Cut Email Brief Writing Time 75% with Claude & Klaviyo

Quick Summary

How We Cut Our Email Brief Writing Time by 75% With Claude

This article explains how ECD built an internal tool called the Email Brief Writer on top of Claude — and what that system actually does under the hood. Writing ten email briefs used to take about an hour across six open tabs: promo calendar, need-to-know doc, Klaviyo dashboard, previous brief templates, Asana, and a blank Google Doc to pull it all together. Today, those same ten briefs take fifteen minutes.

The post walks through the full architecture: how Claude reads a client’s promo calendar from Google Sheets, proposes a campaign plan, writes each brief voice-matched to the client, and simultaneously creates a send-schedule row, spawns an Asana parent task with eight role-assigned subtasks, drafts the Klaviyo campaign with audience and send time already populated, and generates the copy doc in Google Drive. The main takeaway is that the time savings are not coming from a smarter AI — they are coming from encoding what the team already knows and giving Claude a precise system to implement it against.

The value is in the prompt layer and the data wiring, not the model itself Each client’s brand voice, segment logic, naming conventions, discount rules, and standing restrictions live in a structured profile. When Claude builds a brief, it does not have to be reminded — it reads the profile and applies every rule automatically, the same way every time.
The system compounds over time, not just saves time upfront Claude pulls monthly Klaviyo performance data — opens, clicks, conversion rates, which subject lines worked, which campaigns underperformed — and those insights flow back into the same system that writes the next round of briefs. Six months in, the system has a sharper picture of each client’s audience than any single person could hold across a full portfolio.
Brief approval triggers four simultaneous downstream actions A send-schedule row, an Asana parent task with eight subtasks, a drafted Klaviyo campaign, and a Google Drive copy doc all generate at once. Nobody opens Klaviyo to set the campaign up manually, because Claude already has.

Writing a brief to create marketing emails used to mean having 6 tabs open at once. The client’s promo calendar in one tab. Their need-to-know doc in another. Their Klaviyo dashboard for segment names and naming conventions, a previous brief to copy the template, Asana to track which downstream tasks needed spawning, and a blank Google Doc for assembling it all.

As founder and CEO of ECD Digital Strategy, a performance-driven ecommerce marketing agency, this was a huge part of what my team did in a given week. To create 10 briefs for clients took my team about an hour.

Today, the same 10 briefs take about 15 minutes, thanks to a tool we built on top of Claude that we’re calling The Email Brief Writer.

Let’s walk through the architecture and how we actually use it.

How Claude builds the campaign plan

The Email Brief Writer is a web app we host on Render that uses Claude as its engine.

Claude picks a client from a dropdown, takes in how many email and SMS briefs are needed for the cycle, reads a few lines about what the batch is about, and hits “Plan.”

Next, Claude reads the client’s promo calendar straight from a Google Sheet and proposes a draggable, editable plan. A human team member tweaks it.

Claude then writes each brief, voice-matched to that client, with the right Klaviyo segments and naming conventions baked in.

When the plan is approved, 4 things happen at once:

1

A row lands in the client’s send-schedule sheet.

2

A parent task spawns in Asana with 8 role-assigned subtasks covering copy, review, design, build, and QA.

3

A draft campaign appears in Klaviyo with the audience and send time already populated.

4

A Google Doc copy file generates from a template in the client’s Google Drive folder.

In other words, nobody opens Klaviyo to set the campaign up, because Claude has already taken care of it.

The same system handles flow briefs, too

The same system handles flow briefs as well: welcome series, abandoned cart, abandoned check-out, browse abandonment, post-purchase, and win-back. Each flow type has a framework inside the Claude prompt.

It sets:

The email count
The delays between messages
The subject line patterns that match the flow’s intent
The rules about whether the brand discounts heavily or avoids discount-led copy

When Claude builds a win-back flow for a particular brand, for example, it doesn’t have to be reminded what the reactivation angle is, or that they never discount more than 15%. Claude knows, because that brand’s profile lives in the system as structured data, and the prompt is wired to read it.

Standardizing brief and template structure

Claude doesn’t design the visual template. A designer still owns that. But Claude standardizes the structure every designed email uses.

Every brief follows a fixed skeleton:

Hero hook
Body argument
Primary CTA
Secondary CTA
Audience block
Send time
Naming convention

The skeleton is filled from the client’s profile and the campaign type. So, when the designer opens a brief, they already know:

Where the hero copy belongs
What the secondary product mention is
Which segments the email targets

While the visual template is still a human craft, the brief that feeds it is now consistent across every person on the team.

Crafting the custom prompts

The whole system runs on long, specific system prompts that live in version control. When we onboard a new client, we paste their need-to-know doc into a “wizard” (a dynamic living document). Claude pulls brand voice, standing rules, and the product catalog into a JSON profile and commits it. The next brief written for that client respects every rule.

1. Prompt 1: user instructions always win

ABSOLUTE PRIORITY ORDER:

1

USER INSTRUCTIONS are the single source of truth for what to plan. If the user says “15% off sitewide today only,” plan a 15% sitewide sale anchored on today.

2

CLIENT PROFILE: voice, segments, recurring monthly briefs.

3

PROMO CALENDAR: informational only. Never substitute “what the brand usually does” for what the user EXPLICITLY ASKED FOR in this batch.

