Turning Instagram engagement into multi-channel nurture
Instagram comments that become contacts without getting creators flagged — then Kajabi’s first Visual Editor, so DM → email → SMS → offer lives in one canvas instead of ManyChat trees.
9 min read
Role
Lead Product Designer
Timeline
Oct 2024 – Feb 2025
Team
1 designer · 1 PM · eng
My role & constraints
Lead Product Designer — sole designer with PM David Wang through Visual Editor handoff.
Owned: Comment to DM end-to-end (safety defaults, setup UI) and the first Automations 2.0 Visual Editor (canvas, branching, performance) through eng-ready specs.
Constraint: Maternity leave at handoff — another designer owned build and launch. Comment to DM shipped; VE launch metrics sit after my ownership.
Business context
Social visibility wasn’t turning into owned contacts, and deeper nurture lived in ManyChat because Kajabi Automations only supported one trigger → one action.
30%
of heroes use ManyChat for lead capture — Comment to DM is its #1 adoption driver
50%
of Kajabi contacts are imported from outside the product (85% via bulk upload)
3.5%
of contacts convert to paid within 35 days; 83% who hit GMV in their first 30 days had ≥5 contacts
Audience lives on social and in other tools — Kajabi needed native capture and deeper automation to own the journey.
Aimed at Hobbyists and Builders. Two versions:
v1
Comment to DM
Instagram → DM in today’s Automations.
v2
Visual Editor
Canvas for multi-trigger flows — up to 50 actions.
Later
Conversation Hub
View and reply to Automations 2.0 conversations in one place inside Kajabi.
Insights
These came from creator interviews about off-platform automation trees, survey and discovery work on social → contacts, and 7 usability sessions in Dec 2024 on Comment to DM setup.
01Creators couldn’t see how automations connected
Biggest pain: only isolated this→that pairs. Creators walked me through deep, wide trees in ManyChat, GoHighLevel, and Zapier. They wanted that sophistication in Kajabi — one view, and room to build elaborate journeys.
Creators reach audiences on email, Facebook, and Instagram (71% use Instagram DMs; 63% Messenger). Social is visibility; email is nurture. Viewers aren’t contacts yet — so they can’t convert inside Kajabi. 50% of contacts are still imported from outside (85% via bulk upload).
03Native capture needed safety defaults
Creators wanted native Comment to DM to drop ManyChat (“too many tools… under one umbrella”). They asked for delay guidance so Instagram wouldn’t flag them — and picked 5–10 seconds. Setup UI split by context: for v1 Comment to DM we tested a step-by-step drawer vs. a one-page modal (most preferred seeing the full setup at once); for v2’s Visual Editor, one-step config that stays tied to the canvas favored a drawer.
“There’s just too many tools and we just want things under one umbrella.”
Reelika · usability interview · ex-ManyChat
“There is a little bit more of an authentic feel to it if you can have a little delay there.”
Robin · usability interview
Those findings set the product calls below — delay and keywords as defaults, a Visual Editor canvas, and setup patterns that differ for Comment to DM vs. the canvas.
Decisions
Calls that shaped Comment to DM and Automations 2.0 — product rules first, then the setup UI for v1 vs. v2.
Product rules
01Delay DMs by default — don’t get creators flagged
Core responsibility: Automations 2.0 must not get creators flagged as spam or bots on Instagram (and later other social). Chose: configurable delay (0s–5min), up to 10 varied replies, auto-like. Creators picked 5–10s in testing. Business: safe native capture vs. ManyChat. Design: human-feeling defaults, not buried settings.
02Keywords by default; all-comments optional
Most creators want “comment HELPER for the freebie” — not a DM to every comment. Keywords first (max 5); all-comments stays for boost posts.
03Capture permissions now; don't penalize early adopters
Capture Hub-ready DM permissions in v1, but don’t count Instagram handles against plan limits until Conversation Hub can support two-way messaging — so creators can view and reply to Automations 2.0 conversations in one place inside Kajabi. Avoids punishing early adopters and re-consenting later.
04New Visual Editor canvas — don’t stretch the old UI
I designed the initial Automations 2.0 Visual Editor. Extending Rails-era one-to-one UI for long chains would make it unusable — and Comment to DM is only one recipe on a much bigger canvas. Chose: a new canvas (pan/zoom) with multi-trigger, branching, and stacked actions across text, email, and (later) more social platforms — up to 50 actions per flow, not unlimited — plus per-step performance on the journey, and a drawer for detail config so creators keep the canvas in view. Business: deep multi-channel trees stay in Kajabi. Design: clear journey view with data on what works; intricate one-step edits (SMS copy, etc.) open beside the flow, not over it.
Triggers
Actions
UI
v1 — Comment to DM setup
In Dec 2024 usability sessions we tested both patterns for Comment to DM. Most creators — especially Builders — preferred seeing the full setup at once over a forced Next → Next path. They called the one-page version clearer and “better to visualize”; the step-by-step drawer split decisions across screens.
Explored
Side drawer · step-by-step
Select options, then press Next for the next set. Clear for some first-timers, but felt linear and fragmented in testing.
