What 152 DTC welcome
popups say about the
offer, before Q4 changes it.

Every brand has an
opinion about its popup.
Almost none have looked
at anyone else's.

The welcome popup is the highest-traffic piece of marketing most DTC brands own and the least benchmarked. Everyone has a view on it, usually inherited from whichever agency set up Klaviyo, and almost nobody has opened fifty competitors' sites in a fresh browser and written down what happened. So I did, for 235 of them, and the result is a tracker on The Index that will get a second column the week of Black Friday.

The brands are the tier-one and tier-two rows of The Handoff directory, which is to say the brands whose ads a practitioner has cited or whose creative sits in a serious swipe file. A headless browser loaded each homepage from a US address, in a fresh session with no cookies, on a 1440 by 900 desktop and a 390 by 844 phone. It waited up to 32 seconds for a timed popup, scrolled, and on desktop pushed the cursor out through the top of the window for exit intent. Nothing was typed into any form.

Where a popup opened with a "Yes, please" or a quiz question, one click was allowed to reveal the field, and no more. That matters for reading the numbers: "email first" below means the first field the visitor actually sees, not what the brand's flow eventually asks for. A handful of sites refuse data-centre traffic; those were checked by hand in a normal browser and are marked as such on the tracker.

One caveat I will repeat because it shapes every figure here: this is a first visit, in one session, in the first week of September. The 76 brands that showed nothing are not all abstainers. Klaviyo's default timer is 30 seconds on site, Alia caps frequency, and plenty of flows fire on the second pageview. The tracker lists what was loaded on each of those pages so you can tell a deliberate quiet from a slow one.

Percent off wins, and
15 is the number.

In September 2026, 84 of the 152 welcome popups I captured (55%) state a percentage off, and the median is 15%. 21 brands say 15%, 19 say 10%, 14 say 20%. Above 20 the offers thin out fast and change character: the 40% at Tushy, the 50% at Cuts, the 60% at True Classic and the 67% at Mars are Labor Day sale headlines with an email field underneath, not standing welcome offers. Fifteen popups in the set carried the words "Labor Day" in the first pane, the same sale-wearing-the-welcome-offer pattern I found on 243 ad-to-page teardowns in August.

Dollar offers are rare and they are not random. Avocado Green Mattress offers $100 off a furniture order of $1,000 or more; Feno offers $100 off a toothbrush system; Ernesta puts $50 toward a rug; Quince gives $40 off a $250 order, in the app only. On a $1,000 ticket, 10% reads as a bigger gift than the brand wants to give, and a fixed dollar figure caps the cost while still anchoring. At the other end, Mid-Day Squares gives $5 and Jones gives $10, because on a snack or a mixer a small percentage looks like nothing.

Then there is the group that has stopped stating a number at all. Twenty-nine brands, 19% of the set, lead with a mystery discount, a spin, a scratch card or a sweepstake. Eighteen of those twenty-nine run on Alia, which tells you this is partly a template decision. It is also a margin decision: a mystery discount lets the brand hand out 5% to most people and 20% to a few, and the visitor plays for the 20%.

Only 36 of the 152 popups say "first order" in the pane the visitor sees. Most brands leave the condition to the code, which is fine until the code gets used twice. Four tie the offer to a subscription (AG1, Arrae, David, Cometeer), which is the cleanest version of the trade: a bigger number, 20 to 30 percent, in exchange for a recurring order rather than a one-off.

Seventy-six popups ask a
question before they
ask for anything.

In September 2026, 76 of the 152 popups (50%) show no field on the first pane. 43 open with a segmentation question, "What's your hair goal?" or "Who are you shopping for today?", and 33 open with a yes/no, "Want 20% off?" with a decline link underneath. The field appears on the second pane, after the click. Two years ago this was a minority pattern; at the top of the market it is now the coin flip.

