Three things happened to email over the past two years and most DTC programs have adjusted to one of them. The deliverability floor became a hard gate at three inbox providers rather than a set of best practices. The headline metric everyone still reports became uninterpretable. And the flow stack, the boring part, kept quietly earning the majority of the revenue per send while campaign volume ate the calendar.
The correction is not complicated. Send less to fewer people, measure revenue per recipient instead of opens, and build the flows nobody wants to build.
The sender bar is now a gate at three inboxes, not one
Google and Yahoo moved first. Since 1 February 2024, senders of 5,000 or more messages a day to personal Gmail accounts must have SPF, DKIM and DMARC set up on the sending domain (DMARC policy may be `p=none`), must support one-click unsubscribe via RFC 8058 headers on marketing and subscribed mail with a visible unsubscribe link in the body, must have valid forward and reverse DNS, must transmit over TLS, and must keep the Postmaster-reported spam rate below 0.30% (Google, Email sender guidelines).
Microsoft joined on 5 May 2025. Domains sending 5,000 or more messages a day to outlook.com, hotmail.com and live.com must pass SPF and DKIM and publish DMARC at minimum `p=none` with alignment. The consequence is harsher than Google's: non-compliant mail is rejected with a 550 5.7.15 error, not filed to junk (Microsoft, Outlook's requirements for high-volume senders).
Three practical consequences most programs have not internalised:
0.30% is the cliff, not the target. Google's own guidance is to stay below 0.10% and never reach 0.30%. On a 200,000-send campaign that cliff is 600 complaints away, which is one bad Black Friday send.
Connect Google Postmaster Tools. If you cannot see your spam rate, you are blind on the one metric with a hard enforcement threshold. This takes twenty minutes.
A dedicated, warmed sending subdomain is not optional at volume. Sending marketing from your root domain conflates transactional reputation with promotional reputation.
Open rate is not a metric, and the preview line just got worse
Apple Mail Privacy Protection pre-fetches images and marks an open whether or not a human saw the message. That has been true since 2021 and it makes open rate uninterpretable for any list with meaningful Apple share. The scale of the problem is circular: Litmus put Apple at 51.52% of tracked email opens in February 2026 across more than 1.1 billion opens (Litmus, Email Client Market Share). Those are Litmus-tracked opens, the very metric MPP distorts, so read it as an order of magnitude, not a precise share.
iOS 26 made the second-order problem worse. With Apple Intelligence on, Mail replaces the preview line under each sender with a generated summary of the message. Assume the inbox line is now an AI paraphrase of your email's content: the content itself has to carry the offer clearly and early, and a decorative preheader gets thrown away.
Report click rate, placed-order rate, and revenue per recipient. Disclose the attribution window on every figure, every time. Klaviyo's default attribution will not match Shopify's, GA4's, or your books. Reconcile once, document the delta, stop re-litigating it monthly.
List quality is a revenue decision, not hygiene
The sunset conversation always sounds like a compliance chore and it is actually a margin decision. Sending to unengaged subscribers to hit a revenue number in November raises your complaint rate, which lowers inbox placement, which reduces the revenue you earn from the engaged segment in January. You are borrowing from next quarter at a bad rate.
Three things that should be written down and running:
An engagement-based sending policy. Default audience is engaged; the unengaged window widens only for specific, deliberate sends.
A sunset flow. Re-permission, then suppress. Actually suppress, not "tag and keep sending."
A capture standard. Every list-growth mechanic gets checked for consent language, disclosed terms, and whether the subscriber will recognise you in 30 days.
SMS: the rules loosened, the costs did not
The regulatory picture changed twice and both changes are widely misunderstood.
One-to-one consent is gone. The Eleventh Circuit vacated the FCC's one-to-one consent rule on 24 January 2025, holding it exceeded the agency's statutory authority, and the FCC formally eliminated it later in 2025 (Wiley; Kelley Drye). A single opt-in can cover multiple sellers if the consent language is clear. This does not weaken the underlying requirement for prior express written consent for promotional texts.
Revocation got stricter. Rules effective 11 April 2025 require that opt-out requests be honoured through any reasonable method (including "stop," "quit," "end," "revoke," "cancel," "unsubscribe," a website form, an email to a business address, or a voicemail), that revocation apply across both calls and texts regardless of the channel it arrived on, that it be honoured as soon as practicable and within ten business days, and that at most one clarification message be sent within five minutes, containing no marketing content (Bryan Cave Leighton Paisner).
If your SMS platform only recognises exact-match "STOP," you have a gap. Audit the keyword list.
The economics. Pass-through fees keep climbing. The Campaign Registry raised 10DLC brand registration to $4.50 and standard brand vetting to $41.50 effective 1 August 2025. Carrier surcharges sit around $0.003-$0.005 per message and rose again in January 2026, with T-Mobile and US Cellular both increasing A2P pass-through fees. On top sits your platform's own send cost; Klaviyo's US SMS rate is quoted at $0.006 per credit. These are vendor-published schedules, not an independent survey, so verify against your own invoice.
The discipline follows: compute revenue per SMS send, not per SMS subscriber. Use SMS for time-sensitive, high-value and transactional-adjacent moments, not a newsletter cadence. Monitor delivery rate, not send count, because carrier filtering drops messages silently.
Honest benchmark ranges for flow revenue share
Here is the number you came for, with the population attached, because a curated set is not a sample.
Klaviyo's 2026 benchmarks, drawn from more than 183,000 Klaviyo customers across ecommerce verticals, report that flows generate about 41% of email revenue from 5.3% of sends, that flow click rate averages 5.58% against 1.69% for campaigns, that flows deliver roughly 13x the placed-order rate of campaigns, and that top-decile flows reach $7.79 revenue per recipient (Klaviyo, 2026 Email Marketing Benchmarks).
