Barton Springs Mill · Fluid IQ Digital

Sending went quiet for a week. The August promotion is still sitting in draft.

Barton Springs Mill sends 11–20 campaigns a month and has done for years. Nothing went out between 5 and 12 August. Separately — and more expensively — almost none of it has ever been tracked.

Prepared 12 August 2026 Klaviyo state as at 19:15 UTC Source: Klaviyo API + Google Analytics 4 (property 322152846) Read-only audit

The headline

There are two separate problems, and they need different responses. One is acute and fixable this week. The other has been quietly costing you visibility for years.

7 days
silent, 5–12 Aug.
Normal cadence is ~3 sends a week
5 days
the August Mill Digest and BOGO
class sale have sat in draft
0.02%
spam-complaint rate in July —
the list is healthy, no warm-up needed
~0
of 505 campaigns carry tracking,
so the revenue is invisible

Problem 1 — a week of silence, and the promotion is still stuck. Nothing went out between 5 and 12 August against a normal cadence of roughly three a week. Sending restarted this afternoon with two recipe emails. But the three campaigns written on 7 August — the August Mill Digest, the August BOGO Class Sale and its follow-up — are still in draft five days later.

That pattern is worth reading carefully: the recipe emails move, the revenue emails don't. It suggests the hold is specific to the promotion rather than a technical fault with the account.

Problem 2 — the campaigns have never been tracked. Klaviyo has sent between 52,000 and 285,000 emails a month for the last year. In Google Analytics, essentially none of it is attributable, because the campaign links carry no tracking parameters. That revenue has been landing in "Direct" and getting credited to nothing. Every report anyone has ever run has understated email.

A note on method

An earlier draft of this document was wrong. Here is what changed and why.

Before we had Klaviyo access, this analysis was built from Google Analytics alone. GA4 could see the automated flows and no campaigns at all, and the conclusion drawn was that no newsletters had been sent in 2026.

That was wrong. Klaviyo shows 505 sent campaigns, 11–20 every month, right through the period. GA4 could not see a single one of them because the campaigns carry no UTM tracking, while the flows do.

It is worth stating plainly because it is the finding: an outside analyst looking at your analytics would conclude Barton Springs Mill does not send marketing email. That is how completely the tracking gap hides the work. Everything below is now measured against Klaviyo's own record.

What we found

Critical · Issue 1

A week of silence, and the August promotion is still unsent.

Campaigns sent per month has been stable for over a year — 11 to 20, every single month. August has three, and seven days passed between the first and the next two.

CampaignWrittenStatus
8.9 sourdoughSent 5 Aug
— seven days, nothing sent —
8.15 pasta fredda12 AugSent 12 Aug
8.16 baguettes12 AugSent 12 Aug
August Mill Digest7 AugStill draft
August BOGO Class Sale7 AugStill draft
August BOGO Class Sale Ending7 AugStill draft

So the account is not abandoned — content is being written and, as of this afternoon, sent again. But the three campaigns from 7 August have now sat for five days. The Mill Digest is the monthly flagship. The BOGO class sale is a revenue campaign with a calendar attached to it, and every day it waits is a day of that promotion gone.

Snapshot note: two of these were still in draft when this audit began and were sent while it was being written. The account is being actively worked in — treat the status column as accurate to 12 August, 19:15 UTC.

Critical · Issue 2

505 campaigns, no tracking on any of them.

The automated flows were built with UTM tracking — GA4 sees them, tagged (misspelled) as klayvio. The campaigns were not. So in 19 months of analytics, the only Klaviyo revenue GA4 can attribute is $4,984, essentially all of it from flows.

That figure is not what email earned. It is what email could be measured earning. The real number is materially larger and currently unknowable, because it is sitting inside Direct.

This is very likely a chunk of the Direct-traffic discrepancy we reconcile every month in your reporting. Fixing it is a settings change in Klaviyo, not a rebuild — and it makes every future report honest.

Good news · Issue 3

Deliverability is healthy. No warm-up needed.

July 2026 (last complete month)CountRateRead
Emails delivered52,944Healthy volume
Bounced1970.37%Well under the 2% danger line
Spam complaints110.021%Excellent — the practical ceiling is 0.10%
Unsubscribes80.015%Very low
Clicked3,5346.7%Solid for a food/recipe list

This matters for what happens next. A list that has been dormant for months needs a slow, staged warm-up before you can send to it again. A seven-day gap on a list with a 0.02% complaint rate does not. You can resume normal sending immediately — the drafts can go out this week.

One caveat: the reported open rate is very high (78%), which is normal and not quite real — Apple Mail auto-opens messages for a large share of subscribers. Judge performance on clicks and revenue, not opens.

Serious · Issue 4

203 lists, and only 7 real segments.

The account has accumulated 203 lists, almost all of which are one-off class sign-up sheets with single-digit membership — 8.9 gnocchi (4 people), 3.1 puff pastry (3), 5.3 late cake sign ups (2). They are historical records, not audiences, and they make the account hard to work in.

The actual addressable audiences are the seven segments, and they are genuinely good:

SegmentProfilesWhat it's for
All Buyers45,364Everyone who has purchased
All Profiles — Newsletter41,785The main send audience
sour dough11,386Product affinity — already built
100 Miles of Mill7,906Local / geographic
pizza geeks4,477Product affinity
Previous class / tour purchasers4,262Class marketing
Placed 5+ orders3,738VIP / loyalty

You are further along than the account looks. The affinity segments most brands never build already exist. What is missing is using them — and the replenishment and lapsed-buyer segments, which are the two biggest gaps.

