How I'd launch an institutional DeFi brand again

On 7 October we took Hare by GSR, a credit desk for onchain finance, from no name and no account to a public launch on a conference stage, a CoinDesk story and a day of partner and press coverage. This is my breakdown of the method: what we did in each step, why, and what I'd keep next time.

Pedro Verdades · October 2026

Disclosure: I run marketing at Turtle, which is a partner in Hare. Everything below is public record. No budgets, no internal numbers.

The stage before the reveal.

If you only read one thing

Audit the category before you draw a logo. Find the uniform, then the whitespace.
Write the positioning down as decisions with dates. One line, one parent, one audience order.
The founder's answers are the voice. Don't invent a tone, transcribe one.
Build templates by post type, static and motion, before launch. Not assets by occasion.
Nothing public before the stage. Holding page, anonymous account, no partner marks until cleared.
Brief every partner under embargo with their role in one sentence and a post already written for them.
Work with a PR firm. Their job is the embargo, the journalists and the first story.
Earned first, owned second, paid last and smallest.
Post when the desks you want are awake, and only after the outlet has published.
Have a community before the day. The first hour is carried by people who already care.

01I audit the room before I design anything

Most brand projects start with a mood board. I start with a spreadsheet. Before Hare had a name we pulled ninety days of posts from sixteen accounts: the vault curators and credit desks it would sit next to, and the protocols it would launch on. Around 1,800 posts, each tagged by intent, format and who it was written for, then read for two things: what performed, and what the category looks like when you squint.

Three findings did most of the work. Name-led launch posts beat thesis posts, and thesis posts do badly for accounts under fifty thousand followers, which every new brand is. Methodology shown beats methodology claimed: the few posts that walked through how a position was assessed carried the strongest signal in the whole set, and they were a small fraction of what anyone publishes. And the category wears a uniform: landscape cards, abstract marks, institutional register, and a constant leak into the host protocol's dashboards.

Once you can see the uniform, the whitespace is obvious. A warm, chromatic primary in a room full of navy and black. House charts instead of screenshots of someone else's analytics. Motion, which almost nobody in the category uses. A desk that explains its underwriting in public. I think this step is where most crypto brands lose: they skip it, so they end up wearing the uniform without having chosen to.

What I'd keep. Rank the competitors on distinctiveness, consistency and sophistication, separately for site, social and blog. Write the uniform down in one paragraph. Pick two or three places to break it on purpose. Every later decision gets easier.

02Positioning is a list of decisions, each with a date

We settled the positioning as a short list before the design went final, and I'd argue the list matters more than the deck it ends up in. The line: the credit desk for onchain finance. The parentage: Hare by GSR, with GSR named as the principal partner on every first mention. The audience order: allocators first, issuers second, because an allocator reads a vault like a credit book and asks two questions, what happens when everyone wants out at once, and am I being paid for the risk. Two products with their partners named exactly. And a list of retired lines that nobody was allowed to use again.

The voice I didn't invent. Connor Milner, Hare's CEO, answered five questions for a launch-day interview with GSR: why now, where the gap in DeFi credit risk is, what happens on the worst day, why gold, where this goes. Those answers became the about text, the tone of voice and the first week of posts. Precise, unhurried, plain English, and honest about risk before anyone asks. "Calling these products riskless would be a lie" is a sentence you can build a brand on, and no agency would have written it.

Five content pillars gave every future post a home: issuance, underwriting in public, reporting, doctrine, and the desk in the room. Each has its own formats. Doctrine, for instance, rides on proof and never posts alone, because the audit told us thesis posts die on small accounts.

What I'd keep. A one-page sheet: the line, the parent, the audience order, the products with exact partner wording, the pillars, the naming rules, the retired lines. Plus a claims gate: no yields, no dates, no capacity, no partner marks until each partner has cleared them. Everyone who writes for the brand works from that sheet, including me.

03Build a kit, not a pile of assets

The designer delivered a Figma kit: marks in four treatments, two grounds, one accent used once per composition, a display face, a body face, a mono for labels, and an illustration grammar. We wrote it up as a design system with hard rules, so that anyone on the team could produce a card without reopening the art direction. That document did more for consistency than any review call.

Then we built templates by post type, not assets by occasion: product card, partner card, statement, notice, report, each as a static and as a short motion piece, in light and dark. A library of 44 short films in seven groups existed before launch, one for every post type the calendar would need, on a review page where the team marked each film approve, revise or reject. Avatars and banners came out of the same kit. The website went live two weeks early as a holding page with the name and the line and nothing else, so launch day only had to switch content, not ship a site.

