All KPIs

The 30-day scorecard

Rolling window, 8 September to 8 October 2026, against the 30 days before it. List and deliverability tiles are the current month to date, because they are counted by calendar month. Each sparkline is that measure's own trend.

One line per section. Green is good news, red is bad news, grey is context; colour never means up or down on its own. Click a line to open its section.

Contents

16 sections.

Why the 30-day change moved daily, 2 Aug – 8 Oct

The headline compares a rolling 30 days against the 30 before it. When an exceptional week crosses the boundary it leaves the current window and joins the baseline, so it counts twice in opposite directions and the percentage swings further than trade did.

The weekly three

Change is green when the news is good and red when it is bad, so unsubscribes falling is green. Sends and recipients are volume, not a verdict, and stay grey.

Flow, acquisition, campaign. The same three blocks with the same rows every week, measured over the rolling 30 days, 8 September to 8 October 2026, against the 30 days before it. Nothing is added or dropped week to week: a row that appears only when it looks good cannot show a trend, and the trend is the whole reason to read this weekly.

1 · Flow

Automated sends, every live flow in the account, no exclusions.

2 · Acquisition

New opt-ins by the surface that captured them, counted from the signup events themselves.

The prior 30 days include 9 days before the pop-up two-step fix on 2026-08-18, so any SMS change above is mostly that fix landing, not a new trend. Compare two windows both after 2026-08-18 to read the trend.

3 · Campaign

Broadcast sends, GlockApps seed tests and internal test sends excluded.

Totals

The absolute numbers every rate on this page is measured against.

Efficiency each tile states its own window

Absolute totals say how big the programme is. These say what it is worth per unit, which is what moves when something improves or decays.

Attributed against actual by month, Oct 25 – Oct 26

Every revenue figure on this page is Klaviyo-attributed: an order is credited to email or SMS when it follows a click or an open inside Klaviyo's attribution window. That includes orders the store would have taken anyway. Shopify's own store revenue is the denominator, so the share is stated here rather than left to be assumed.

Flow attribution is available from Klaviyo only as a twelve-month total, so it appears in the total row and not by month. Campaign-attributed by month is both channels.

Margin on what the channel earns by month, Jul 26 – Sep 26

Every revenue figure above is turnover. This converts it to what the business keeps, from Shopify order records multiplied by each variant's unit cost. 90,476 orders across 3 complete months; 352 line items carried no unit cost and are excluded from COGS.

What the programme costs, against what it keeps per month; revenue is the 12-month average

What the channel costsmeasured March to August 2026

Billing read from the Klaviyo invoices, March to August 2026. The full cost report, with the SMS split, is at bfs-klaviyo-costs.vercel.app. Cost is against attributed revenue, so the share above inherits the attribution caveat.

Why April spiked, then fell by month, Jan 26 – Oct 26

What is actually shrinking by month, Jan 26 – Oct 26

Two real declines sit underneath the import, and neither is visible in a net-growth chart.

What Shopify says about the email side by month, May 26 – Oct 26

Klaviyo can only report what it was told. Shopify holds the original consent record, so it settles whether the checkout changed.

List growth

The question this panel answers is the one that has never been on a slide: are we net growing or net shrinking, on each channel, month by month. Bars above the line are months the list grew; bars below are months it shrank.

Email 12 mo: Nov 25 – Oct 26 MTD

Email has grown every month for twelve months. The best organic month was January 2026 at +14,243; April's raw +19,941 included the 11,681-address import. October 2026 added 16,731 and lost 12,940 at run rate, against a first-quarter average of 21,729 joined and 8,265 left. Both joins and churn are inside the first-quarter range.

SMS 12 mo: Nov 25 – Oct 26 MTD

SMS since the migrationmeasured to 2026-08-28

Postscript was switched off on 19 June 2026. The list fell through July and August and has grown since. Each part of that has a measured cause.

Why the list shrank: people replied STOP to one campaign sequence by month, May 26 – Oct 26

Every SMS opt-out event carries its own reason, so the question answers itself.

