Your Feature Announcements Are Written in a Language Your Customers Don’t Speak
Company Newsletter

Your Feature Announcements Are Written in a Language Your Customers Don’t Speak

Picture the feature your team fought hardest for this year. The one that took two quarters, ate a sprint plan alive, and finally shipped on a Thursday afternoon to quiet celebration in the team channel.

You wrote the feature announcement yourself. “Introducing advanced workspace permissions with role-based access controls.” You listed the capabilities, linked the docs, and hit send on the update email. Then you waited for the replies.

Silence.

Three weeks later, a support ticket arrived from one of your best accounts. “Could we get some way to stop the whole team from editing the billing settings?” The exact job your feature does, described by a customer who received your announcement, opened it, and felt nothing.

Months of engineering, one paragraph of announcement, zero recognition.

My cofounder Cagri made the case this week that SaaS companies pay dearly for silence, and that a feature your customers never heard about performs like a feature you never built. His post answers whether to tell your customers what you shipped, and how often. This post answers the harder question underneath his. Why do customers so often feel nothing when you do tell them?

Why Do Customers Ignore Your Feature Announcements?

The announcement about workspace permissions was accurate. Every word of it was true. It was also written in a language only your team speaks.

In 1990, a Stanford graduate student named Elizabeth Newton ran an experiment that explains your empty reply column better than any marketing audit. She split participants into tappers and listeners. Each tapper picked a well-known song, something like “Happy Birthday,” and tapped its rhythm on a table while a listener tried to name the tune. Before the guessing began, Newton asked the tappers to predict their listeners’ success rate, and they predicted 50%.

Across 120 tapped songs, the listeners named three. That’s 2.5%.

Read that number again, then reread your last release note.

The gap between 50% and 2.5% exists because the tappers had the melody playing in their heads while they tapped. To them, every knock on the table carried the whole song. The listeners heard knuckles on wood.

Your team has been humming the melody of that feature for months. You sat through the roadmap debates, the edge cases, the naming meeting. By launch day, the phrase “granular permissions” carries the entire song for everyone in the building. Your customer hears taps.

Unfair? Completely. Your team wrote that announcement in a state of total fluency, and fluency is the one thing your reader lacks.

Psychologists call this the curse of knowledge. Once you know something, imagining the mind that lacks it becomes nearly impossible. The curse writes most of the world’s changelogs, and it explains why the smartest product teams often produce the least felt announcements. Expertise is the infection route.

I’ve written about the insecurity that makes company newsletters sound corporate. The curse of knowledge is the second disease, and it hits confident teams hardest. Jargon born of insecurity wants to impress. Jargon born of fluency simply forgets that anyone lives outside the building.

What Language Do Your Customers Actually Speak?

The language of their own Tuesday.

There’s an old line in marketing, repeated so often that tracing its origin became a full research project. Harvard’s Theodore Levitt loved telling his students that people buy a quarter-inch drill because they want a quarter-inch hole, a line Levitt himself credited to the ad man Leo McGivena. Millions of drills sold, and every buyer was picturing a hole.

Your feature is the drill. An announcement your customer can feel describes the hole. And the hole always lives inside her week: the report that builds itself before the Monday meeting, the intern who can no longer touch the billing settings, the renewal call where your champion finally looks prepared.

Kathy Sierra built a whole philosophy of product on this shift. In Badass: Making Users Awesome, she argues that the products people evangelize are the ones that upgrade their users, because nobody brags about a tool. People brag about what they can now do.

An announcement is the same craft in miniature. Describe the upgraded customer and she recognizes herself in the first line. Describe the upgraded product and you’ve written a mirror for your own team.

Customers fall in love with the Tuesday a feature hands back to them. The feature itself is the delivery van.

The evidence here runs deeper than taste. Marketing researchers Grant Packard and Jonah Berger analyzed over 1,000 real customer conversations and found that customers were more satisfied, more willing to buy, and actually spent more when companies spoke to them in concrete language. A one standard deviation increase in concreteness lifted satisfaction by about 9% and actual spending by at least 13%.

Their explanation should hang above every product marketer’s desk. Concrete words signal that you’re listening. “Your March invoices” proves attention in a way “enhanced billing functionality” never will.

A changelog entry written in product language reads like a postcard from the factory floor. Accurate, dated, and addressed to no one in particular.

How Do You Announce a New Feature So Customers Actually Care?

Three moves carry most of the translation, and all three fit inside the announcement you were already going to write.

Make the customer the subject of the sentence. “You can now lock billing settings to two people” puts your reader inside the sentence, doing something she has wanted to do for months. “We shipped advanced permissions” puts your team in the sentence, taking a bow. Same feature, same length, entirely different reader.

Grammar is the cheapest empathy tool you own.

Borrow the words from your support tickets. Your customers have already written this announcement for you. They wrote it in the tickets, the exit surveys, and the sales calls, phrased in their own vocabulary. “Stop the whole team from editing billing” is a finished headline sitting in your help desk right now.

Support tickets are your customers drafting your feature announcements for you, in their own words, months in advance.

Give each feature one scene. A before moment and an after moment, two sentences each. The controller who used to export three spreadsheets on the last Friday of every month now opens one dashboard on Friday morning and leaves at five. Your reader will remember that controller long after she’s forgotten the feature’s name.

