YundaDeskRight after a shipping wave, the first thing that fills up the inbox is never a complaint — it is “where is my package.” The same question, asked hundreds of times, in different languages, across different channels: one on WhatsApp, one by email, another as an Instagram DM. Agents are not bad at answering it, they simply cannot keep up. Every WISMO (Where Is My Order) question requires looking up the tracking number before replying, which eats two or three minutes each. Once volume spikes, the questions that actually need human judgment get pushed further down the queue.
This is exactly the kind of question automation is built for — the answer is deterministic, verifiable, and involves no financial decision. Here is how we recommend building it out.
Handling the WISMO Surge: conversation-volume stress test from baseline to peak
Understand why WISMO always clusters
Logistics questions are not evenly distributed. There is a clear pattern:
- Days 3 to 7 after shipment are the peak window, as customers start checking tracking numbers and message in when updates stall
- Customs clearance and overseas-warehouse transfer points commonly show tracking data that appears “stuck,” which triggers a batch of inquiries around the same order cohort
- One to two weeks after a big sale is the single largest WISMO surge of the year, as shipment volume and inquiry volume scale together
If these three patterns are not identified ahead of time, the team stays reactive — reacting to volume that has already doubled instead of preparing for it. Flagging these windows in advance is what lets you put both automation and staffing where they actually matter.
The knowledge base needs to cover more than “where is it”
Many teams assume WISMO automation just means wiring up a tracking lookup. In practice, customers ask a wider range of questions than that. The knowledge base should cover at least:
| Question type | Typical phrasing |
|---|---|
| Basic tracking | Where is my package, roughly when will it arrive |
| Timing explanation | It says shipped but tracking has not updated |
| Exception handling | Tracking has been stuck in one city for days |
| Customs questions | Do I need to do anything for customs, will I owe duties |
| Address changes | Can I still change the delivery address after shipment |
| Damage or loss | Tracking shows delivered but I never got it |
The first four categories can almost always be answered directly using the knowledge base plus order data. The last two — address changes and damage/loss — involve an action or compensation and should be escalated to a human; the next section covers how to tier them.
Let AI support read the order, not just look up a tracking number
The core of WISMO automation is not bolting on a tracking API — it is letting AI support connect “this message” to “this order.” Cross-border CRM fields like country, language, and time zone, tied to the order record, mean the AI already knows before it replies:
- What logistics stage this customer’s package is currently at
- Whether the current timing is normal or an exception, based on average delivery times for the destination country
- Whether it is currently daytime or the middle of the night for the customer, and whether tone and wording should adjust
With that context, AI support can give an answer specific to that order, instead of asking “please provide your order number” and waiting for a reply — every extra round trip just spends the customer’s patience. When the AI cannot find an answer or cannot make the call, it says the question has been received and is being verified, and hands the full context to a human — it does not fabricate a delivery date to fill the gap.
Tier the handling: what AI takes, what goes to a human
Applying the AI-first, human-backed boundary, WISMO scenarios split into three tiers:
| Tier | Scenario | Handling |
|---|---|---|
| Low risk | Routine tracking questions, timing explanations | AI answers directly using the order record |
| Medium risk | Tracking stalled for an extended period, customs hold | AI responds, confirms it has flagged the case for verification, and commits to a follow-up |
| High risk | Suspected lost package, an upset customer, a compensation request | AI de-escalates and collects information, then hands off immediately with full context |
This tiering is not about giving the AI less to do — it is about freeing up human attention for the cases that were actually overwhelming the team in the first place: lost packages and compensation calls.
Do not wait for the question: push tracking updates proactively
The other half of solving WISMO is not answering faster — it is customers never needing to ask. Sending a proactive update at key logistics milestones (picked up, shipped, cleared customs, out for delivery, delivery exception) intercepts a meaningful share of messages before they are ever typed.
For proactive outreach to actually help rather than annoy, the guardrails have to be in place:
- Cooldowns and frequency caps, so the same logistics event does not trigger repeat pings
- Quiet hours, so nothing lands in the middle of the night in the customer’s local time zone
- No interrupting an active conversation with an unrelated proactive message
- Sensitive actions always route to a human — proactive outreach never commits to a refund or compensation on its own
The three modes — observe only, confirm each message, or send automatically — can be opened up gradually as trust builds; teams do not need to start fully automated. See how proactive outreach avoids annoying customers for the full mechanism.
Turn every WISMO conversation into the next one’s answer
WISMO automation is not a one-time setup. Every time an agent handles a logistics question the AI missed — a specific overseas warehouse’s customs quirk, a special note about one shipping lane — that knowledge should get captured, not stay locked in the agent’s head.
The mechanism: when an agent answers or corrects the AI, the system generates a pending learning suggestion, which only takes effect once a manager reviews and approves it in the review queue. It is then written into skills or the knowledge base — fully traceable, testable, and reversible with one click. The next time a similar logistics question comes in, AI support can answer it correctly the first time instead of escalating again. This is also why the knowledge base keeps getting more accurate over time — see how the knowledge base feeds the AI.
Run one pass on real data before going live
Before turning this on for real, we recommend a small rehearsal:
- Pull real WISMO conversations from the last shipping peak and check each AI answer for accuracy and correct order details
- Deliberately test a few customs-exception scenarios the knowledge base does not cover, and confirm the AI hands off instead of inventing a delivery date
- Check that proactive push timing matches the guardrail settings (no repeats, no interruptions)
- Run a pass in each target-market language and confirm time zone and tone are matched correctly
Once this checks out, teams notice the shift immediately: logistics questions stop being a flood that drowns the inbox, and instead split cleanly into “the majority the AI answers instantly” and “the minority that genuinely needs a human.”
WISMO was never a question that required cleverness — it requires wiring order data, the knowledge base, and proactive outreach together correctly. Get that pipeline right, and the time the team gets back can go toward the refunds and complaints that actually need human judgment.