Order tracking: the most underrated support automation

It is 11

. A customer named Priya ordered a birthday gift six days ago, the party is Saturday, and the tracking page she bookmarked has said "in transit" since Tuesday. She is not angry. She is nervous. So she opens the chat on your site and types the most common sentence in all of ecommerce support:

"Where is my order?"

And then nothing happens until your support inbox opens at 9am, at which point a person spends four minutes looking up an order number, pasting a carrier status, and typing a delivery estimate. Multiply that by every Priya, every night.

Depending on the store, "where is my order" (the industry calls it WISMO) is 20 to 40 percent of all support tickets. And I want to convince you of something that sounds too tidy to be true: WISMO is the single best first support automation you will ever ship. Not because it is impressive. Because it is boring in exactly the right ways.

Why the most boring ticket is the best one to automate

Here is the counterintuitive part. When teams pick their first automation, they instinctively aim at the hard stuff: the angry escalations, the nuanced refund judgment calls, the tickets that eat an hour each. That is backwards. The hard tickets are hard because they need judgment, and judgment is the one thing automation is worst at.

The ideal first automation candidate has three properties, and WISMO has all of them in their purest form:

It is predictable. Every WISMO ticket has the same shape: an identity (order number or email), a lookup, a status, an estimate. There is no creative variation. Priya at 11pm and Marcus at 7am are asking structurally identical questions about different rows in the same table.

It is high volume. Automating a ticket type that arrives twice a month saves you nothing. WISMO arrives constantly, spikes after every marketing send, and floods in during carrier delays, which is exactly when your human team is most underwater.

The answer is verifiable. This is the underrated one. A WISMO answer is either right or wrong, checkably, against the carrier's own data. Compare that to "which plan should I choose?", where a bad automated answer can sound plausible for weeks. When the ground truth is a tracking API, you can audit your automation every single day.

There is a fourth property that is easy to miss: nobody's feelings are on the line. A refund denial needs empathy and discretion. A tracking status needs accuracy and speed. Customers do not want a thoughtful human meditation on their package's journey. They want to know where the package is, at whatever hour the worry strikes. This is the rare ticket where the automated experience is genuinely better than the human one, because the human one comes with a ten-hour wait.

I have written before about how AI support reduces repetitive work, and WISMO is the canonical example: the ticket your team has answered ten thousand times with the same three sentences.

What does a good automated WISMO exchange look like?

Bad WISMO automation is easy to picture, because you have met it: a bot that replies "Please check the tracking link in your confirmation email!" That is not automation. That is a rejection letter with a smiley tone.

A good exchange does four things, in order:

First, it identifies the order with minimal friction. Ask for an order number or the email used at purchase. If the customer is chatting on a channel where you already know who they are (logged into their account, or messaging from the phone number on the order), skip the question entirely and confirm: "I found your order #4817 from June 2nd, the ceramic table lamp. Is that the one?"

Second, it gives the real status, not a euphemism. "In transit" is what the carrier page says; your automation should translate: "Your package left the Leipzig sorting facility this morning and is on a delivery vehicle in your region." Specificity is what separates an answer from a brush-off.

Third, it gives an estimate with its source. "The carrier currently estimates delivery on Thursday, June 12th." Note the phrasing: the carrier estimates. We will come back to why those two words carry the whole system.

Fourth, and this is the one everyone skips: it includes an escape hatch that triggers itself. If the package has not scanned in four days, if the status says "returned to sender," if the address failed validation, the automation should not cheerfully recite the broken status. It should say: "This shipment has not moved since Saturday, which is longer than normal. I am flagging this for our team, and a person will follow up by tomorrow morning." A stuck shipment is no longer a WISMO ticket. It is a problem ticket wearing a WISMO costume, and the automation's job is to notice the costume.

This is the same triage logic I laid out in not every customer question deserves AI: the healthy pattern is structured automation for the structured cases and a fast, obvious path to a human the moment the case stops being structured.

The trust rule: never guess a delivery date

If you take one sentence from this article, take this one: your automation must never state a delivery date the carrier did not state.

The temptation is real. Customers ask "will it arrive by Friday?" and a confident "yes, it should!" ends the conversation pleasantly. But an automated system that guesses dates is a promise-generating machine, and every broken promise lands on your human team with interest. The customer does not remember that a bot guessed. They remember that your store told them Friday.