What I'd keep. List the post types the first month needs. Build one template per type, static plus motion, both themes, with placeholders in brackets so nothing uncleared can ship by accident. Put them on a page the client can review film by film. One accent, one job per typeface. The rules are what make five people's output look like one brand.

04The partners are lined up before the day, not on it

Every partner got a briefing pack under embargo: what Hare is, what the announcement says and does not say, their role in one sentence, publication guidance (coming soon, no yields or dates, link to the main post), and a draft post written for them to quote or adapt. Each partner had its own group chat with the people who would actually press publish. One shared calendar carried one row per post with an owner, a date, the dependencies and the approval state, so "who is waiting on what" always had an answer.

The press went through a PR firm, and I'd do that again every time. Their job is not the marketing team's job: carry the release to journalists under embargo, answer the follow-up questions with a quote from the CEO, and secure the first story with one outlet that will publish at the agreed minute. Give them the release early and let them work.

Two more things were set in advance. The stage: a panel slot at DAS Asia in Singapore, where the announcement would be made in person to the allocators and institutions in the room, with the reveal held until the CEO's slide. And a teaser the day before from the parent company's account, which said something was coming and named nothing.

What I'd keep. Briefing pack, one chat per partner, one calendar with owners, a PR firm with the release a week out, a stage, and a claims list that has to be clear before anything goes out. The embargo is not paperwork. It is what lets a tier-one outlet publish first.

05Launch day: the order matters more than the volume

I designed the day so that each step gave the next one something to point at. We moved the posting time from the morning to 13:00 London so European and US desks were both awake, and we held our own post until the outlet's story was live. Here is how it ran.

10:40 SGT The stage: Connor revealed Hare by GSR on the DAS Asia panel. The audience that matters most heard it first, in person, before anything was online.
12:34 UTC The first story: CoinDesk published from the embargoed release, with Connor's answer on how the capital is used. Earned, not paid. Everything after this links back to it.
12:40 UTC Our own post: The announcement went out from the Hare account with the launch film, tagging every partner. Turtle quoted it two minutes later; GSR posted an hour later.
first hour Partners: GSR's head of communications quoted it to his own audience. Paxos Labs posted on the gold vault. Chainlink and Aave replied from their main accounts. Each had the draft in hand before the day.
first hours Press that follows press: Crypto Briefing wrote its own piece. PANews, ChainCatcher, TokenPost, BloomingBit, BlockTempo and Wu Blockchain carried the story into Asian-language markets, and the aggregators followed. None of it was pitched. It came from the first story.
afternoon Large accounts: The biggest reach of the day came from news pages and analysts picking the story up on their own. A small number of placements with news pages and creators were added after the organic wave, not before it, labelled as paid where the platform asks for it.
all day Team and community: Everyone on both teams posted in their own words. Partners' people, investors and the community built around the parent company over two years carried it into their own networks. The event's media published the stage clip by evening.

The thing I'd underline: the paid layer was the smallest and the last. Paid media on launch day is a multiplier on something that already exists. If the organic wave isn't there, you are paying to amplify silence.

What I'd keep. Stage, then earned press, then your own post, then partners, then press that follows press, then paid, then everyone you know. Hold your own post until the outlet has published. The story makes the post credible, not the other way round.

06The week after is part of the launch

Launch day carried the company announcement and nothing else. The products and the partners each got their own day afterwards: the two vaults with Aave, then GSR, then Turtle, then Paxos Labs, then the data and monitoring partners once their role lines were agreed, then a risk explainer in Connor's words. Each post had its template, its partner's wording cleared, and a quote-post ready from the other side.

Spreading the partners across a week does three things. The account posts daily with something genuinely new. Each partner gets a day where the attention is theirs, which is what makes them show up. And the brand's voice settles in public before the first vault opens, when the questions get harder.

What I'd keep. Plan the eight days after launch before the launch. One partner or product a day, with their post and your reply both drafted. Keep the claims gate running; the pressure to add a number is highest in week one, and that is exactly when a wrong one costs the most.

Hare by GSR launched on 7 October 2026 at DAS Asia, Singapore. Hare USD Earn is powered by Aave. Hare Gold Earn is powered by Aave and in partnership with Paxos Labs. Opening dates and terms are announced separately. I run marketing at Turtle, a partner in Hare. Nothing here is investment advice.

> cd ..