88% ignored it. Then it took four minutes.
Making email automation usable for merchants who think in outcomes, not architecture.
- Role
- Product Designer
- Team
- Team of 6
- Timeline
- 9 Months
- Type
- Full-time
At a glance
8 min read
- Problem
- Merchants knew email could help but had no idea what to build, what order to build it in, or how to measure success.
- Insight
- They didn't need a better automation tool. They needed a starting point. Discovery, not configuration.
- Result
- 4:47 average setup time, 41.2% abandoned-cart email open rate (marketing average 15–25%), 20,000+ active installs on WordPress.org.
20,000+
Active installs on WordPress.org
7%
Avg abandoned carts recovered
$1.6K
Monthly recovery, 500-order store
“Project data showed 88% of merchants weren't using email automation. Not because the tools didn't exist, but because every tool assumed you already knew what to build. The real problem wasn't automation. It was discovery.
Imagine you run an outdoor gear store. You know email makes money for other people. You open the automation tool and it shows you a blank canvas, a node library and a trigger dropdown, and asks what you would like to build.
That was James. Forty minutes in he said, "I give up. I'll just hire someone." Six of the ten merchants I sat with said a version of the same sentence.
The Team That Shaped the Product
A copywriter, a deliverability specialist and a support lead each changed what shipped.
Three collaborators fundamentally changed what we built. The copywriter pushed back on my merchant-first layouts, arguing "Lead with the product, not the store brand." In the project's copy test, open rates rose 12% and click-through rose 34%. The deliverability specialist caught that no merchant would self-configure SPF/DKIM/DMARC, turning authentication into a required activation step (94% success against the under-30% self-serve benchmark used by the team). The lead engineer co-designed a sub-100ms event cache that made the 1-hour abandoned cart window actually viable on WordPress. I drove the design through these collaborations, but this product is better than what I would have designed alone.
What 810 Merchants Told Us
68% did not know what was possible. Merchants think in outcomes; platforms think in architecture.
The survey (800 merchants): I surveyed FunnelKit's merchant base with one open question: "What email automations do you want to set up?"
- 68%: "I don't know what's possible"
- 23%: "Whatever makes the most money"
- 9%: Had specific workflows in mind
| Response | Value |
|---|---|
| Did not know what was possible | 68% |
| Whatever makes the most money | 23% |
| Had specific workflows in mind | 9% |
The contextual sessions (10 merchants): I sat next to merchants while they tried to set up automations using Klaviyo. James, an outdoor gear store owner, 40 minutes in:
"I give up. I'll just hire someone."
That exact exchange happened in 6 of 10 sessions. Almost verbatim.
The diagnosis: merchants think in outcomes ("email people after they abandon"). Platforms think in architecture ("create a list, build a flow, add a delay node, configure trigger conditions"). The translation between those two is where every merchant got lost.
Why Existing Tools Kept Failing
SaaS was too expensive and assumed too much fluency. WordPress plugins were too basic. Neither told merchants what to build.
SaaS platforms (Klaviyo, Mailchimp, Drip): $50–500+/month, multi-hour setup, data living outside WordPress. API keys and webhooks required before a single email sends.
WordPress plugins: Free or cheap, but limited to basic transactional sends. No behavioural triggers. No real automation.
Three specific failures compounded this:
The three failures, in detail
Setup required fluency merchants don't have. Klaviyo's abandoned cart segment: "Users where Cart Abandoned Date/Time is in the last 1 hour AND Email is not blank AND Consent Status equals Subscribed..." One merchant: "I don't know what half these words mean. I just want to email people who left items."
Email templates were generic to the point of uselessness. Standard plugin abandoned cart email: "Hi {{first_name}}, you left something in your cart! [Button]." No abandoned products. No personalisation. Zero brand voice. "This looks like spam."
Nobody told merchants what to build. Platforms are tools. Merchants needed strategy: which automations matter most, in what order, and how to measure them.
The Design That Changed Three Times
A blank workflow canvas, then a recipe gallery, then an outcome-first wizard: 15 of 15 activated, averaging 4 minutes 12 seconds.
Version 1: Visual workflow builder
The obvious starting point: drag-and-drop flow builder with trigger, delay, and email nodes.
Tested with 8 merchants across skill levels.
- Advanced (2): "Great. Very powerful."
- Intermediate (3): "I think I get it but I'm not confident."
- Beginners (3): "I have no idea where to start. Can you show me an example?"
75% of my target users didn't know what to put on it. The copywriter, sitting in on testing, offered the reframe: "What if we show them finished examples first and let them edit?" I was thinking about the tool. She was thinking about the merchant.
Version 2: Recipe library
A gallery of pre-built automations: Abandoned Cart Recovery, Welcome New Customers, Win-Back, Post-Purchase.
9 of 10 merchants activated in under 5 minutes. 7 of 10: "Way easier than I expected."
Remaining problem: 3 merchants asked "How do I change the emails?" They'd launched, but didn't know what to do next. Activated ≠ confident.
Version 3: Outcome-first guided wizard
Step 1. Choose your outcome: Recover abandoned carts / Welcome new customers / Re-engage past customers / Increase repeat purchases.
