The tax on growth: what November does to your queue and your margin
Sell a hundred orders and a familiar pile of questions shows up. Sell a thousand and you get ten piles. Growth in an online store is almost linear in one disappointing way: every package you ship is a future conversation, and most of them land within two weeks of dispatch.
For most of the year the pile is a cost you absorb. Then November arrives, and the pile arrives with it.
What November actually does
Peak season does not break your marketing or your checkout. It breaks your queue. The mechanics are boring and ruthless:
- Volume compounds. More orders means more of every question you already get, in roughly the same proportions.
- WISMO takes over. “Where is my order” was already the biggest pile you had. Carrier lag over peak shipping makes it worse, because the answer is genuinely later than the customer wants.
- Your answers get slower. The same six questions arrive at 3x volume while you are also the person negotiating with suppliers, fixing the ad account and packing the BFCM orders.
- Slow answers cost twice. A refund chased twice, a chargeback filed in frustration, a one-star review that mentions your silence. The queue stops being a time cost and becomes a revenue cost.
The uncomfortable part is the arithmetic on your own hours. If peak volume adds two hundred cases a month and each one takes you four minutes including context switching, that is thirteen hours of your November gone on questions you have answered a thousand times. Thirteen hours you needed for selling.
Your three realistic options
Owners facing this generally pick from three lanes. None of them are magic and all of them have a real cost model.
1. Absorb it. You answer everything yourself and lose the hours. Costs nothing on paper, costs the most in practice, because your hours are the store’s most expensive input.
2. Hire temporary help. A seasonal support person is real labor and honest labor. Train them for two weeks, pay them for two months, watch them leave in January with everything you taught them. Works, and repeats every year like a subscription you pay in onboarding.
3. Automate the repetitive pile. An AI teammate answers the where-is-my-order, returns, order-edit and policy questions from your approved sources, and hands the rest to you. The cost is per case, the knowledge stays, and it does not leave in January.
The right answer differs by store. But the decision deserves actual numbers, so here are the published rates in this market as of September 2026.
What automation costs, in published numbers
The category has consolidated around per-conversation pricing on top of platform fees. Published rates:
| Option | Published cost model | Fine print |
|---|---|---|
| Gorgias AI | $0.90 to $1.00 per AI conversation | On top of helpdesk subscription; billed per conversation |
| Siena | $750/mo platform fee + $0.90 per automated conversation | Enterprise-oriented; own reviews suggest smaller teams are not the fit |
| Yuma AI | ~$850/mo floor for small merchants | Billed per resolved case on top |
| Loqum | $500/mo for up to 350 cases | $1,000/mo for 850, $3,000/mo for 4,000. Managed: we build, train and run it |
Per-case math for a mid-size store doing 400 peak-season cases a month: at Loqum that is the $1,000 plan, about $2.50 a case, with a named engineer accountable on a monthly call. At per-conversation rates, an unanswered billing question costs the same as a resolved WISMO. Read the fine print on what counts as a conversation before you compare any of these numbers.
A managed option is not the right call for everyone. Below roughly 150 cases a month, the honest answer is that good macros and a strong FAQ will carry you. Past that, the hours come off your margin either way, and the only question is which line item they hide in.
What you can do this week, whatever you choose
- Write the six answers once. Where is my order, how do returns work, can I change the address, where is my refund, do you have a code, my package arrived damaged. One paragraph each, in the tone you want customers to hear.
- Turn them into saved replies in your helpdesk. Every platform has them. This is the cheapest automation that exists and most stores never finish it.
- Put the carrier tracking link in your dispatch email. A slice of WISMO is people who never found the tracking link. Kill that slice.
- Watch the ratio for a month. Cases per hundred orders. It is the single number that tells you when the pile has become a tax you should not be paying.
The point
The tax on growth is not a technology problem. It is a labor problem with a boring shape: repetitive questions arriving in proportion to your success. Software that answers them is now priced like labor, which means you can shop for it like labor, in cost per case, and hold it to the same standard you would hold a new hire.
That is the whole thesis of what we build at Loqum: an AI teammate that works inside the helpdesk you already use, trained on your policies, managed by a named engineer, at a fixed monthly cost you can put next to your margin. If reading the arithmetic made you want to run it on your own numbers, that conversation starts with Anna.