Loomcraft Studio sells handloom textiles and home goods online, with a two-person support team answering nearly everything through a chat widget on their site. Order volume is naturally seasonal — a festive-season sale in late 2025 tripled orders almost overnight — and support volume tripled right alongside it, because most of what customers were messaging about was some version of "where's my order."
Illustrative example. Loomcraft Studio is a composite scenario built to show how AIVA works for this kind of business — not a verified customer account.
The problem
Founder Meera Iyer's team was getting roughly 200 messages a day at peak, and the large majority were order-status questions that required the same manual steps every time: find the order, open the courier's tracking page, copy the status back into chat. None of it was hard. All of it was constant, and it was crowding out the messages that actually needed a person — a damaged item, a wrong size, a refund request — which sometimes sat for a day behind a queue of tracking lookups.
The worst of it landed on the sale's first weekend. Orders came in faster than the two-person team could even open them, let alone answer questions about them, and the chat widget's queue backed up to the point that some customers were waiting the better part of a day just for someone to check a tracking number. A few of those customers messaged again to ask if their order had actually gone through at all, which added more volume on top of volume that was already unmanageable. Meera remembers checking her phone at midnight that weekend and seeing the queue longer than when she'd last looked, hours earlier.
"'Where's my order' used to eat our afternoons," Meera says. Hiring seasonal support for four months a year wasn't something a small team could easily justify — the cost of training someone for a role that mostly disappears again by January never quite penciled out against four months of relief.
Getting it live
Connecting AIVA to Loomcraft's actual order and shipment data took a few days, most of it spent making sure product names, order statuses, and courier tracking fields mapped correctly rather than approximately. That kind of integration is common groundwork for any online seller — most run on a platform like WooCommerce or a similar storefront system where order and shipment status already exist in a structured, connectable form; the setup work is mapping that existing structure accurately, not building it from scratch. That mapping mattered more than it might sound: Loomcraft ships through more than one courier depending on the destination, and each courier's tracking data comes back in a slightly different format.
That difference caused the one real hiccup during setup. Early testing surfaced a mismatch on one specific courier, where its "out for delivery" status wasn't being read correctly and was showing up as a generic "in transit" instead — not wrong exactly, but less specific than it should have been. It was caught during the test period, before launch, by running AIVA against a batch of real recent orders and checking the answers against what the courier's own site actually showed. The mapping was corrected the same week, and by the time AIVA went fully live on the chat widget, status answers were accurate across every courier Loomcraft uses.
Answer the routine, flag the rest
We connected AIVA to Loomcraft's order and shipment data so it can answer a status question with the real answer — carrier, current location, expected delivery — instead of a generic "it usually ships in 3-5 days." AIVA handles the routine lookups directly through the chat widget and flags the genuine exceptions — damage claims, wrong items, refund requests — to a human, with the order details already attached so nobody has to ask the customer to repeat themselves.
The team didn't get faster at answering "where's my order." They stopped having to answer it at all.
What a support day looks like now
Most conversations now start and end without either of Loomcraft's two support staff seeing them. A customer asks where their order is, AIVA pulls the real shipment status and delivery estimate, and that's the whole exchange. During the most recent big sale, that pattern held even as volume spiked — the difference from the year before wasn't that the team worked faster, it's that most of the spike never reached them at all.
The messages that do reach a person now look different than they used to. A customer writing in about a saree that arrived with a pulled thread, or a cushion cover in the wrong size, gets routed to a human with the order number, product, and purchase date already pulled up — no back-and-forth just to get everyone looking at the same order. Meera says the team's average response quality on those went up, not down, simply because nobody's rushing through a damage claim to get back to a tracking-number queue.
Meera also started watching the sale differently than in previous years. Instead of judging how the launch was going by how backed-up the chat queue looked, she checks AIVA's dashboard for message volume and how much of it AIVA is resolving without help — a number that stayed high even at the peak of the most recent sale, which told her within the first few hours that the site could keep taking orders without the support side buckling under it. That's a different kind of visibility than "how long is the queue," and it's the number she now watches first on any big sale day.
The result
- Daily message queue reaching a human: roughly 200/day → about 12/day, a 94% drop in what the two-person team has to personally handle
- Peak-season order volume handled: the 3x spike went through without adding seasonal headcount
- Response time on a status question: instant, versus a queue wait during peak hours
- Team focus: shifted almost entirely to the exceptions that actually need a human decision
What Meera would tell another online seller
Her advice is to test against real orders before flipping the switch, not just sample data. "We found our one real gap — a courier's delivery status coming through vague instead of specific — by running it against orders that were actually mid-shipment, not by reading through settings," she says. "If we'd just trusted the setup and launched, a real customer would have hit that gap during our biggest weekend of the year instead of us catching it quietly, days earlier, while testing."
She'd also tell another founder not to think of it as a queue-reduction tool first. "The number we track most now isn't messages per day, it's what our two people actually spend their time on. The queue dropping was the visible part. The part that mattered was that a damaged-item message doesn't sit behind fifty tracking questions anymore."
"AIVA answers instantly and only flags the odd one out," Meera says. "Our queue went from 200 messages a day to about twelve. I don't know how we would have handled the season otherwise — we'd have needed to hire, and even then we'd have been slower than this." She's also glad the cost tracked the business rather than the other way around — Loomcraft pays for what it actually uses under AIVA's pricing, so the quiet months between sales don't carry the same cost as the peak ones.
Other retailers can read how a similar setup applies to boutique fashion retail, or see what it actually takes to add a web AI assistant without a developer on staff. Sellers with their own seasonal spikes can check usage-based pricing for seasonal businesses and start free before their next big sale.