Inventory the questions you answer every week
Those become evergreen menus. If staff types it daily, automate it carefully.
HOURS, BOOK, MENU, PROMO — keywords are how customers drive your system without learning a new app. Reliable match → multi-step path → human exit when trust requires it.
Disjoint sets, evergreen vs campaign separation, escalation by design.
Those become evergreen menus. If staff types it daily, automate it carefully.
No two live automations should fight for the same words.
Campaigns expire. Menus should not die with them.
Location pickers, booking links, FAQ branches — complexity must earn its keep.
HUMAN, AGENT, or escalation rules for when automation stops.
Keyword automation is how local brands, support-leaning social teams, and campaign ops publish always-on entry points. Ranking intent: keyword ops — menu design, conflict avoidance, campaign codes vs evergreen trees, escalation. Different job from comment-to-DM spikes; different job from broadcasts.
The rules that shape this job
Official APIs onlyMatching is practical, not magic. Design for humans typing on phones.
Where this job goes wrong
One mega-bot with overlapping phrases double-fires and confuses customers.
Dead campaign codes left live break trust when someone messages last month’s promo.
No human exit turns a booking exception into a one-star review written to the bot.
Local businesses, support-leaning social teams, and agencies that ship evergreen menus clients can staff.
Channels this job touches
Bio and post codes
ConnectedPage inbox, same Meta Graph adapter
Not connectedWhere the regulars already chat
ConnectedBot API only — never scraping
Not connectedOnly with consent and real urgency
Not connectedAttach it when you staff it
Not connectedConsistency, coverage, and calmer staff inboxes.
Core questions resolve while the team serves people in the room.
Time-bound phrases live apart from evergreen trees.
Disjoint ownership makes behavior predictable.
Humans own the exceptions automation should never touch.
Reuse menu structure with local details.
A keyword can start a longer path without a second product silo.
Matching is designed for practical social text, not compiler-grade tokens. See product docs for exact modes and normalization. Design keywords people will actually type.
Yes — lists in the trigger model. Prefer coherent sets (HOURS/OPEN/SCHEDULE) rather than unrelated campaigns fighting one trigger.
Operator confusion and double-fires. Prefer disjoint live sets. Clear ownership of phrases beats clever overlapping graphs.
Complaints, booking exceptions, refunds, and anything where a canned branch damages trust. Inbox roles exist so automation is infrastructure, not a wall.
Menus are evergreen ops (hours, booking, FAQ). Campaign codes are time-bound promos. Separate automations and tags so a dead promo code does not break the menu tree.
Yes. After match, free Logic can collect fields, wait, tag, and hand off in the same contact model as other use cases.
Start free, ship an evergreen menu with a human exit, and keep campaign phrases on a short leash.