So the rule is: say what the carrier says, attribute it, and be honest about uncertainty. "The carrier estimates Thursday. Estimates can shift by a day during busy periods, and this page will always show the latest status." That answer is slightly less satisfying in the moment and enormously more trustworthy over a hundred orders. Trust compounds; false reassurance compounds too, in the other direction.

The same rule generalizes: never let the automation speculate about why a package is delayed, whether customs will charge a fee, or whether a lost package will be refunded. Facts from systems of record, yes. Predictions and promises, no. Those belong to humans.

What your team does with the hours back

Run the numbers for a mid-sized store: 1,000 tickets a month, 30 percent WISMO, four minutes each. That is 20 hours a month of copy-pasting carrier statuses. But the hours are the smaller prize.

The larger prize is attention. WISMO tickets do not just consume time; they interrupt. An agent who is halfway through untangling a duplicate-charge dispute, and stops to look up a tracking number, pays a context-switch tax on both tickets. Clear the WISMO layer and the queue that remains is smaller, slower-moving, and made of exactly the problems humans are for: judgment, exceptions, and the stuck-shipment escalations your automation flagged overnight.

Which means your escalations get better, not just fewer. The 11pm nervous customer got her answer at 11pm. The genuinely stuck package reached a human with the context already attached. Nobody spent their morning as a human tracking API.

Launching WISMO automation in one week

You can ship a credible version of this in a week. Here is the shape of it:

1. Pull one month of tickets and tag the WISMO ones. Count them. This number is your baseline and your business case. Note the sub-variants: normal status checks, stuck shipments, wrong-address panics.
2. Connect your order and tracking data. Whatever platform you use, the automation needs live access to order status and carrier tracking, keyed by order number and email. If you sell on Shopify or WooCommerce, this is typically an integration toggle, not a project.
3. Write the exchange, including the failure branches. Draft the happy path (lookup, status, attributed estimate), then draft the escape hatches: no scan in X days, returned to sender, order not found, customer says "this is wrong." Every branch ends in either a verified answer or a handoff, never a dead end.
4. Set the escalation rules with your team. Agree on what "stuck" means in days, who receives flagged shipments, and the promised follow-up time. Write these down; the automation will hold your team to them.
5. Launch on one channel, watch every transcript for three days. Pick your busiest channel and read every single WISMO conversation for the first 72 hours. You are checking two things: is the status accurate, and did every odd case reach a human?
6. Measure resolution, then expand. Track what share of WISMO conversations ended without a human and without a reopen. When that number is healthy and stable, roll out to the next channel and the late-night hours you were never covering anyway.

If you would rather not wire this together yourself, this is a standard pattern in support platforms with ecommerce integrations; Fetchply ships order tracking over the chat widget, WhatsApp, and email out of the box, and the free plan is enough to run the one-week pilot above.

• WISMO is 20 to 40 percent of ecommerce tickets and the ideal first automation: predictable shape, high volume, and an answer you can verify against carrier data.
• A good exchange looks up the order with minimal friction, gives the real status, attributes the estimate to the carrier, and escalates itself when a shipment looks wrong.
• Never let automation guess delivery dates or make promises; say what the carrier says and be honest about uncertainty.
• The payoff is not just hours saved but focus: humans keep the judgment work, and the stuck-shipment escalations arrive with context attached.
• Pilot on one channel, read every transcript for three days, and measure resolutions without human touch and without reopens.

What percentage of tickets is WISMO for a typical store?

Most ecommerce stores see 20 to 40 percent, with spikes after promotions and during carrier delays. Tag one month of your own tickets to get your real number; it is usually higher than teams guess.

Should the automation answer WISMO for orders that look stuck?

It should recognize them, not answer them normally. A shipment with no carrier scan for several days should trigger an honest message ("this is taking longer than normal") plus an automatic escalation to your team. Reciting a stale status for a stuck package destroys trust.

Can I automate WISMO without any AI at all?

Largely, yes. The core exchange is a structured lookup: identify the order, fetch the status, present it. A guided, step-by-step flow handles that deterministically. AI helps at the edges, like understanding "my parcel vanished lol" as a WISMO question, but the backbone is plain data.

How do I measure whether the automation is working?

Three numbers: the share of WISMO conversations resolved with no human involvement, the reopen rate on those conversations, and the volume of WISMO tickets reaching your human queue compared to your pre-launch baseline. If resolution is high, reopens are low, and human WISMO volume drops, it is working.