Shipping into the app store in December

We put Vigilance into public beta in November and got into the Shopify App Store this month. Which means we launched a checkout product during the highest-volume weeks of the ecommerce year, and I've been asked several times whether that was on purpose.

Partly. It's a genuinely mixed decision and worth writing down while I can still feel both sides.

The case for launching now#

The problem is at maximum visibility. Coupon-code injection is a leak that scales with traffic. In March a merchant might not notice. In the last week of November, the volume of extension-driven code attempts is high enough that it's visible in the numbers, and a merchant who's paying attention can see it happening to them in real time. Selling an unfamiliar problem is enormously easier when the customer can watch it occur.

The value is legible immediately. Blocked codes are a number. During peak, that number gets large fast, and multiplied by average order value it becomes a sentence a merchant repeats to other merchants.

Merchants are actually looking at their checkout. Eleven months of the year, checkout is something that works and nobody thinks about. Right now it's the thing everyone is staring at.

The case against, which is also strong#

Nobody wants to touch checkout in December. This is the obvious one and it's real. The instinct of any competent operator during peak is to change nothing. We're asking merchants to install something that sits in the most sensitive part of their funnel during the weeks where a mistake is most expensive. A significant number of interested people have told us, reasonably, to come back in January.

A bug now is catastrophic in a way it isn't in April. The blast radius of a checkout problem scales with the same volume that makes the value visible. Same multiplier, both directions.

Support load lands during the period you're least able to absorb it. Every new install is a configuration conversation, and they all arrive in the same three weeks.

What we did about it#

Nothing clever. We were honest with merchants about the timing tradeoff, and we told people to wait if waiting was the right answer for them. That cost us installs this month and I think it was correct — a merchant who installs reluctantly during peak and has any issue at all becomes a bad review that outlives the season.

The other thing: we went into public beta in November rather than a full launch, deliberately, so the framing set expectations. Beta is a real signal to a sophisticated merchant and the merchants who install during peak are self-selecting for sophistication.

What I actually think we bought#

Not installs. Evidence.

The thing we most needed was proof that this problem is as large as we say it is, on real stores, at real volume, with numbers we can point at. Peak season generated more of that in six weeks than a normal quarter would have in six months. Merchants who are seeing the blocked-code counter go up right now are going to be our entire sales argument for next year.

We're getting Shopify Plus merchants signing up. The math for a brand at that volume isn't a stretch — it's arithmetic, and now it's arithmetic with our own data behind it.

Ask me in the spring whether the peak-season cohort retained, because that's the number that actually decides whether this was smart. Right now all I can tell you is that it was loud.