The two gates are doing different jobs. The yes/no is a conversion mechanic: the first click is a micro-commitment, and a visitor who has said yes to the offer is more likely to finish the form than one who was shown the form cold. Attentive, Brooklinen, Bombas, Liquid I.V. and Eberjey all run it. The quiz gate is a data play. AG1 asks your top health focus; Act + Acre asks your hair concern; Fly By Jing asks how spicy you like it. The answer lands on the profile before the email does, and it becomes the segment the brand emails for the next year, which is where the welcome flow either earns its keep or sends everyone the same five emails.

Twenty-eight of the 43 quiz gates run on Alia, and that is the story of this dataset in one line. Alia is on 73 of the 235 sites and rendered 56 of the 152 popups, level with Klaviyo's 55, on a much smaller install base. Brands keep Klaviyo for the list and hand the front door to a tool whose whole product is the gate, the mystery reveal and the scratch card. If you have wondered why every wellness brand's popup suddenly asks what problem you are trying to solve, this is why.

Email is still the first
ask. SMS lives on
step two.

Of the 108 popups where a field was visible, 104 asked for email first, 2 asked for email and phone together (Dermalogica, SAXX), and 2 led with a phone number: Metabiotics on the first pane, and HygieneLab behind its quiz. Eleven popups mention texts anywhere in the flow, and in seven of those the phone ask arrives on a second pane after the email is in. Nobody at this end of the market asks for a phone number cold.

That is a more conservative picture than the SMS vendors' decks would suggest, and it matches what I see on the P&L side. Email is the list that pays for itself; SMS earns its 12% click rates on urgency and costs per send, so the brands that run both put the cheap ask first and the expensive one behind it. The "sign up for email and texts" line on the button is doing quiet work on the consent side: the button text is the disclosure, and the phone field only appears once the visitor has already said yes to something.

The 42 popups marked "behind the gate" on the tracker are the quiz and game flows where a single click did not reveal a field. From the ones that did reveal, the pattern is consistent: Alia's second pane is an email field, and its third, where there is one, is the phone. If your welcome flow asks for the phone first, you are running a test the top of the market has already run.

Seven seconds in, a
medium card on desktop,
a takeover on the phone.

In September 2026, 114 of the 152 popups (75%) fired on a timer, and the timer is tight: a quarter fired by 5.5 seconds, the median was 7.3, three quarters had fired by 10.8. Thirteen appeared inside 2.5 seconds of load. Twenty-one waited for a scroll, or for a 25-second timer the capture cannot tell apart from one. Four waited for exit intent: Happy Mammoth, Nice Laundry, Pickie and Zesty Paws, all between 31 and 46 seconds in.

The reflexive complaint about popups, that they hit before the page has painted, is a minority case here. What is more striking is that the delay is the same on the phone. Vendors ship one timing rule and brands rarely split it by device, so the mobile visitor, still watching the hero load on a cellular connection, gets the popup at roughly the same wall-clock moment as the desktop visitor who has already read the fold.

Size is the card itself, not the dimmed backdrop. On desktop the centre of the distribution is a medium card covering 20 to 45% of the window. On a phone the same creative becomes a large card or a takeover, because the vendor scales the card to the width and the copy does not shrink to match. Thirty-one desktop popups leave most of the page readable; on mobile that number is five. If you have only ever looked at your popup in the vendor's desktop preview, you have not seen what most of your visitors see.

Two vendors, one voice,
and the jokes live in
the decline link.

Klaviyo is loaded on 185 of the 235 sites, and Alia and Klaviyo between them rendered 111 of the 152 popups (56 and 55). Attentive rendered 19, almost all yes/no gates with an SMS second step, which is the shape its per-message pricing rewards. Fifteen popups were custom builds, and they belong to the biggest brands in the set: Bombas, True Classic, Eight Sleep, Fabletics. At scale the popup becomes product surface and leaves the app. Postscript is on 48 sites and rendered 2 popups the capture could read, which is not a verdict on Postscript; it is a note that most brands running it use it for the SMS list and let another tool own the door.

