Your delivery rate can look healthy while Gmail clicks and sales quietly fall. Check this before increasing your Black Friday sends.

As the November email calendar fills up, I keep coming back to one question: are the numbers in the campaign report telling the whole story?

The latest Gmail figures made me stop. An email can be marked as delivered and still land in spam, so the report can look healthy while Gmail clicks and revenue quietly fall. I would check that before adding another Black Friday send.

In Validity’s data, average Gmail inbox placement fell from 90.97% in Q1 to 81.57% in Q3 2026. That is a drop of 9.4 percentage points. Spam placement rose from 5.4% to 9.5% over the year.

Those figures do not prove that your emails have the same problem. They are a good reason to check.

Take your last three comparable campaigns and split the results by mailbox provider. Compare Gmail clicks, orders and revenue per recipient with the rest of the list. Do not stop at the delivery rate.

If Gmail is slipping, check whether the audience or sending frequency changed. Also verify that your setup meets Google’s sender requirements.

I would do that before changing the offer or rewriting every email. Sending more often to the same unresponsive people can make the problem worse.

DoorDash’s October 29 earnings call offered a quieter picture than many AI-shopping headlines. CEO Tony Xu said third-party agents had generated only “several hundred orders” combined since they began operating. You can read the original wording in DoorDash’s Q3 2026 earnings-call transcript filed with the SEC.

Agent shopping matters, but the headline does not tell you how much it matters to your store.

Start with Shopify’s AI-channel reports, if they are available to you. Check what the figures include. A shopper who clicks an AI recommendation and completes the purchase on your site is different from an agent completing the purchase.

Before building a strategy around AI shoppers, find out whether they are already bringing you orders and what kind of orders they are.

Knowing where an order came from leaves another question: did that marketing create a sale that would otherwise not have happened?

A coupon partner can collect the last click from someone who was already about to buy. The order gets attributed to the partner, but how much did the partner add?

A holdout test gives you a better indication. Leave part of a comparable audience out of one promotion, then compare purchases across both groups.

An illustrative result: 5,000 subscribers receive the promotion and place 250 orders. Another 5,000 eligible subscribers do not receive it and place 200. The observed difference is 50 orders, not all 250.

Check revenue and margin too. Extra discounted orders are not automatically extra profit.

For a first test, choose one promotion and randomly split eligible subscribers into a send group and a holdout group. Keep their other marketing consistent and agree on the measurement period before sending.

Look past the headline numbers.

“Delivered” does not necessarily mean inboxed. An AI-attributed order was not necessarily completed by an agent. An attributed order was not necessarily created by the campaign.

Start with Gmail. Compare your last three campaigns and see whether clicks and revenue per recipient are holding up. If they are slipping, investigate before increasing your sending frequency.

I would rather find the problem now than after the biggest campaigns have gone out.

Corné
Emercify

Do customers ever ask, “Which one is right for me?”

Other shoppers may have the same question, but leave without asking.

We have a way to help Shopify stores understand what those visitors need and help them make a choice.

We’re offering a complimentary test period to a few stores where it could be a good fit.

If you sell products people need help choosing, reply with your store URL and I’ll take a look.