Now the caveats that make those figures usable rather than misleading:
The population is brands that chose Klaviyo, mostly DTC, skewing SMB to mid-market. It is not a sample of ecommerce.
These are Klaviyo-attributed figures on Klaviyo's default window. Attributed revenue is not incremental revenue. A welcome flow gets credit for purchases that would have happened anyway.
The repeated "email drives 30% of revenue" claim is an attribution artefact as much as a performance fact, highly sensitive to window length and to how much of your traffic is already high-intent.
Use the Klaviyo figures to sanity-check whether your program is in the right postcode. Use your own trailing 24 months, banded p25/p50/p75 with seasonality removed, to decide anything.
The one durable structural insight in that dataset: flows do the majority of the per-send work on a tiny fraction of the volume, and a larger share of flow revenue comes from first-time buyers than campaign revenue does. If your calendar is 90% campaigns and your flow stack has four flows in it, you are spending your list's attention in the wrong place.
The flow architecture that still earns its keep
The core library, in rough order of contribution: welcome, abandoned checkout, abandoned cart (distinct from checkout), browse abandonment, post-purchase, review request timed to delivery plus a usage window, replenishment for consumables only, cross-sell from real co-purchase data, winback segmented by prior value, VIP recognition before discount, sunset, and the trio that is most often unbuilt and highest performing: back-in-stock, price-drop and low-stock.
The most common structural defect we find in flow audits is not copy. It is suppression. Parallel flows firing at one profile, no explicit exit conditions, browse abandonment competing with abandoned checkout for the same person on the same day. Written exit conditions and cross-flow suppression are what separate a flow stack from a spam machine.
The second most common defect is a segment with nowhere to send people. We built a category-branching signup popup for an apparel client this year that split subscribers three ways and fed a conditional welcome email. One branch pointed at a collection URL that did not exist. The segmentation was elegant and it bought nothing, because a segmented email landing on a generic page is just a slower unsegmented email. Check the destination before you build the branch.
Zero-party data, and AI that helps rather than flattens
Zero-party data is a good idea with a low ceiling: ask only for what you will actually use. A three-option preference question at signup that routes to genuinely differentiated content beats a ten-field profile nobody acts on.
On AI-assisted content, the useful discipline came from Nielsen Norman Group this month: "One AI Output Is an Example, Not an Evaluation" (NN/g, 14 August 2026). One good generated subject line is not evidence the approach works, any more than a winning A/B result you peeked at early is a result. Klaviyo now sells its agents as paid add-ons, so this is a cost line, not a free feature.
AI earns its keep in lifecycle on variant volume for testing, on QA (broken links, missing alt text, unsubstituted merge tags), and on performance summaries. It reliably destroys value on brand voice. If the founder's voice is why people opened the email, do not let a model paraphrase it.
What to do this week
Connect Google Postmaster Tools and look at your spam rate. If it is above 0.10%, that is this quarter's priority over any campaign.
Confirm SPF, DKIM and DMARC pass on your sending subdomain, and that one-click unsubscribe headers are present on marketing sends.
Delete open rate from your reporting template. Replace it with click rate, placed-order rate and revenue per recipient, with the attribution window printed next to each.
Audit your SMS opt-out keywords against the FCC's "any reasonable method" standard, and pull your true cost per send from your invoice.
List your live flows and their exit conditions. Any flow without written exit conditions is a suppression bug waiting for a busy week.
Check that every segmented email has a real destination.
Sources
Google, Email sender guidelines, requirements effective 1 February 2024. https://support.google.com/a/answer/81126
Microsoft, Strengthening Email Ecosystem: Outlook's New Requirements for High-Volume Senders, enforcement from 5 May 2025. https://techcommunity.microsoft.com/blog/microsoftdefenderforoffice365blog/strengthening-email-ecosystem-outlook%e2%80%99s-new-requirements-for-high%e2%80%90volume-senders/4399730
Litmus, Email Client Market Share, February 2026 (1.1bn+ tracked opens). https://www.litmus.com/email-client-market-share
Wiley, 11th Circuit Vacates FCC's One-to-One TCPA Consent Rule, 24 January 2025. https://www.wiley.law/alert-UPDATE-11th-Circuit-Vacates-FCCs-One-to-One-TCPA-Consent-Rule
Kelley Drye, Eleventh Circuit Vacates TCPA 1:1 Consent Rule, 2025. https://www.kelleydrye.com/viewpoints/blogs/ad-law-access/eleventh-circuit-vacates-tcpa-11-consent-rule
Bryan Cave Leighton Paisner, The TCPA's New Opt-Out Rules Take Effect on April 11, 2025, 18 March 2025. https://www.bclplaw.com/en-US/events-insights-news/the-tcpas-new-opt-out-rules-take-effect-on-april-11-2025-what-does-this-mean-for-businesses.html
Klaviyo, 2026 Email Marketing Benchmarks by Industry (183,000+ Klaviyo customers). https://www.klaviyo.com/products/email-marketing/benchmarks
Nielsen Norman Group, One AI Output Is an Example, Not an Evaluation, 14 August 2026. https://www.nngroup.com/articles/
The Campaign Registry 10DLC fee schedule effective 1 August 2025, via carrier and platform documentation. https://signalhouse.io/guides/a2p-carrier-fees-2026/
Salesmsg, SMS, MMS, and A2P 10DLC Carrier Fees, 2026. https://help.salesmessage.com/en/articles/6001981-sms-mms-and-a2p-10dlc-carrier-fees
Apple, Use Apple Intelligence in Mail on iPhone, iOS 26. https://support.apple.com/guide/iphone/use-apple-intelligence-in-mail-iph9ae667055/ios