Step 2. See a recipe with expected results. Numbers sourced from FunnelKit's merchant cohort data, cross-referenced against published studies: ranges, not points. A 30-day check-in shows the merchant's actual recovery rate against the benchmark. Transparency upfront, context when it matters.
Step 3. Customise only essentials: Sender name, discount amount, preview emails.
Step 4. Activate: "Your automation is ready. It will run automatically starting now."
Step 5. Optimise later (optional): "Want to go deeper? Edit timing, add emails, or modify copy."
15 merchants tested the final design. All 15 activated. Average: 4 minutes 12 seconds. 13 of 15: "Easiest automation tool I've used."
Three Decisions That Shaped Everything
A 3-email cart sequence that holds the discount back to hour 72, accessibility across all 47 templates, and authentication as a required step.
1. The 3-email abandoned cart sequence
Why each email lands when it does
Email 1 (1 hour): Exact abandoned products with images. Direct "Complete Your Order" CTA. Why 1 hour? Cart is still top of mind. Why images? CTR was 67% higher than text-only in testing.
Email 2 (24 hours): Customer reviews of the abandoned products. No discount. Why social proof? At hour 24, customers have second-guessed their decision. Other people's confidence reduces purchase anxiety without requiring a price reduction.
Email 3 (72 hours): 10% discount code. Why save the incentive? Discounting immediately trains customers to abandon carts expecting discounts. Email 3 reaches people who genuinely needed a nudge, not everyone. Why 3 emails and not 5? Testing across 500 merchant accounts showed emails 4 and 5 recovered under 2% additional revenue while meaningfully increasing unsubscribes. 3 is where the ROI turns.
2. Accessibility in email templates
The template accessibility standard, and how it was tested
All 47 templates were built to a consistent standard before launch. Alt text on every product image (the actual product name and context, not "image"). Plain-text fallback versions. Minimum 14px body text. 4.5:1 contrast ratio. Single-column layout for screen reader compatibility.
I tested rendering across 60+ clients using Email on Acid, including NVDA on Outlook and VoiceOver on Apple Mail. Three templates failed the initial screen reader test: product names were embedded in images without alt text. Fixed before launch, not after.
3. Deliverability as a required activation step
Authentication isn't a technical config; it's the gate between "sent" and "received." Treating it as optional meant the entire product promise was hollow. I designed an authentication wizard that detects which DNS records exist, explains what's missing in plain language, provides copy-paste instructions for the domain registrar, and verifies configuration before allowing any sends. I tested the explanation copy with 8 non-technical merchants: all 8 successfully authenticated without external help.
94% of merchants now authenticate successfully. Industry average for self-serve: under 30%.
Evidence note: The 94% figure is the observed FunnelKit merchant authentication rate after the wizard shipped. The under-30% figure is the self-serve benchmark used by the project team; it is context, not a like-for-like controlled comparison.
Results in Context
About $19,700 a year for a 500-order store, at 2.7 active automations per merchant.
For an illustrative store with 500 monthly orders: 350 abandoned carts × 7% recovery × $67 average value = $1,640/month, or about $19,700/year from one automation. This is a scenario calculation, not a promised merchant result.
Merchants averaged 2.7 active automations, tracked via FunnelKit Analytics at 30-day intervals. Smaller stores most often added the welcome series second. Larger stores added win-back. That segment difference now informs how V2 sequences recipe recommendations.
What I'd Change
Non-English markets were an afterthought at 31% of installs, and the email editor should have shipped in V1.
Read the post-launch corrections
International merchants were an afterthought. 31% of installations from non-English markets: English-only templates, US timezone defaults, "Complete Your Order" translating awkwardly. Two tiers of experience. Internationalisation from day one.
I should have built the email editor in V1. 47% of merchants wanted to customise layout, colour, or fonts. 12% of 1-star reviews mentioned template inflexibility directly. I'd treated customisation as advanced when it was core.
Individual contact visibility was not edge case. Second most-requested feature post-launch. The use case I hadn't imagined: a customer emails the store asking why they received a message. Without visibility into which automation a specific customer is in, a routine support question becomes a 20-minute investigation. (Shipped post-launch.)
What Surprised Me
Activation is not retention, and one specialist refusing to make authentication optional changed the whole product.
Activation is not retention, and I only learned that after launch. I should have built retention dashboards from day one. Post-launch: 91% of activated automations still running at 90 days, 84% at 180 days. Good numbers, but I didn't have them until 6 months after launch. If they'd been bad, I'd have had no early signal.
The deliverability specialist changed the whole product. I'd have shipped with authentication as an optional setting. He caught that this made the entire product promise hollow. The fact that a specialist without a UX background reframed a fundamental design problem is the most useful reminder I got on this project: the best insight can come from anywhere on the team. Staying open to that is part of the job.
Live product: FunnelKit Automations · Full email template library (47 templates), A/B results, and architecture documentation available on request.
20,000+
Active installs on WordPress.org
7%
Avg abandoned carts recovered
$1.6K
Monthly recovery, 500-order store
41.2%
Abandoned-cart open rate (marketing avg 15–25%)
6.8%
Click rate (ind. 2–4%)
4:47
Average setup time
4.9/5
Rating, 342 reviews