A scene like this is also exactly the material a generic AI model can never supply, because the model never met your customers. I’ve cataloged what that erasure costs elsewhere. The scene has to come from someone who has heard the tickets and the calls, which is the reason translation is a craft in the first place.

You aren’t alone if every draft still comes out in the builder’s tongue first. Every product team we talk to writes the feature version before the customer version. The craft lives entirely in the second pass.

One rule guards the whole exercise. If a sentence in your announcement would survive under your competitor’s logo, it still speaks the builder’s language.

The Subject Swap: A 20 Minute Test for Your Last Announcement

You can measure your own translation gap this week with a printout and a pen. Run the five steps in order.

Step 1: Print your most recent feature announcement. The email, the changelog entry, or the update section of your last customer newsletter. One send is enough.

Step 2: Underline the grammatical subject of every sentence. The noun doing the verb. Skip greetings and signoffs.

Step 3: Sort every underline into three columns. Builder: we, our team, the product, the dashboard, the new release. Customer: you, your team, your reports, your month-end close. Ghost: abstractions that belong to nobody, like “this update” or “efficiency.”

Step 4: Count the columns. Most first drafts run heavy on Builder and Ghost, and the ratio is the measurement. An announcement where the Customer column wins is an announcement written in your reader’s language. Anything else is tapping.

Step 5: Rewrite the three sentences tied most directly to revenue until the Customer column owns them. The subject becomes your reader. The verb becomes something she does in her actual week.

The swap takes twenty minutes and costs nothing. Run it on every send for a quarter and the ratio starts drifting on its own, because once you can see the columns, you can hear the language while you write. The exercise works with any toolchain, and with none at all.

Who Does the Translating, Month After Month?

The craft is learnable. The calendar is the killer.

The founder speaks the customer’s language fluently, because the founder sat on the sales calls and read every early ticket. The founder also has no Thursday afternoons left. The marketer has the Thursday afternoons and joined two years after those sales calls happened. Cagri named the crack this work falls into: at most SaaS companies under 50 people, telling customers what got built belongs to nobody.

The HeyNews Company Newsletter was built to live in that crack. The engine studies how your company already sounds, from your site, your docs, and your past sends, and it learns your customers’ vocabulary alongside your own. Point it at your changelog feed and your blog, and every shipped feature becomes a scored story candidate for the next issue.

On your cadence, a drafted issue arrives with the translation work already begun. The feature, the job it does, the customer scene. Automations draft on schedule and hold everything for your review, and nothing sends without your approval, on every single issue. You read the draft the way your most demanding customer would, adjust the three sentences that matter, and approve.

There are two ways to run it. Keep production in house and use the platform as your drafting engine, with your team holding the pen. Or hand the recurring production to our team, keep the final read, and spend your minutes on judgment alone. Either way, the loop closes itself: open and click data flows back into the system, and future drafts learn which translations your customers actually reward.

Remember the tappers? Their fix was a singer. Volume had nothing to do with it.

If you’d sooner start small, the standard HeyNews platform runs the same engine on your own list. Connect what you’ve already published, let the AI Writer learn your voice, and refine each draft in plain language until it sounds like you on a good day. The 14 day free trial includes five drafts, which is enough to run the Subject Swap against a machine draft of your own next announcement and judge the translation with your own eyes.

It All Comes Down To…

  • Customers ignore feature announcements written in the builder’s language, and the curse of knowledge guarantees fluent teams write that way by default. Newton’s tappers predicted 50% recognition and got 2.5%.
  • Your customers speak the language of their own week. Describe the hole, the upgraded Tuesday, the scene after the feature. Concrete language also pays measurably: satisfaction rose about 9% and spending at least 13% when companies spoke concretely.
  • Your support tickets already contain your next announcement, phrased in customer vocabulary. Borrow the words before you invent new ones.
  • Run the Subject Swap this week: underline every grammatical subject in your last announcement, sort them into Builder, Customer, and Ghost, then rewrite the three revenue sentences until the Customer column owns them. No tool required.
  • Whoever drafts the issue, one rule protects the relationship: a human who knows the customers reads every send before it travels.

The Ticket That Answered Your Announcement

Go back to the feature your team fought for. It’s still there, still solving the exact problem your best account described in that support ticket three weeks after launch. The distance between the two was always one sentence long, written in the wrong language.

Your customers keep handing you the phrasebook. Their tickets, their calls, and their renewal questions hold every word your next announcement needs. Translate the next send into their language, and the silence after you press the button starts turning into replies.

See how the Company Newsletter carries the translation every month while your team keeps the final read, or book a call and walk through your changelog together. Already sending email and ready to hold the pen yourself? Start the 14-day free trial.

Eren Daşkesen, Co-founder of HeyNews

Eren Daşkesen

Co-founder & Chief Creator Officer of HeyNews. Eren wrote the novel "Kürek," managed projects for 15+ years, and now spends his time teaching AI to write like a person, not a press release. He brings a background in marketing and brand management, and his main job at HeyNews is making sure the AI output reads like something a human would actually want to send.

Try HeyNews free for 14 days

Start Free Trial