I read every headline. Twenty-three of the 152 have a voice you would recognise as the brand's. Liquid Death: "Wanna cheat on your other beverage company?" Poppi: "sour apple is here. sweet." Candy Kittens: "Looks like you're new here!" Bero: "Fancy a free tote?" The other 129 say "Get 15% off your first order" in one of about four ways, and that is not a criticism. The plain version converts, and the templates push you toward it.

Where personality shows up most reliably is the decline link. Twelve brands write the "no" as copy rather than "No thanks": Sunday Citizen's "No, I'll pay full price", Beast's "no thanks, I'd rather not save", MindBelly's "Nah, I'm good", Nutrops' "No thanks, I don't like savings", Unplugs' "I'll keep paying full price". It costs nothing, it is the one place the template lets you be a brand, and it is the cheapest A/B test on the site.

1. Write the decline link. "No thanks" is the template's copy, not yours. The twelve brands that wrote theirs are on the tracker; steal the shape, not the line.
2. Look at the phone first. Open your own popup on a 390-pixel screen before you approve it. Sixty-seven of 141 mobile popups in this set are large cards and 17 are takeovers, and most of those were medium on the desktop preview.
3. Decide what the quiz is for. If you gate the field behind a question, make the answers segments you will actually email by. Twenty-eight of the 43 quiz gates are Alia templates; the question is the only part that is yours.
4. Put the cheap ask first. Email, then phone, on a second pane, with the disclosure on the button. That is what 104 of 108 visible forms do, and the four that do not are running a test you do not need to repeat.
5. Know which offer you are running. A standing 15% and a Labor Day 45% are different products. Fifteen brands in this set had the sale wearing the welcome popup's clothes, and the BFCM recapture will show how many more do it in November.

Below the top of the
market it is 10% off,
Klaviyo, and a plain field.

The 152 popups above are the top of the market. The Handoff directory has another 2,482 rows below them, mostly from the 1800dtc harvest, and the same capture ran on all of them in September 2026: 2,365 loaded, 1,188 showed a capture popup (50%, against 65% at the top). Same rules, same viewports, same first visit.

Three things separate the wide market from the top of it. The offer is smaller: the mode drops from 15% to 10% (271 brands say 10%, 204 say 15%), though the median holds at 15%. The gate is rarer: 362 of 1,188 popups (30%) hide the field behind a quiz or a yes/no, against 50% at the top, and the quiz specifically falls from 28% of popups to 12%. And the stack is Klaviyo's: it rendered 589 of the 1,188 (50%) where Alia rendered 206 (17%); at the top the two are level. Alia is a tier-one habit that has not reached the long tail yet.

What does not change is the interesting part. Email is first almost everywhere (933 of 983 visible forms), texts are a second-step ask (95 of 1,188 mention them), the popup is timed rather than on load, and the copy is plain. The wide market is the top of the market with the same defaults and fewer decisions layered on top. That is the whole lesson: the brands that are winning did not find a different popup, they made three more choices about the same one. The offer they picked on purpose, the question they gate it with, and the words on the decline link.

Same script, same 235
brands, the week of
Black Friday.

This baseline exists for its second column. The capture re-runs on the same brands with the same rules the week of Black Friday, and the comparison renders on the tracker: who raised the percentage, who swapped the standing welcome offer for a sale headline, who added a phone field for the season, and who turned the popup off because the site-wide banner was doing the work. It is the one line of the peak-season playbook that almost every brand changes and almost none of them measure. My expectation, on the record, is that the median moves from 15% to 20%, the "up to" rows triple, and the gated flows stay exactly where they are, because nobody rebuilds a quiz in November.

For the brands reading this before their own Q4 build, the useful move is narrow. Find the five brands in your category on the tracker, open their mobile captures, and compare the offer and the first ask against yours. If you are the only one still stating 10% on a plain email field, you are not wrong, but you are now the control group. And if the welcome popup is one of a dozen things you want read against the market before the quarter, that is what The Read is for; the tracker stays free either way.

Questions brands ask
about the welcome
popup.