match against, which means items somebody has recorded.

STEPS

  1. Open Reply rules for the workspace, or go to /app/workspaces/{workspace}/reply-rules.
  2. Add a rule: give it a name, a keyword to match, and the reply template.
  3. Choose the mode. "Suggest" drafts for review. "Auto (gated)" records a governance decision as well, and still leaves the draft for review.
  4. Pause or resume a rule from its row.
  5. Open the workspace inbox - the "inbox" link at the top of the page goes there - and choose "Draft replies" on the item you want answered. This is what makes a rule run; a rule does nothing on its own.
  6. Come back to Reply rules, read the suggestions table underneath and choose "Approve" on a draft you are happy with.
  7. Copy the approved text and post it on the platform yourself.

WHAT YOU SHOULD SEE

An approved suggestion, with its text on the page. There is no Send button, because there is nothing behind one.

WHAT THIS WILL NOT DO

  • It will not post a reply anywhere.
  • It will not run by itself. A rule is evaluated against ONE item, and only when somebody asks: "Draft replies" on an inbox row, or your own integration calling the same endpoint. Nothing watches the inbox and nothing evaluates a rule on a schedule, so an item nobody asked about has no drafts against it.
  • It will not generate wording. A template is filled in, not written.
  • It will not let you edit a rule after it is made. Only the paused or active state can be changed; the name, keyword and template are fixed.
  • It will not dismiss a suggestion. There is no dismiss control.
  • It will not do any of this on your phone. The mobile screen lists rules and nothing else.

WORTH KNOWING

The "auto" mode sounds like automatic sending and is not. It records a governance decision alongside the draft, and the governance policy for an action at that risk level asks for a human confirmation, which nothing in this product provides. The suggestion is still written as a suggestion.

IF IT DOES NOT WORK

  • "A rule needs at least one of kind, sentiment or keyword." A rule with no criteria would match everything, so it is refused.
  • "reply or moderate permission required."
  • "That rule could not be saved." and "That suggestion could not be approved." are the fallbacks.
  • "No rules yet." is the empty rules state. The empty suggestions state says the same thing this guide does: a rule does not fire on its own, so a draft exists only where somebody asked for one.
  • "No active reply rule matched this item, so nothing was drafted." means the request worked and nothing matched. That is different from it failing, and the page says so.
  • "reply or moderate permission required." also covers asking for a draft.
  • If you reach the send endpoint directly on a network with no determination, it answers 409 with "Nobody at easyMarketing has determined whether <network> permits posting a reply, so it has not been attempted." It used to answer 503 saying the capability did not exist, which told you the product was broken when the truth was that a question had not been asked.
  • "This inbox item was recorded inside Social Studio and carries no provider reference, so there is nothing on a network to reply to." means the item was typed in rather than read from a network.

COMMON QUESTIONS

Why build rules if most networks refuse the send?

Because the drafting and the approval are real on every network, and a consistent, reviewed answer to a recurring question is worth having ready whether the product posts it or you do. Where a determination is missing, the product names the network and says the gap is easyMarketing’s rather than implying you have something to fix.

Does a rule cost credits?

No. Templates are filled in locally.

  • Triaging your inbox

    What the unified inbox holds, which networks its comments can actually be read from, the difference between an empty inbox and an unread one, and why empathy analysis always refuses.

  • WhatsApp flows

    Setting up trigger-and-reply flows for WhatsApp, how inbound messages actually reach the inbox, and the two separate reasons nothing is ever sent.

  • The contacts behind your inbox

    How the social CRM builds itself from recorded inbox activity, what you can add to a contact, and the things it deliberately does not do.