By the Onsites AI team · Last updated · 5-minute read
Dropshipping support has an awkward truth at its center: the store makes promises about shipping it cannot fulfill and depends on suppliers it cannot see. Everything else — the WISMO flood, the tracking questions, the refund requests, the "it's been 3 weeks" emails — follows from that structure. And the volume curve is dropshipping's special cruelty: a product catches on TikTok and the desk's load doubles overnight, at exactly the moment the supplier relationship is still informal. This guide is the playbook that survives that curve: tracking numbers synced instead of typed, supplier chase organized as threads with owners, a refund ladder written before the first angry email, and a desk shape — three free seats, $15 beyond, no per-ticket meter — priced to spend on the flood rather than the flood's invoices.
The dropshipping desk fails in a known order. The tracking treadmill: "where is my order?" is 60–80% of volume; hand-typing tracking replies dies at 50 orders a day and is unmanageable at 200. The supplier chase: "where is EISU-4821? has been restocked? why is the tracking stale?" — chased across supplier chats by whoever notices. The refund improvisation: every refund negotiated ad hoc — one buyer gets a full refund, another gets 40% and an apology — the inconsistency shows up in reviews. The peak pile: the flood lands on the desk's weakest process; unanswered tickets age into angries, angries age into chargebacks. Each failure is a missing system, not a missing person — which is why the playbook fixes systems first.
The single highest-leverage fix in dropshipping support. Sync tracking into threads: orders and their tracking numbers flow into the desk so every "where is order #1041?" is answered from data — status, carrier, last scan, expected delivery — inside the reply; the reply itself references the shipment's live state rather than yesterday's memory. Proactive beats reactive: the shipment's scan stops moving; the desk drafts the delay note to the buyer before they notice — announced delays are forgiven, discovered ones are charged back (the delay-email guide writes the note). The WISMO macro: the 20% that remains hand-typed drafts from the template and ships at the copilot's speed with human review — the credits line at flood volume stays single-digit (published rates). The treadmill's arithmetic: a store doing 100 WISMO asks weekly saves a person-day a week by syncing alone.
The supplier relationship is support's invisible half, and it fails when it lives in personal chats. One chase thread per supplier: the desk's team-chat or supplier channel holds the running chase — stock questions, tracking staleness, the EISU-4821 mystery — searchable by order number, visible to the whole team; the buyer's thread links to it. Chase templates with dates: "tracking not updated in 96 hours" is a trigger, not a hunch — the chase asks for a status with a deadline, and the desk's notes record the promise. The escalation is written too: the supplier who misses twice moves to the alternate-supplier note; the desk's memory holds the history. The chase is support work — treat it as threads with owners, reviewed weekly like any other queue. During the flood: the chase system scales because it is written; the informal chat system fails because it is memory. Suppliers come and go — the desk's threads of what they promised, and failed to, compound.
The refund ladder is policy, and policy written at scale beats policy improvised at 2 am. Write the ladder: (1) item not arrived past the promised date → replacement or full refund, buyer's choice, no argument; (2) arrived damaged → photo, refund or replacement, same choice; (3) "changed my mind" → return window policy stated plainly; (4) chargeback incoming → refund immediately, the fee fight costs more than the refund. Empower the desk: the policy lives in the handbook the desk reads (the writing the AI uses); the reply drafts from it, human-approved; the refund-reply guide writes the four responses. The consistency dividend: reviews stop mentioning unfair refunds because refunds stopped being negotiated; the ladder's cost is predictable, and its consistency is the store's reputation. The desk's shape — seats and credits — means the ladder's execution at 10× volume costs the same per reply as at normal volume.
Work one real flood to see the systems pay. A store's product catches on social: orders jump from 40 a day to 180, support volume from 25 tickets to 140 daily, the supplier's factory is at capacity. The unprepared desk: tracking replies typed one by one until answering stops at 6 pm with 40 unanswered; two buyers get different refunds for the same fault; the supplier chase lives in one owner's phone; by day five, three chargebacks have landed and reviews mention ghosting. The prepared desk: tracking synced — 70% of the flood answers itself from data; delay notes went out day one because the factory warning arrived before the questions did; the refund ladder answers identical faults identically; the chase thread asks the factory for a capacity date with a deadline; the extra help (one seat, $15) clears the remainder on a clock. Same flood, opposite outcomes — and the desk's cost during the store's best-ever week stays near its quiet-week rate, because the meter rides seats and cents, not resolutions. The flood is dropshipping's constant; the difference between stores is entirely whether its systems were written before it arrived.
The shape question is pricing and structure. Structure: one thread per buyer (their whole history), orders synced, the supplier chase beside it, the handbook behind the drafts — a store doubling overnight adds tickets, not systems. Pricing: seats are the flood's cheap problem ($15 beyond three free — the flood hires help, not tiers); credits ride usage (the flood's AI line is tens of dollars, not a subscription's); no per-resolution meter taxes the store's best month twice (the resolution-meter guide works that trap). Self-hosting exists for the store whose margins or data rules need it ($15/seat annual, 10-seat minimum, 60-day trial). The 30-day check after the flood: the queue drains daily, the tracking reads from data, the refund ladder's reviews show consistency, and the desk's costs during the store's best week — the meter that does not spike — are themselves a fact the supplier chase can point at with pride.
What breaks first in dropshipping support at scale?
The tracking treadmill (60–80% of volume is WISMO), then supplier chase — chased across personal chats by whoever notices — then refund consistency (improvised at 2 am), and finally the peak pile where unanswered tickets become chargebacks. Each is a missing system, fixable by playbook.
How do tracking replies stop being hand-typed?
Sync orders and tracking into the desk so every reply reads live status, carrier and expected delivery from data; draft delay notes proactively before buyers notice; the remaining WISMO replies draft via the copilot with human review. A store doing 100 WISMO asks weekly saves roughly a person-day.
How should supplier chasing be organized?
One chase thread per supplier, searchable by order number, with dated triggers ("tracking stale 96 hours") and recorded promises. The buyer's thread links to the chase. Suppliers change; the desk's memory of what each promised compounds.
What refund policy survives scaling?
The written ladder: not arrived past the promised date — a replacement or a full refund (the buyer's choice); damaged — photo then the same choice; changed mind — stated return window; chargeback incoming — refund immediately. Consistency in reviews is the dividend of policy over improvisation.
What desk shape handles volume doubling overnight?
One thread per buyer, orders synced, supplier chase alongside, handbook behind the drafts; seats at $15 beyond three free, AI credits at published prices, no per-resolution meter. Help during a flood prices the same per reply as normal volume.