2. Prompt 2: exactly 3 segments, in order

Always include EXACTLY 3 INCLUDE segments, in priority order:

1

PRODUCT

interest (180D)

2

COLLECTION

interest (90D)

3

BRAND

full-engagement (90D): opens + site visits + adds to cart (NOT “Email Engaged”; cast the wider net)

3. Prompt 3: one client’s discount rules, encoded

AVOID sitewide percentage-off as the main hook. 12+ months of 25% off sitewide trained the list to wait for sales. Default promo format: tiered AOV ($90 to $10 off, $120 to $20 off, $150 to free gift).

The time savings, broken down

Overall, we’ve cut our brief writing time by about 75%:

Our 10-brief planning session went from 1 hour down to 15 minutes.
Our 30-brief monthly calendar went from 2 hours down to 30 minutes.

Across the team, that’s hours per person per month that used to go to assembly. Now, those hours go to strategy work and client conversations instead.

The improvements and time back we’re seeing aren’t from a smarter AI. They’re from encoding what we already know:

Voice rules
Segment logic
Naming conventions
What each client refuses to do

That knowledge goes into a system Claude implements against precisely.

And it doesn’t stop at brief assembly. The same backbone is becoming a single data layer for every client we run marketing for. Each ECD client database is wired into its own Shopify account, so SKUs, inventory levels, and recent order data flow into the system in real time. The brief writer knows what’s in stock today, what’s a top seller this month, and what just went live.

On a monthly and sometimes bi-weekly cadence, Claude also pulls a full report from each Klaviyo account via the MCP connection, revealing:

Opens
Clicks
Conversion rate
Which segments engaged
Which subject lines worked
Which campaigns underperformed

And those insights don’t sit in a PDF in a folder. They flow back into the same system that writes the next month’s briefs.

That’s where the value actually compounds. The system isn’t generating the same copy every cycle. It’s reading what worked for that client 3 months ago, what worked last month, what shifted in their catalog, and what segments grew or shrank, and it’s shaping the next round of briefs around all of it. 6 months in, the system has a sharper feel for each client’s audience than any single person could hold in their head across a full portfolio.

The model is becoming a commodity. The prompt layer and the data wiring are the product. The agencies that figure that out now will be the ones that keep their margins as the rest of the industry catches up.

Frequently Asked Questions

What is the Email Brief Writer and what problem does it solve?

The Email Brief Writer is a web app ECD built on top of Claude, hosted on Render, that automates the assembly of email and SMS campaign briefs for ecommerce clients. Before building it, creating ten briefs required about an hour of work across six open tabs — the client’s promo calendar, their need-to-know doc, Klaviyo for segment names and naming conventions, a previous brief to copy the template from, Asana for task tracking, and a blank Google Doc to assemble it all. The same ten briefs now take fifteen minutes. The tool also handles flow briefs for welcome series, abandoned cart, browse abandonment, post-purchase, and winback sequences.

What happens automatically when a campaign plan is approved?

Four things happen simultaneously. A row lands in the client’s send-schedule Google Sheet. A parent task spawns in Asana with eight role-assigned subtasks covering copy, review, design, build, and QA. A draft campaign appears in Klaviyo with the audience and send time already populated. And a Google Doc copy file generates from a template in the client’s Google Drive folder. Nobody opens Klaviyo to set the campaign up manually — Claude has already handled it as part of the brief approval process.

How does the system stay consistent across different clients and team members?

Each client has a structured profile — a JSON file that captures brand voice, standing rules, segment logic, naming conventions, discount restrictions, and product catalog details — generated when the client is onboarded through a wizard that reads their need-to-know doc. Every brief Claude writes for that client reads from this profile automatically, so no one on the team has to manually remember that a client never discounts more than 15%, or that their segment naming convention works a specific way. The brief skeleton itself is also fixed — hero hook, body argument, primary CTA, secondary CTA, audience block, send time, and naming convention — so every designer opens a consistent brief regardless of who wrote it.

How does the system get smarter over time rather than just repeating the same output?

On a monthly or bi-weekly cadence, Claude pulls a full performance report from each Klaviyo account via an MCP connection — opens, clicks, conversion rate, which segments engaged, which subject lines worked, and which campaigns underperformed. Those insights do not sit in a PDF. They flow back into the same system that writes the next month’s briefs. Combined with real-time Shopify data — SKUs, inventory levels, recent order data — the system shapes each new round of briefs around what actually worked recently, what shifted in the catalog, and what segments grew or shrank. Six months in, the system has a sharper feel for each client’s audience than any single person could hold across a full portfolio.

What is the actual source of the time savings, and what does the team do with the hours back?

The savings come from encoding institutional knowledge — voice rules, segment logic, naming conventions, discount restrictions, and what each client refuses to do — into a system Claude implements against precisely. It is not a smarter AI. It is a better-structured prompt layer and data wiring. Ten briefs went from one hour to fifteen minutes. The monthly thirty-brief calendar went from two hours to thirty minutes. Across the team, that is hours per person per month that used to go to assembly work. Those hours now go to strategy and client conversations instead. The broader point is that the model is becoming a commodity — the prompt layer and the data wiring are the product, and the agencies that build that infrastructure now will be the ones that protect their margins as the rest of the industry catches up.

How we cut our email brief writing time by 75% with Claude