Case study
Two years of sending, one number that never moved, and a measurement window that guaranteed it would not.
The founder wanted proof that email was paying for itself. His own dashboard was set up to say it wasn't.
The founder owns a US online fragrance retailer: designer and niche perfume at discount prices, plus bath and body, sold to price-aware US shoppers and gift buyers at peak. He came back to us for a second engagement in November 2025, a two-month BFCM and Christmas sprint that then ran month to month. The account had a large list, a full campaign calendar and an owner who reads his own reports carefully. The channel still looked like a rounding error.
- The problem
- Email's best month in two years was 12% of store revenue, and a two-day, click-only attribution window guaranteed the number would never move.
- The plan
- Fix what the inbox sees, then what the customer sees, then what the founder sees.
- The result
- Deliverability 65 to 80, open rate 45% to 60%, and a side by side showing how much the attribution setting was hiding.
Deliverability score and monthly open rate
Both on a 0 to 100 scale. Shaded band is the platform's Good deliverability range (75 to 89). Open rate is the monthly average; the final point is the seven-day figure in early February.
Deliverability score and campaign open rate, November 2025 to February 2026
-
Step 1 · November 2025
Audit
Five findings from the account, each one measurable, before the first campaign went out on 14 November.
- Deliverability 65, Fair band
- No incentives in high intent flows
- Transactional mixed with promotional
- Thin signup capture
- Two day, click only attribution
-
Step 2 · November 2025 to February 2026
Strategy
Three phases, given to him in this order, each one a precondition for the next.
- Repair the signals
- Give the flows a reason to convert
- Agree the yardstick
-
Step 3 · December 2025 to February 2026
Results by February 2026
80Deliverability score, the platform's Good bandFrom 65 in the Fair band at the audit, held at 80 through January58.6%Campaign open rate, January 2026From 45.4% at the audit, above 60% in early February1.08%Campaign click rate, early February 2026From 0.29% to 0.50% on early January sends, against a 1.2% target22%Email share of revenue under standard attribution, trailing 30 days to February 2026From 4% to 6% under the account's two day, click only setting, the same 30 days
A large list, a full calendar, and a channel that looked like a rounding error
This retailer had a large list, a full campaign calendar and an owner who reads his own reports carefully. The channel still looked like a rounding error. Three depths.
The villain is the wrong yardstick. Not a platform, not a person.
A two-day, click-only attribution window is a defensible choice for a brand that wants a conservative floor. It becomes a trap when it is the only number anyone looks at, because it turns every open-then-buy, every remembered offer, every visit three days later into “not email”. The channel can improve on every fundamental and the yardstick will still read flat. That is exactly what happened here.
External problem: email's best month in two years had been 12% of store revenue, and most months sat far below it. Deliverability was in the platform's Fair band. High-intent flows (abandoned cart, winback, VIP) carried no incentive at all. Transactional and promotional content were mixed in the same flows. Signup capture was thin relative to site traffic. And the account's attribution was set to a two-day, click-only window, which counted almost nothing that email actually caused.
Internal problem: he could not tell whether email was underperforming or under-measured, and every report made the question worse. When we showed a lift, he asked for the year-on-year view instead. When the year-on-year view came, it was distorted by a much heavier discount the previous year. He had paused paid retargeting because it was unprofitable, which put more pressure on email to prove itself, on a dashboard configured to make that impossible.
Philosophical problem: a founder who refuses to buy growth with margin-destroying discounts should not be punished for it by a metric. A retailer that has decided profitability matters more than headline revenue deserves a measurement of email that reflects what email did, not how many people clicked within 48 hours.
What the audit found in November 2025
- Revenue Email's best month in two years had been 12% of store revenue, and most months sat far below it.
- Deliverability Score of 65, in the platform's Fair band.
- Flows High-intent flows (abandoned cart, winback, VIP) carried no incentive at all.
- Content Transactional and promotional content were mixed in the same flows.
- Capture Signup capture was thin relative to site traffic.
- Attribution The account's attribution was set to a two-day, click-only window, which counted almost nothing that email actually caused.
We had worked with him before, briefly, and it had not gone well. He came back anyway.
The first engagement in 2024 lasted a few weeks and ended with him hiring someone else. That matters to this story because it means the second engagement started without illusions on either side. He knew our process. We knew he would check every figure. In November 2025 he came back for a tightly defined two-month sprint covering BFCM and Christmas: flow optimisation, a new pop-up, a campaign calendar, and a deliverability and segmentation roadmap.
We had seen the account's shape in other fragrance and beauty retailers: heavy promotional cadence, a list full of two-year-old sale buyers, flows built once and never touched, and a founder who is right to be sceptical of discount-driven revenue because he has watched it eat his margin. The plan we gave him was built to move the fundamentals first and to argue about measurement second, with the fundamentals as evidence.
Fix what the inbox sees, then what the customer sees, then what the founder sees.
Three phases, given to him in this order, each one a precondition for the next.
