Two very different worlds collided this week: newsletter people obsessing over open rates and AI search people obsessing over citations, and it turns out they're fighting the same battle. Consistency wins in both. A newsletter writer who sent 366 daily emails without missing one, the newsletters hitting 55% open rates by doing nothing cleverer than showing up on schedule, and our own benchmark of 10,000 local businesses proving that ChatGPT "knowing" a business is worth roughly nothing if it never recommends it. Here are your five.
1. What 366 days of daily emails taught one newsletter writer (that most weekly ones never learn)
Ephraim Gopin spent all of 2024 sending a daily email. 366 issues, not one skipped, for a non-profit newsletter called Email 366. Properly relentless stuff. He wrote everything two to six weeks ahead so he was never scrambling on the day it went out, built in recurring weekly slots (a guest expert on Sundays, something lighter on Fridays) so he wasn't carrying the whole thing solo, and signed off every single day with the same line, "Back at it tomorrow," until skipping it became such a big deal that readers started emailing to ask what was wrong. Here's the bit that should make you feel better about your own hit rate: Gopin's own rule was that one in three posts landing well is a good year, not a failure. Two out of three can be perfectly ordinary and you're still winning. The organising and the recurring structure are what let him hit that bar 366 times without burning out, not raw talent or an unhealthy amount of coffee. Inbox Collective
2. The newsletters hitting 55% open rates aren't doing anything clever, they're just relentlessly consistent
Beehiiv dug into its best-performing newsletters (1440, Marketing Examined, TIME, among others), which collectively earned creators over $25 million in 2025. Top performers average 38% open rates against a 20% industry benchmark most of us would happily settle for, and the outliers hit 55%. The common thread isn't production value, it's boring old repetition: the same format, the same slot, every single time, subject lines under 20 characters, and a main call to action placed roughly 60% through rather than buried at the bottom. So no, a redesign won't fix flagging opens. Consistency will, and it's a lot cheaper than a rebrand. Readers reward routine over polish: 1440 hits inboxes daily without fail, and that predictability is the entire trust mechanism, not the layout. beehiiv
3. AI search cites listicles and comparison pages more than anything else, so stop burying them
AI sessions are now 56% the size of all search traffic, so this isn't a niche concern any more. Reads to Leads has been tracking what actually gets cited across client work, and "best X" listicles make up 43.8% of all page types ChatGPT cites, with comparison or alternatives pages accounting for another 32.9% of LLM citations across a 177 million datapoint study. Freshness beats quality too, which stings a bit: a six-month-old article that's never been touched loses to a well-updated competitor page, regardless of which one's actually better written. The fix isn't a brand new content strategy, thankfully. It's updating what already exists, and writing the comparison page AI will otherwise write itself anyway, usually with half the facts wrong. Reads to Leads
4. Google will finally show you AI Overview impressions, just don't mistake them for a scoreboard
Google's rolled out a dedicated Generative AI performance report inside Search Console, showing impressions from AI Overviews and AI Mode broken down by page, country, device and date. Handy, until you notice what it doesn't show: queries, clicks, click-through rate, or which passage actually got used. An impression only counts once a link's been scrolled or expanded into view too, so it isn't even counted the same way as an ordinary search impression. AI and organic impressions shouldn't get blended into one tidy "visibility" number, and a graph moving three days later shouldn't get credited to whatever you changed last Tuesday. There's too much noise (seasonality, competing sources, Google quietly shifting its own systems) for that kind of claim to survive scrutiny. Search Engine Journal
5. We checked 10,000 local businesses in ChatGPT and Perplexity, and "known" doesn't mean "recommended"
We ran our own AI Visibility Benchmark: 10,000 US local businesses through ChatGPT and Perplexity. 93.59% were known to ChatGPT, so outright invisibility is rare, but only 55.58% of those results fully matched the business's actual Google Business Profile details. And before anyone celebrates early: 40.88% were never recommended by ChatGPT in unbranded, intent-driven searches (68.79% for Perplexity), even once they were known about and described positively. A good write-up isn't the win it looks like. AI rates the businesses it knows positively 79.8% of the time, so that's basically table stakes, not proof it'll actually put a business forward when someone asks for a recommendation. The one signal that did predict it: review volume. 133.4 average reviews for businesses surfaced by both platforms, versus just 10.7 for the ones surfaced by neither. Reviews, not vibes. Insites
Small business shout-out
This week's shout-out goes to Storyville Letters, which sends subscribers a mystery through the post: 24 real letters over 12 months, stuffed, stamped and hand-sealed, each one carrying the next chapter, clue, map or artefact in a story that unfolds two envelopes at a time. Founder Haley Jackson spent years building story worlds for film, TV, museums and VR, with credits stretching from NASA and the Smithsonian to James Cameron and Barbie, before starting Storyville after watching how much a single letter could mean to her mother during Covid lockdowns. Know a small business we should feature? Reply and let us know.