Serious · Issue 5

Two flows are sitting in draft.

Five automations are live and working: Abandoned Cart, Browse Abandonment, Customer Thank You, Customer Winback, and the Welcome Series. Two are in draft and therefore not sending — Customer Thank You — New vs. Returning and Test flow when someone hits 5 orders.

Both look like abandoned experiments rather than accidents (last touched 2022). Worth a decision either way: finish them or archive them, so the account stops carrying ambiguity about what is running.

The evidence

Campaigns sent per month

Klaviyo send record. August 2026 is 1–12 Aug (partial, shown hatched).

Stable at 11–20 a month for over a year. August has three — one on the 5th, then nothing for seven days, then two this afternoon.
View as table

Emails delivered per month

Klaviyo "Received Email" metric. August 2026 partial; today’s two sends had not yet registered at the time of the snapshot.

Between 52,000 and 285,000 a month — a substantial programme. August so far: 805, a 98% drop against the recent median.
View as table

Addressable segments

The seven real audiences, by profile count.

Enough volume in the main newsletter segment (41,785) to run properly powered A/B tests — see the testing plan below.
View as table

The plan

The first two steps are this week and cost almost nothing. Everything after that is building on a programme that already works better than its reporting suggests.

Step 1

Release the three August drafts, and find out what is holding themThis week

The Mill Digest and the BOGO class sale have been finished since 7 August. Deliverability is healthy and other campaigns went out today, so there is no technical reason these cannot follow. The open question is what specifically is holding the promotion — an approval, a date, or a discount decision.

The BOGO class sale is time-bound. It is the one with a real cost per day of delay.
Step 2

Turn on UTM tracking for campaigns15 minutes

A single Klaviyo setting adds tracking parameters to campaign links. While it is open, correct the flows' misspelled source (klayvioklaviyo) and standardise the medium to email, so campaigns and flows report as one channel instead of four fragments.

Do this before the drafts go out and the August sends become the first ever properly measured. Do it after, and we wait another month for clean data.
Step 3

Resume normal cadence — no warm-up ramp

A seven-day gap on a list complaining at 0.02% does not need staged re-entry. Go straight back to the three-a-week rhythm. (If the pause turns out to have been longer than it looks — say sending was throttled before August — we revisit this and warm up in engagement tiers instead.)

Step 4

Resolve the "people who missed it" question

James asked about queueing people who did not receive earlier emails. Now that we can see the account, this is narrower and easier than it sounded:

  • Nobody missed a campaign — the campaigns were sent, they just weren't tracked. There is no backlog of skipped recipients.
  • The real opportunity is resending to non-openers. Klaviyo does this natively — take a campaign that performed, resend to those who did not open, 48–72 hours later, with a different subject line. Low risk, and it is free incremental revenue on work already done.
  • The two draft flows are the only genuine gap — anyone who hit those triggers since 2022 never got that email. Worth deciding whether that content still matters.
Step 5

Build the two missing segments

The affinity segments already exist. The two that don't are the ones that matter most for a mill:

  • Replenishment timing — flour is a consumable with a predictable repurchase interval. A "you're probably running low" email at the right moment is the single highest-ROI automation available here, and it does not exist today.
  • Lapsed / at-risk — bought before, nothing in twice their usual gap. Catches people before they're gone rather than after. Customer Winback exists but fires on a fixed window, not on each customer's own rhythm.
Both build on data Klaviyo already has. Neither needs new creative to start.
Step 6

Start testing — you have the volume for it

With 41,785 on the main newsletter segment, tests here can actually reach significance, which is unusual and worth exploiting. Details below.

A/B testing — what to test, in order

One variable at a time. With a 41,785-profile segment you can split roughly 20,000 per arm, which is comfortably enough to detect a 1–2 point difference in open rate — most brands cannot test properly and have to guess. You can.

#VariableWhy it sits here
1Subject lineBiggest swing in open rate, cheapest to produce two of. Start here.
2Send day & timeFree to test and often a larger lift than new creative. Your recipe sends cluster oddly — worth probing.
3Offer framingFree shipping vs % off vs bundle. Moves revenue, not just opens.
4Primary CTAConverts the open into a click — copy and placement together.
5Creative & layoutWhere the design refresh gets validated rather than assumed.
6From-nameTest last — changing sender identity often costs recognition.

Three rules that keep results honest

On the design refresh: worth doing, and worth doing as test #5 rather than as an assumption. Ship it as an A/B against the current template on a real send. If it wins, roll it out with evidence; if it doesn't, you've saved the rebuild.

Next steps

Ask · James What is holding the August promotion?

The recipe emails went out today; the Mill Digest and BOGO class sale from 7 August did not. That points at an approval or a pricing decision rather than a technical fault — and it is the one answer that changes what we do first.

Ask · James Can the three August drafts go out?

August Mill Digest, August BOGO Class Sale, and the Ending follow-up. The BOGO sale is time-bound and has already lost five days.

Fluid IQ Turn on campaign UTM tracking

15 minutes, no creative work, and it makes every future report honest. Best done before the remaining August drafts send, so those become the first cleanly measured campaigns.

Fluid IQ Build replenishment + lapsed-buyer segments, then start testing

The affinity segments already exist and aren't being used. Those two are the real gaps, and both build on data Klaviyo already holds.