Repair the signals
Stop sending to dormant two-year-old sale segments. Tighten entry conditions on every live flow so unengaged profiles fall out. Rebuild the welcome flow and relaunch the pop-up in time for BFCM. Watch bounce, spam and unsubscribe like a hawk during the heaviest sending weeks of the year.
Why first: deliverability is a 30-day rolling judgement by inbox providers. If BFCM volume goes out to a dormant list, the score drops in December and takes January to recover. Everything after this phase is measured on inbox placement, so placement had to be defended before it was tested.
Give the flows a reason to convert
Activate the revised abandoned cart, VIP and winback flows with the incentives they had never had, structured as spend thresholds rather than blanket percentages so they protect margin. A/B test the cart flow with and without a discount so the founder gets his answer from his own customers. Separate transactional from promotional content and fix the compliance issues in the post-purchase and shipping flows.
Why second: campaigns create intent; flows convert it. Running a heavier campaign calendar before the flows could catch the intent would have produced more traffic and the same conversion, which is the pattern that had made him doubt the channel in the first place.
Agree the yardstick
Put the account's two-day click-only figure next to the platform's standard attribution for the same period, explain what each one counts, and agree a benchmark based on the last twelve months so that any performance-linked arrangement rests on a number both sides accept.
Why last: arguing about measurement before the fundamentals had moved would have sounded like an excuse. Arguing about it after deliverability, opens and clicks had all visibly improved was a conversation about what those improvements were worth.
What the founder did
He gave access, approved a sprint, and held the line on discounting.
- Signed a defined scope. A two-month agreement with a fixed deliverable list, signed 8 November 2025, with the first campaign out on 14 November.
- Approved the flows against dates. Welcome flow live in November, VIP and winback approved and live by 19 December, abandoned cart with a discount A/B test.
- Set the discount rules himself. He was clear that discounts should not hurt the business, that blanket automated discounts were not acceptable, and that incentives should go to customers who value them. The spend-threshold structure in the flows is his constraint, not our idea.
- Kept the site honest during a warehouse move. For ten days in mid December products were not visible on the site. Rather than push sales emails at an empty shelf, he agreed to a storytelling send that produced strong opens and, by design, no revenue.
- Held his nerve on volume. When January campaigns produced weak revenue, he accepted a 14-day test plan focused on open rate, click rate and deliverability rather than demanding more sends.
- Did not pretend to agree. When we presented a 61% lift, he said the comparison window was not meaningful and asked for a better one. He was right, and the report was redone.
The stakes
What this avoided, and what it became.
Failure avoided
- A BFCM sent to two years of dormant sale buyers, dragging the domain into the Fair band for the whole holiday quarter. One such send did go out early on and opened at 12.6% with no revenue; it was the last of its kind.
- Abandoned cart, VIP and winback flows continuing to run with no incentive, no purchase exit and mixed transactional content, through the highest-intent weeks of the year.
- Sales emails pushed at a site with no visible products during the warehouse move, burning trust for nothing.
- A performance-fee negotiation built on a number that measured 48-hour clicks and nothing else.
Success it became
- Deliverability from 65 to 80, the platform's Good band, and held there through January.
- Monthly open rate from 45.4% to 58.6%, with the last seven days of January consistently above 60%.
- Click rate climbing from 0.82% to 1.08% inside a fortnight of the test plan, against a 1.2% target.
- The founder acknowledging, in the February sync, that deliverability, opens and clicks had all improved.
- A side-by-side that showed the standard attribution model crediting email with roughly 22% of the trailing 30 days' revenue while the account's own setting showed a fraction of that.
He can now see two numbers instead of one, and he knows what each of them means.
The honest version of this story is that the engagement ended in March 2026 without agreement on how email revenue should be measured, and therefore without agreement on what a performance-linked fee should rest on. What changed for the founder is not that email suddenly became a third of his revenue. It is that the fundamentals underneath the channel are now visibly healthy, and he has seen, on his own account, how much of the channel's contribution his chosen attribution window was hiding.
Under the account's own two-day click-only setting, email sat between roughly 4% and 6% of store revenue across the whole window, which is below even the account's historic best. Under the platform's standard attribution for the same trailing 30 days in February, the figure was roughly 22%, inside the band most established stores land in. Neither number is a lie. They answer different questions, and the founder now knows which question each one answers.
What he can say with confidence today: the domain is in good standing, the list is no longer being mailed into the ground, the flows carry incentives he chose, opens are running at roughly double the category average, and the campaign click rate, still the weak line, was moving in the right direction when the work stopped.
Email share of store revenue, two yardsticks
Two-day, click-only window, across the whole engagement. Below even the account's historic best of 12%.
The platform's standard model for the same trailing 30 days in February 2026. Presented side by side on 18 February.
Gap acknowledged by both sides; a benchmark methodology based on the last twelve months was proposed, and not agreed before the engagement closed.