What the broken replacement cost, and the fix 19 Aug – 1 Sep

Postscript's pop-up stopped on 19 June. Its Klaviyo replacement was misconfigured until 18 August, so for ten weeks the site had no working SMS pop-up. The checkout consent field was never affected.

Against the Postscript baseline each row states its window

Checked on independent measures, because any one of them can look fine on its own. The question is not "is SMS sending" but "is it behaving at least as well as it did before the migration".

The consent step is still the leak 21 June to 18 August 2026

Capture is ahead of the old baseline, and the form still loses most of the people who try. It has its own report: https://bfs-sms-gap-recovery-deploy.vercel.app/.

Open rates, decline rates, bounce rates 12 mo: Nov 25 – Oct 26 MTD

Open and click are unique per send, from campaigns, against delivered. Bounce, decline and complaint are account-wide, flows and campaigns together, against emails received in the month. Each panel is its own scale, so read them one at a time rather than against each other.

Opens are drifting down and that is the one to watch. The unique campaign open rate has fallen from 43.6% to 36.9% over twelve months, a loss of 6.8 percentage points, while reach rose. Declines and complaints both improved over the same period (0.359% to 0.295% and 0.037% to 0.029%), and bounces are steady near 0.21%. So the list is not complaining more; it is opening less. That pattern points at who we mail rather than what we send.

SMS, from the Klaviyo migration onward by month, Jun 26 – Oct 26 · per received

SMS only became a Klaviyo channel in June 2026, when the Postscript list was migrated. Before that Klaviyo recorded a few dozen text events a month, so a rate calculated on those months divides by nothing and produces a meaningless number. The panel therefore starts at June 2026, and anything earlier lives in Postscript.

Opt-out rate is unique unsubscribes against texts received.

SMS click rate, and why it is not an engagement measure campaigns since 15 Jul · per delivered

SMS clicks are recorded and reported here, because the number gets asked for. What follows is what the number is made of, so nobody optimises against it.

The mechanism consistent with every observation is carrier and handset link pre-fetching: a prior hour-level count of one send put 95.5% of clicks inside the delivery hour. Use conversion rate and revenue per recipient to judge an SMS send; use click rate only to spot a broken link, where it drops to near zero.

Flow performance

Every flow that sent anything in the last 30 days, one row per flow per channel, ranked by revenue in USD. A flow that runs both email and SMS appears twice on purpose, because a blended row would hide which half is working. Rates are against delivered; bounce and failure are against recipients.

Email flows email 30d: 8 Sep – 8 Oct · per delivered

SMS flows SMS 30d: 8 Sep – 8 Oct · per delivered

Reorder and Subscribe & Save 30d: 8 Sep – 8 Oct vs prior 30d

The reorder estate is 63 of the 90 flow rows and 26% of flow revenue. It moved -12.9% against the prior 30 days while the rest of the estate moved -27.0%. The group total hides the breadth, so this reads it both ways: how the dollars net out, and how many flows moved which way.

Flows that are live but sent nothing 30d: 8 Sep – 8 Oct

5 live flows sent nothing in the last 30 days: 4 carry send steps and short delays (dark), 1 has no send step by design. 3 more were earning last month and sent nothing this month.

Campaign performance

Twelve months of email first, then both channels over the period they have shared Klaviyo. Email is the only channel with a full year of history here; SMS ran on Postscript until June 2026 and its earlier months are not in this account.

Email campaigns, November 2025 to October 2026 by month; open and click per delivered, revenue per recipient

Rev / recipient is campaign revenue divided by recipients for that month. October 2026 is month to date, so its sends, recipients, orders and revenue are partial; its rates and per-recipient figure are not. The twelve-month campaign revenue in Totals starts on 8 October 2025, so it also counts 8 email sends and $339,556 from the part-month October 2025 this table leaves out; the rows here plus that stub sum to the Totals figure.

All three lines start at 100 in November 2025. Volume and revenue share an axis only because both are indexed; no absolute comparison between them is implied.

