Dandelion Chocolate’s list already opened above 70%. The lever left was where the email landed: the primary inbox or the promotions tab. We put Maverick’s inbox optimization on every campaign from May 2026 with a control held back on each send, and read the results the way we’d read any test. Across 32 campaigns and 1.7 million Maverick-optimized emails, opens rose 5.9%, clicks 18.3%, and the Maverick versions out-earned the control on 23 sends out of 32.
Facts
Client: Dandelion Chocolate
Industry: Craft chocolate (CPG, gifting)
List Size: 100K+
Partners: Maverick
ESP: Klaviyo
Focus area: Inbox placement optimization
Results
- +5.9% open rate across 32 campaigns, 70.9% to 75.1%, significant at 95% on 26 of the 32 sends
- +18.3% click rate pooled across the same sends, 2.74% to 3.25%
- 23 of 32 campaigns where the Maverick versions out-earned the control per recipient
- About $153K net incremental campaign revenue on the recipients Maverick handled, May to October 2026
Client Insights
Dandelion Chocolate is a San Francisco bean-to-bar maker with a gifting business that peaks hard in December and a list that behaves like a community: campaign open rates above 70% are normal, and the Advent calendar sends in July convert like a product drop. We’ve run their lifecycle email for over two years.
On a list this engaged, copy and design tests move the numbers less than they’d elsewhere. What still moves them is placement: whether Gmail files the email under Primary or Promotions, and whether Outlook and Yahoo let it through clean.
The Challenge
Inbox placement isn’t something you can A/B test by hand. The signals providers use to sort mail (HTML structure, sender configuration, the technical shape of the message) sit underneath the design, and a brand team can’t rebuild the plumbing of an email three ways and ship all three at once.
The revenue at stake isn’t small, though. Dandelion’s campaigns do most of their year in a few seasonal sends, and on those sends a few points of open rate are worth more than a month of newsletters. So the question was whether an optimization layer could add placement without touching the creative, and whether it would still be adding it five months later, not just in the first demo.
Our Process
1. What Maverick does, and what it doesn’t touch
Maverick is an inbox placement optimization platform. In the hours between scheduling and sending, it tests a range of backend configurations for that specific campaign and implements the best ones as A/B/C variations inside Klaviyo. Design, copy, images and calls to action stay exactly as we built them. The variations differ in the parts of the email that subscribers don’t see.
2. A control on every send
From May 21, 2026, every Dandelion campaign went out with a control variant, usually about a fifth of the audience, and the rest split across Maverick’s variations. Over five months that came to 441,000 control recipients against 1.74 million Maverick-optimized ones, on 32 campaigns, from a Monday newsletter to the Advent calendar pre-order. When a content test ran on top (a story-led version against a product-led one), Maverick ran inside both arms, so the two tests didn’t contaminate each other.
3. Reading it like a test, not a dashboard
Every comparison here is inside its own campaign: Maverick’s pooled variations against that campaign’s control, with a two-proportion z-test at 95% confidence on opens and on conversion. Incremental revenue is the difference in revenue per recipient between the Maverick versions and the control, multiplied by the recipients the Maverick versions actually received, summed over all 32 campaigns, so the figure is net. We report the pooled Maverick result, not the best variant’s, because the best of five always looks good in hindsight.
The Results
Read from Klaviyo on October 8, 2026, recipient-weighted across the 32 campaigns:
| Metric | Control | Maverick versions | Lift |
|---|---|---|---|
| Open rate | 70.9% | 75.1% | +5.9% |
| Click rate | 2.74% | 3.25% | +18.3% |
| Campaigns where Maverick out-earned the control | 23 of 32 | ||
| Campaigns with a significant open-rate win | 26 of 32 | ||
| Net incremental revenue, May to October | about $153,000 |
The first month set the pattern. In the May to June read, four campaign-level tests averaged +11.6% opens and +22.5% clicks, two of them significant on conversion, with $7,181 of confirmed incremental revenue between them. Over five months and 32 sends the open-rate lift held at around 6%: positive on 30 campaigns and significant on 26. That’s about as consistent as an email result gets. Clicks followed, up 18% pooled.
Revenue followed the clicks. The Maverick versions out-earned the control on 23 of the 32 sends, and the net incremental revenue on the recipients they handled came to about $153,000. Where it came from: the Advent calendar sends in July, where a 29% lift in conversion on the largest audience of the year added more in one week than the entire first month; the August product launch, where the Maverick versions nearly doubled the control’s revenue per recipient; and the September subject-line send, where the best Maverick version made twice the control per recipient.
Where To From Here?
The control stays on every send, so every result keeps its receipt. The next read is the one that matters most, the December gifting run, where the audiences are largest and a few points of open rate carry the year. It’s the send this whole setup was built for.