Evidence
Before, achieved, benchmark
Every line we could source a published benchmark for. Several are below. They stay in.
| Metric | Before | Achieved | Benchmark | Standing |
|---|---|---|---|---|
| Deliverability score | 65 (Fair) | 75 (mid Dec) · 80 (Dec) · 80 (Jan) | 75 to 89 = Good | Good |
| Campaign open rate | 45.4% | 53.6% (Dec) · 58.6% (Jan) · 60%+ (7 days, Feb) | 31% all industries | Above |
| Campaign click rate | 0.29% to 0.50% (early Jan sends) | 0.82% (Jan avg) · 1.08% (early Feb) | 1.69% all industries | Below |
| Email share of store revenue, account setting (2 day, click only) | Historic best 12% | ~4.9% (Nov) · ~3.8% (Dec) · 5.7% (Jan) | 25 to 40% | Below |
| Email share of store revenue, standard attribution | Not measured | ~22% (trailing 30 days, Feb) | 25 to 40% | Just below |
| Spam complaint rate | 0.01% | 0.008% (Dec) · 0.01% (Jan) | Under 0.01% healthy | At the line |
| Bounce rate | 0.18% | 0.31% (Dec) · 0.40% (Jan) | Under 1% healthy | Healthy, drifting up |
| Unsubscribe rate | 0.20% | 0.22% (Dec) · 0.21% (Jan) | Under 0.3% healthy | Healthy |
| Flow share of email revenue | Majority (flows partly inactive) | Majority (Dec) · below half (Jan) | ~41% | Not reported as a figure |
Before figures from the 14 November to 2 December 2025 report and meeting minutes. Achieved figures from the December and January reports, the 2 February update and the 18 February minutes. Benchmarks: the platform's 2026 email benchmarks (all industry, no published fragrance vertical), the platform's deliverability guidance and hub bands, and the 25 to 40% email revenue share range for established DTC stores.
Results by programme
| Initiative | Outcome | Figure |
|---|---|---|
| Segmentation and flow entry conditions | Dormant sale segments removed from sends; unengaged profiles filtered out of flows. Deliverability recovered from Fair to Good within four weeks of BFCM. | 65 to 80 |
| Welcome flow and BFCM pop-up | Welcome flow relaunched with BFCM content in November; seasonal pop-up live before Black Friday and updated for Christmas. Sign-ups ran below the previous year because the incentive was deliberately lighter. | n/a |
| Abandoned cart, VIP, winback | Revised with spend-threshold incentives set by the founder; VIP and winback live 19 December; cart flow A/B tested with and without discount. | n/a |
| BFCM and Christmas campaigns | Cyber Monday and Winter Favorites were the top sends. Two discount campaigns only over Christmas because of the warehouse move. | 62.1% open (best) |
| Storytelling send during warehouse move | Intentionally non-commercial while products were offline; strong opens, no revenue, by design. | 61.0% open · 0 revenue |
| Platform and compliance audit | Post-purchase, shipping and delivery flows found mixing transactional and promotional content; fixes specified. Loyalty and reviews platforms audited alongside. | n/a |
| 14-day test plan (late Jan to Feb) | Targets of 60% open, 1.2% click, 80 deliverability. Opens hit target; click reached 1.08%; deliverability held at 80. | 2 of 3 met |
| Attribution comparison | Account setting versus standard model for the same 30 days, presented 18 February. Gap acknowledged by both sides; benchmark methodology proposed; not agreed before the engagement closed. | About 5% vs about 22% |
In the client's words
No verbatim client text is on file. These are paraphrased from meeting minutes and are used here without a name, at the founder's discretion.
Revenue alone is not enough to judge this. Revenue was well down on last year and the business was more profitable, because we discounted less.
The founder, Paraphrased from the monthly review call, 5 December 2025
Discounting should not hurt the business. Automated discounts to every customer are not the answer. Discounts should reach the customers who value them.
The founder, Paraphrased from the monthly review call, 5 December 2025
Deliverability, open rates and click-through have all improved. Repeat orders week over week have not, and with retargeting paused, email has to carry retention.
The founder, Paraphrased from the weekly sync, 11 February 2026
Your turn
If your dashboard says email isn't working, check the yardstick before you check the emails.
-
Repair the signals.
We audit deliverability, segmentation and every live flow, and tell you what the inbox providers currently think of you.
-
Give the flows a reason to convert.
We rebuild them around incentives you set, tested against your own customers.
-
Agree the yardstick.
We show you what your attribution setting counts and what it leaves out, before anyone argues about the result.
This engagement did not end in a handshake, and we have written it up anyway, because the fundamentals moved and the measurement lesson is the most useful thing on this page. If you recognise the pattern, start with the audit.
Start with the audit, before anyone argues about the result
We audit deliverability, segmentation and every live flow, show you what your attribution setting counts and what it leaves out, and give you a plan in the order that worked here: repair the signals, give the flows a reason to convert, agree the yardstick.