A 30-person ecommerce support team takes messages in three places: the website form lands in a shared Gmail, the LINE official account, plus Facebook DMs. Complaints get dropped, the same customer gets two replies from two agents, and "average response time" at month-end is a guess. When you get serious about "we need a help desk," the first fork is: SaaS like Zendesk, or self-build? This covers planning, cost, and pitfalls.
When It Fits vs When It Does Not
- A ticketing system fits when: steady multi-channel volume, you need routing and tracking, you want SLAs and agent metrics, complaints need an audit trail, team of 3+.
- Self-build fits when: you already have a Laravel admin to integrate, you need tickets tied deeply to order/member data, per-seat SaaS gets too expensive long-term, or you have data-residency/compliance needs.
- Not for you (yet): low daily volume one person handles on LINE; you only want a FAQ page, not a ticket flow; the team has not defined "done" before buying a system.
Alternatives Matrix
| Option | Pros | Cons | Cost band |
|---|---|---|---|
| Zendesk / Freshdesk (SaaS) | Out-of-box, multichannel, full reports | Per-seat fee, limited customization, data offsite | ~US$25–115/seat/mo |
| Intercom (conversational SaaS) | Strong live chat + AI bot | Usage-priced, expensive | Med-high |
| Shared inbox + labels (DIY start) | Near-zero cost, start today | No SLA/routing, hard to scale | Low |
| Self-build Laravel + Filament | Ties to orders/members, no per-seat fee, full custom | Build and maintain | One-off build + maintenance |
Full Process (Tools and Deliverables)
- Week 1 — needs and flow: inventory channels, define the ticket state machine and SLAs. Deliverable: flow in FigJam, SLA table.
- Week 2 — data model and integration: design tickets/messages/agents tables, connect LINE Messaging API and email (IMAP/SMTP or Mailgun). Deliverable: ERD, integration list.
- Weeks 3–4 — admin and routing: build ticket list, assignment, internal notes, status flow with Filament. Deliverable: working admin.
- Week 5 — reports and alerts: response/resolution time, agent metrics, overdue reminders (queue + scheduler). Deliverable: dashboard.
- Week 6 — pilot and handover: two-week agent pilot, tune, write the manual. Deliverable: launch + SOP.
Real Cost Breakdown
- SaaS: 10 seats x US$55/mo ≈ US$6,600/yr (~NT$210k), rising yearly and scaling with headcount.
- Self-build: one-off build (by scope) + host (VPS monthly) + maintenance; most channel APIs are free, use Mailgun (usage-based) for high email volume.
- Hidden costs: LINE official account message-quota overage, agent training hours, data migration, change requests for new reports.
Reality vs What Clients Imagine
- Clients think "a system makes support faster"; truth: without a defined process, a system just digitizes the chaos. Define states and SLAs first.
- Clients think "an AI bot handles 80%"; truth: bots handle repetitive FAQs; complex complaints still need humans, and bots need a knowledge base.
- Clients think "self-build is always cheaper"; truth: small teams usually win with SaaS; self-build pays off with high headcount and deep integration.
Common Pitfalls and Fixes
- No definition of "done": draw the ticket state machine or you get zombie tickets.
- Channels in silos: merge the same customer across channels (dedupe by email/phone) to avoid double replies.
- SLA as decoration: turn overdue into automatic reminders and escalation, not memory.
- Unmaintained knowledge base: assign an owner; distill high-frequency questions into FAQs so the bot has material.
- Reports as an afterthought: define metrics before launch so tables store the right data.
Success Metrics + 90-Day Roadmap
- Day 30: all channels converged into one system; first-response-time baseline set.
- Day 60: average resolution time down, overdue rate < 10%, top 20 FAQs captured.
- Day 90: bot handles repetitive FAQs, agent metrics live, evaluate adding channels.
Decision Checklist
- ☐ Support comes from 2+ channels
- ☐ Team of 3+ needs routing
- ☐ Need SLAs and metrics
- ☐ Complaints need audit/compliance
- ☐ Want deep order/member integration
- ☐ Per-seat SaaS is pricey long-term
- ☐ Already have a Laravel admin to extend
- ☐ "Ticket done" is defined
- ☐ Someone owns the knowledge base
- ☐ Budget covers post-launch maintenance
Six or more checks with the last few leaning "integration + cost control" make a self-built Laravel desk worth evaluating; mostly early checks with a small team mean SaaS is fastest.
FAQ
How long and how much to self-build?
An MVP covering multichannel convergence, routing, and SLA reports usually launches in 5–6 weeks; cost depends on integration depth. Compare "one-off build + monthly maintenance" against SaaS "yearly per-seat." Small teams start with SaaS.
How do I bring LINE into tickets?
Use the LINE Messaging API with a webhook to receive messages and a queue to process replies. Watch official-account message quotas and overage fees; split auto-replies from humans at volume.
I use Zendesk — worth moving to self-build?
Two factors: annual seat-fee total and how badly you need tickets tied to your orders/members. Six-figure NT$ yearly with strong integration needs favors self-build long-term; otherwise stay on SaaS.
Call to Action
Want to know whether SaaS or self-build fits your case? ScriptWalker offers help-desk planning and Laravel self-build, starting with a free needs review:
- Email: [email protected]
- Phone: 0916-224-047
- LINE: @ufv9089p