Preferred
Modal · one page + accordion
All options on one page; accordion opens more as creators select. Preferred in sessions — full setup visible, faster to scan and edit.
v2 — Visual Editor config
On the Visual Editor canvas, the call flipped. Many option layers are already chosen on the nodes themselves; the panel only needs to open for denser, one-step work — writing SMS copy, longer message fields, and similar detail. A modal would cover the canvas those controls still need to stay tied to. Creators preferred a drawer here: one step, beside the flow, interacting with what’s already selected on the Visual Editor.
Explored
Modal over the canvas
Blocks the journey view while creators configure a node — breaks the link between drawer contents and the Visual Editor they’re editing.
Chose
Drawer · one step beside the canvas
Opens when options get intricate (SMS text, etc.); stays connected to the selected node. Preferred for one-step config in the Visual Editor experience.
What we built
v1 — Comment to DM
Shipped into Marketing → Automations (next to Email and Funnels): connect Instagram, pick a post, keywords or all-comments, delay + auto-like, DM with Kajabi link, results (delivered / opened / clicked). Lead capture without waiting on the new canvas.
01
Comment
→
02
Keyword
→
03
Delay
→
04
DM
→
05
Results
Findings · after v1
Creators wanted native capture — and a way to see the rest of the journey, not only the Instagram DM.
What we heard after v1 shaped what came next:
Delay defaults held up — creators kept the human-feeling delay rather than asking for instant sends.
One-page setup won — creators preferred seeing the full setup at once over a step-by-step path.
The DM wasn’t the whole journey — deep, multi-channel trees still lived in ManyChat. That unlocked v2, the Visual Editor.
What we built
v2 — Visual Editor
A canvas far beyond Comment to DM: creators compose journeys across text, email, and — later — more social platforms, with multi-trigger, branching, and stacked actions — up to 50 per flow (vs. 1 trigger + 1 action today). Detail config opens in a drawer beside the canvas — not a modal — so node options stay tied to the flow (SMS copy and other one-step, text-heavy edits). Performance sits on the journey too — creators can see how each step runs, not just that the tree exists. Comment to DM is one recipe on that canvas. Every v1 field was shaped so it doesn’t need a redesign when it lands there.
Up to 50 actions in one flow (not unlimited)
Channels beyond Instagram DMs — text, email, expandable to more social
Per-step performance on the canvas — see how each trigger and action performs
Building the journey is only half the job. Toggle performance on the canvas to see how each step runs — sends, opens, clicks, unsubscribes — without leaving the flow. Creators can spot what’s working and tune what isn’t on the same tree they built.
Outcome
Designed and tested both versions through eng-ready handoff. Another designer took build and launch when I went on maternity leave.
Delivered
Comment to DM — shipped. Live in today’s Automations, with delayed DMs, varied replies, and auto-like so Instagram doesn’t flag creators.
Visual Editor — designed, tested, and ready for engineering. A multi-channel canvas (text, email, future social) with multiple triggers, branching, up to 50 actions, and per-step performance. We tested how creators build on it before handoff.
Documented for what comes next: the order to ship in, plus permissions and contact handling designed so the future Conversation Hub can use them — without asking creators to re-connect or re-approve.
Reflection
What worked:
Shipped Comment to DM first, inside the Automations creators already used — before the Visual Editor existed. Built-in delays and varied replies meant Instagram didn’t flag their accounts.
Interviews made the Visual Editor’s job clear. Creators couldn’t see how their automations connected, and they’d already built complex flows in tools like ManyChat. They wanted that view inside Kajabi.
Tested the canvas before handoff, so engineering got a proven way to interact with it — not just pictures of screens.
What I’d change: ask engineering earlier what Comment to DM needed under the hood. It looked like a quick feature, but rebuilding how contacts and Instagram handles are stored slowed it down more than the Visual Editor.
On the handoff: I went on maternity leave after design and testing. I documented the v1 and v2 plan, the safety defaults, and the permissions (designed so the future Conversation Hub could use them) — so another designer could own the build without re-deciding those things.
What I did beyond design: created new Design System components for the Visual Editor. Kajabi didn’t have an endless (infinite-canvas) editor, so the building blocks were net-new: canvas nodes for triggers and actions, connectors and branching, drawer-based step config, and per-step performance states — patterns other teams could reuse instead of starting from zero.
Lesson: on social, fast replies look like a bot and get accounts flagged. A short delay that feels human is what makes native capture safe. And one Visual Editor puts text, email, and future social in one place — instead of stopping at the DM.
What's next3 roadmap bets
AI that drafts and suggests on the Visual Editor and Comment to DM — creators still control send.
Visual Editor
Kick off a customized tree from goals questions; suggest valid next steps; explain or debug a journey in plain language; review for missed opportunities or stronger configs; migrate ManyChat / Zapier trees into Kajabi. Why: faster time-to-first-flow and deeper multi-step use in-product; reclaim off-platform automation spend; higher conversion from contacts already owned.
Comment to DM
Draft and vary human-feeling DM replies; coach keywords and delay ranges from the post; warn when instant reply risks spam flags; triage comments (buy intent vs. spam vs. FAQ) before the automation fires. Why: more completed native setups; fewer flagged accounts; better-qualified leads into nurture — keep social capture inside Kajabi.
Adoption
Goal → recipe wizard (starter flow + copy), then open in the editor to tweak, plus short inline “why this delay / why this branch” tips. Why: activate Hobbyists and Builders who stall on a blank canvas; cut time-to-value without extra support cost.