Click-to-order lag: judge spend on the right day

By the HonestTag team · Published July 12, 2026

Every "how did ads do this week?" question contains a hidden assumption: that this week's spend produces this week's revenue. For a few stores that is roughly true. For most, it is not, and the gap between when you spend and when the order lands is where good campaigns get cut for the crime of being on schedule.

What click-to-order lag is

For every new customer with a recorded first ad click, HonestTag records the days between that click and their first order, per channel, in six buckets: same day, 1-3 days, 4-7, 8-14, 15-30, and 31+ days. Today the app uses that distribution to set each channel's verdict re-check date (Truth and up); a per-channel lag chart is not in the app yet.

The distribution is the point. Two channels can share a median and behave completely differently: one converts everyone inside three days; the other converts half same-day and the rest over a month. You judge those two channels on different timelines, and the buckets are what capture that.

The silent same-day assumption

Judging this week's spend by this week's revenue assumes same-day conversion without saying so. Suppose your Google buyers take a median of 9 days from click to order. Then Tuesday's spend cannot be judged on Friday. The revenue it will produce has not arrived yet. It is not missing; it is in transit.

Here is the failure mode in slow motion. You raise budget on a channel with a 9-day median lag. For the first week, spend is up and revenue is flat, so the ratio looks worse every day. On day 6 you lose your nerve and cut. The orders from that spend arrive over the following two weeks, land on a smaller spend base, and the ratio suddenly looks great, which teaches you exactly the wrong lesson. Cutting inside the window punishes spend for lag, not performance.

How lag shapes what you read

1. One lag window per change. A change read before one full lag window has passed is read before its evidence exists. If the median lag is 9 days, the revenue from a change made on the 1st has not landed by the 10th.
2. Each channel against its own lag. A universal 7-day reporting window flatters short-lag channels and slanders long-lag ones. Weekly numbers such as NMER and new-customer counts, read over a channel's own window differ from the same numbers read over a shared one.
3. Test windows that do not straddle the judgment date. When a promo or creative test's lag window crosses the day it is reviewed, part of its orders are still in flight. A sale that ends two days before review still has half its orders arriving after the grade.

Lag sets the verdict re-check date

In the app, lag feeds the judgment layer. Each channel's weekly verdict comes with a re-check date, and that date is computed from the channel's own median lag plus 3 days, kept between 7 and 14 days, once the channel has 20 or more measured orders. Before that volume exists, a 7-day default applies.

That construction matters. The check-in date is one your store's own conversion behavior justifies, not an asserted default that happens to match a weekly email cadence. A channel where buyers deliberate for two weeks gets two weeks; a channel that converts same-day gets checked sooner. Mechanics are on the product page.

The dishonest version

Real-time ROAS dashboards imply that today's spend performance is knowable today. For any store with meaningful lag, it is not. The number on the screen is a partial count dressed up as a result, and it will drift upward for days as late orders attach. Refreshing it hourly does not make it truer; it makes it more persuasive.

The subtler version is the silently chosen attribution window: a 7-day click window on a product people take 20 days to buy. Every conversion past day seven simply vanishes from the report, the channel looks weak, and nobody chose that outcome on purpose. The default chose it. An honest tool shows you the lag distribution first and lets the window answer to the data.

Honest limits

Two caveats we would rather state than have you discover. First, lag is measured from the recorded first click, so it exists only for orders that have one; a new customer who arrived with no measurable click is not in the distribution. Second, distributions need volume before medians mean much. A channel with 11 orders does not have a lag curve; it has 11 anecdotes. That is exactly why the verdict re-check math waits for 20 or more measured orders before trusting a channel's own median.

How HonestTag measures it

Everything starts from your store's own Shopify order data. Every order is classified new or returning against store order history, new customers attribute to the first ad click, and every attribution keeps a proof record (click id, timestamps, match method), so the lag between click and order is a stored fact, not an estimate. Google Ads and Microsoft Advertising receive the available retraction or restatement for a refund, and Klaviyo and Google Analytics 4 receive a refund event. HonestTag does not send Meta, TikTok, Pinterest, Snapchat, Reddit or OpenAI Ads a refund or retraction event; the refund is recorded in order proof. The formulas, including the re-check clamp, are published in numbers on the in-app methodology page.

Frequently asked questions