SMS campaigns, November 2025 to October 2026 by month

Klaviyo holds SMS campaigns from June 2026. November 2025 to May 2026 ran on Postscript and are not in this account.

Both channels, since 15 July 2026 15 Jul – 8 Oct, by month · per delivered; bounce+fail per recipient

Every broadcast send in the campaign pull, which starts on 15 July 2026. July 2026 is therefore a part-month and is marked so; the twelve-month tables above carry the whole month.

Campaign revenue since 15 July 2026 is $1,280,841 across 120 sends, $1,022,057 of it from email and $258,784 from SMS.

Every one of the 120 campaigns since 15 July 2026, most recent first, is on its own page: campaign reference.

Is the buying base growing with the list? by month, per day, Jan 26 – Oct 26

The list and the buyers are two different populations. This panel puts them on the same scale so the difference is visible. Everything is per day, which removes the distortion from month length and from October 2026 being a partial month.

Why the two lines separate by month, per day, Apr 26 – Oct 26

Splitting orders into first-time and repeat, from Shopify's own customer records, shows where the growth is actually coming from.

Cohort retention and repurchase timing 2025-09-01 to 2026-10-08

Repeat share says what fraction of orders came from returning customers. It cannot say whether retention is improving. This does: of the customers first acquired in a month, how many came back, and how fast. 190,672 customers across 360,122 orders, 2025-09-01 to 2026-10-08.

The first 3 cohorts of the scan (Sep 25, Oct 25, Nov 25) are excluded: a customer whose first order predates the window reads as new, which inflates their repeat rate. Blank cells are cohorts too recent to have completed that window.

How long buyers actually wait before reordering 2025-09-01 to 2026-10-08

Reorder flow delays should be set against observed repurchase timing, not a guess. Median is the gap between consecutive purchases of the same product by the same customer.

Products with at least 40 observed repurchases. A flow whose first delay sits far from the median for its product is either chasing people who have not run out yet, or arriving after they have already reordered.

Inbox placement by month, Oct 26 – Oct 26

Open rate counts the people who opened. It cannot count the people who never saw the message. Seed testing is the only measure of that: 660 seeded tests, 51,675 seed inboxes.

GlockApps retains roughly the last 700 tests, which at this send volume is about two months. A longer trend needs the history exported before it rolls off.

Identity resolution 2026-02-25 – 2026-10-08

Retention.com matches anonymous site visitors to an email address and writes them into Klaviyo. It affects every per-profile rate on this page, and until now the pack could not see it. Read from Retention.com's own API: 20,552 resolutions between 2026-02-25 and 2026-10-08.

Referrers: (direct) 10,481, www.google.com 3,815, blackforestsupplements.com 3,734, m.facebook.com 635. Resolutions by month: Mar 26 2,865, Apr 26 2,930, May 26 2,067, Jun 26 2,422, Jul 26 2,307, Aug 26 3,363.

Suppression: who we hold back, and why 63 sends, 7 Jul – 30 Sep; sizes as of 8 Oct

Every campaign records the audiences it included and the audiences it excluded. Across the 63 email sends between 7 July and 30 September the exclusion list is close to identical from send to send, so this is a standing policy rather than a per-send decision.

The largest send, reconciled send of 30 Sep; segment sizes as of 8 Oct

"#2 Protein Coffee - Secret Revealed Blog" is the biggest send in the window. Its audience arithmetic is the whole reach gap.

Re-engaging the suppressed pool, and why it did not stickmeasured to 2026-08-21

A re-engagement programme ran out of the backup Klaviyo account from 16 April to 1 June 2026, 18 sends to roughly 73,076 people drawn from the same unengaged pool the stack above holds back. It worked. The handoff did not.

What to do about it

The capture game plan, worked through against the 20,000 a month SMS target — what every surface produces, why the pop-up cannot close the gap alone, and the ranked plan: bfs-capture-plan.vercel.app.

The actions that come out of this pack live on their own page so they can be worked through and ticked off without scrolling this one: bfs-kpi-actions.vercel.app. Four to do this week, three over the next month, and three that need a decision before any work starts.