# Token Math: feedback on the headline, dek, and opening

## The strongest idea

**The person spending the most wants everyone else to spend more.** Normally, being the company's top spender is a reason to cut back. Dan is the top spender by a wide margin (more than 3x the next-highest user), and he reads that gap as a sign the team isn't ambitious enough. That reversal is what makes the piece surprising, and right now the headline and dek don't use it.

The second-strongest idea is the rule that comes with it: **experiment first, then do the token math.** In the body this means Arielle's three questions (what it cost, what it bought, what we learned) and Randy's 4.5-billion-token failure, which still paid off. The dek should carry this idea so the package doesn't read as "spend freely."

## Headline

**Problem:** "Token Math" is a label, not a claim. It names the topic but gives nobody a reason to open the email. The alt, "How Many Tokens are Too Many?", is a generic question the piece never answers with a number. (If you keep it, title case needs "Are.")

**Recommended:**
> **Every's CEO Uses 3x More Tokens Than Anyone. He Wants the Rest of Us to Catch Up.**

It's bold, specific, and fits a subject line. "Catch up" is a bit stronger than the body's "consider ramping up," but it's in bounds for a hed, and the body supplies the nuance.

**Alternatives:**
- **Spend the Tokens. Then Do the Math.** Shorter, and it states the article's rule. Use this one if you'd rather not repeat the 3x figure from the lede.
- **Why Our Biggest AI Spender Wants Us to Spend More**
- **The Case for Burning More Tokens (and Counting Every One)**

## Dek

**Problem:** "What's the ROI on experimentation?" is a second question stacked on a label. It doesn't develop anything. The alt dek is better because it promises practical guidance, but "rein your tokenmaxxing in" works against the piece: the second paragraph says explicitly that the goal *isn't* tokenmaxxing. The alt dek also promises a clear rule for "when it's worth it," and the article offers a process, not a threshold.

**Recommended (pairs with the recommended hed):**
> Expensive AI experiments are welcome at Every, as long as you can say what the run cost, what it bought, and what you learned.

This keeps the key qualification (spending comes with accountability) and previews Arielle's section.

**Alternative (pairs with "Spend the Tokens. Then Do the Math."):**
> Dan Shipper uses more than three times as many tokens as anyone else at Every. Here's how the team decides when big AI runs are worth it, and what a 4.5-billion-token failure taught us.

## Opening paragraph

**What works:** The "But" pivot, the concrete 3x gap, and the closing line ("whether the rest of us are being ambitious enough"). That last line frames the idea as a question rather than a mandate, which is accurate.

**Problems:**
1. **The first sentence is weak.** "Often if you are... you might think about" stacks two hedges ("often," "might") on a hypothetical second-person setup, and it needs a comma after "company." The lede should state the norm plainly so the reversal lands.
2. **The claim strength shifts mid-paragraph.** Sentence 2 says Dan "thinks everyone else should consider ramping up." Sentence 4 says he's "wondering whether" they're ambitious enough. Choose one level of certainty. "Wondering" is the more accurate one, so the paragraph should build toward it.
3. **The 3x fact is buried in sentence 3.** The number is the hook. It should come right after the pivot.
4. **Precision:** the leaderboard is OpenAI tokens. Saying so costs one word and heads off "tokens from which provider?"

**Suggested replacement (preserves "more than three times," "next-highest user," and the tentative framing):**
> Most people who top their company's spending report would take it as a cue to cut back. Every CEO Dan Shipper takes it as a cue to look at everyone else. He uses more than three times as many OpenAI tokens as the next-highest user at the company, and the size of that gap has him wondering whether the rest of us are being ambitious enough with AI.

**Tighter alternative:**
> Being your company's biggest spender is usually a reason to rein yourself in. When it comes to tokens, Every CEO Dan Shipper sees it the other way around. He uses more than three times as many OpenAI tokens as the next-highest user here, which has him wondering whether the rest of us should be ramping up.

Either version works with the second paragraph as written ("The goal isn't tokenmaxxing..."), which supplies the qualification the lede leaves out.

## Small note on the next paragraph

The second paragraph describes tokenmaxxing as "torching tokens for sport." That's fine, but it's the first place the piece says the goal isn't raw spend, so don't cut it for length. It keeps the bold hed honest.

## Assumptions

- The screenshot caption ("Every's OpenAI token leaderboard") means the 3x figure refers to OpenAI tokens specifically. If it covers all providers, drop "OpenAI" from the lede.
- The "us" voice (an insider newsletter from inside Every) is intended, so I kept first-person plural in the hed and lede.