Specimen report. Halden, Verda Supply, Nordkraft and Bloom & Boyd are invented brands and the observations are representative rather than measured. A real teardown carries the actual trigger timestamps, the raw captured emails, and the identity that received each one. We do not ship a figure we did not observe.
Part four
What to test, how we got this, and what it cannot tell you.
These are tests, not forecasts. We measured what four brands send; we did not measure what it earned them, and nobody who only reads inboxes can honestly claim otherwise.
Six tests, in the order we would run them
| # | Test | Why this, from the data | Read it on |
|---|---|---|---|
| 1 | Hold the first cart email at zero offer. Move Halden's discount from +4h to +20h, matching the category floor. | All three competitors hold zero for 19–96h. Halden concedes at 4h. | Cart-flow revenue per recipient, and discount-code redemption rate on email 1 vs 2. |
| 2 | Extend the cart sequence from 2 to 4. Add day-3 and day-7 messages. | Competitors send 3–5 and are still sending on days Halden is silent. | Incremental orders attributed to emails 3 and 4, net of any unsubscribe lift. |
| 3 | Build a browse abandonment flow. Currently none exists. | Two of three competitors run one. Halden has the trigger and no flow behind it. | New-flow revenue, and whether cart-flow volume falls (cannibalisation vs addition). |
| 4 | Test the Nordkraft shape as a variant. Email 2 carries proof — testing, sourcing, reviews — instead of a discount. | Nordkraft caps at 12% with 3 emails and never discounts before day 4. | Margin per recovered order, not conversion rate alone. This trades volume for margin by design. |
| 5 | Split the checkout code from the cart code. Halden reuses BACK10 across both. |
Verda and Bloom use flow-specific codes. This is attribution hygiene, not competitive. | Nothing — this is a measurement fix that makes tests 1–4 readable. |
| 6 | Re-examine the 15% welcome offer. | Halden's signup offer is the highest in the set; Nordkraft offers none for three emails. | First-order margin and repeat rate of welcome-acquired cohorts, not signup rate. |
Run test 5 first
While the cart and checkout flows share a discount code, Halden cannot cleanly attribute recovered revenue to either. Every other test on this list becomes ambiguous until that is separated. It is also the cheapest change on the page.
How this was collected
| Element | Detail |
|---|---|
| Identities | Four, one per market: DE-02, DE-04, SE-01, UK-03. Each aged over 90 days before first use and kept engaged (opens and clicks on unrelated senders) so as not to be suppressed mid-observation. |
| Trigger protocol | Cart and browse triggers fired on the same weekday, within the same local hour, with comparable basket values (€38–€47 equivalent). Deviations are noted per brand where a store's catalogue made parity impossible. |
| Repeat runs | Three: 12 June, 26 June, 17 July. A single run cannot distinguish a flow from a coincidence. |
| Delay measurement | From the trigger timestamp to receipt in our mailbox — what a customer experiences, not the brand's configured delay. |
| Purchases | One: Verda Supply, 14 June, refunded 22 June. Real payment instrument on a registered company. Nothing fabricated. |
| Capture | Full raw message retained — headers, HTML, rendered image, and the resolved destination of every CTA. Available on request for any observation in this report. |
Coverage log
| Date | Brand | Event |
|---|---|---|
| 12 Jun | all four | First trigger run. All four responded. |
| 14 Jun | Verda Supply | Purchase placed to open the post-purchase sequence. |
| 22 Jun | Verda Supply | Order refunded. Dormancy clock starts for winback observation. |
| 26 Jun | all four | Second trigger run. Ladders unchanged from run one. |
| 3–8 Jul | Nordkraft | Coverage gap. Identity SE-01 stopped receiving. Cause unknown — suppression, sending pause and list hygiene are all consistent with what we saw. Five days missing, not interpolated. |
| 17 Jul | all four | Third trigger run. |
| 19 Jul | Bloom & Boyd | Change detected. Cart email 5 moved from 22% to 25% and its subject line changed. Ladder on page 2 reflects the current value; the prior one is retained. |
What this report cannot tell you
- Nothing about their results. We observe what four brands send. We have no access to their open rates, click rates, conversion or revenue. A longer ladder is not evidence of a better ladder — it is evidence of a longer one. Treat every recommendation on this page as a hypothesis to test on Halden's own numbers.
- One variant per brand, per market. Each brand saw one identity in one geography with one basket. If a brand segments by value, history or predicted LTV, we saw a single branch of it. Higher-value baskets in particular may receive a different ladder entirely.
- We may have landed inside an A/B test. Three consistent runs make this less likely but do not exclude it, especially where a brand allocates by profile rather than by send. Bloom & Boyd's 19 July change is equally consistent with a test concluding as with a deliberate edit.
- Email only. No SMS, push or paid retargeting was observed. Where a competitor's sequence appears to have a gap, an SMS may be sitting in it. This is scopeable separately.
- Five days of Nordkraft are missing, and their cart flow is the shortest in the set. We do not believe the gap hid a message, but we cannot rule it out and have not smoothed over it.
Standing offer
Any figure in this report can be traced back to the message that produced it. Ask for any observation and you will get the raw email, its headers, the receiving identity, and the trigger timestamp — or, where we do not have it, a straight answer that we do not.
This one is a specimen
Halden and the three brands it is measured against do not exist. Everything else on these four pages — the format, the method, the coverage log, the sections we left empty — is exactly what a real teardown carries, with actual trigger timestamps and the raw captured emails in place of representative ones.
Name two or three competitors of one client → We come back within a business day with what is already in coverage, what we can see today, and the lead time if they are new.