Tom Savano - Crag Marketing case study

Email grew 93% while the business grew 37%

Tom Savano Cocktails sells premium canned cocktails direct, and kept replacing customers because the ones they had were slow to return. Over nineteen months the second order arrived a month sooner, and half of all orders came from people who had bought before.

27% → 52%

Orders from returning customers

Each repeat order was one the brand did not pay to acquire.

74 → 40 days

Wait to a second order

Money from the same customer arrived a month earlier.

+635%

First-quarter email revenue

An alcohol brand's hardest quarter, against the same quarter a year earlier.

Where they started

Tom Savano sells premium canned and bottled cocktails direct to consumers in the UK. The range is built around occasions: crackers at Christmas, ready-to-serve formats through summer, gift sets year round. Customers do not buy on a cycle here. They buy for a reason on a particular date, which gives the brand a narrow window to be in front of them each time.

The account had three years of history behind it and the programme was in reasonable shape. A regular calendar, welcome and checkout sequences delivering properly, sends that earned.

Most sales came from people buying for the first time. Across those three years, three quarters of orders were first purchases. That meant acquisition had to keep running at full tilt for revenue to stay flat, and the money for that comes out of a different budget.

Customers took longer to come back than the gap between occasions. The median wait from a first order to a second was seventy-four days. On a range built around Christmas, summer and gifting, that is long enough for the next purchase to go somewhere else.

A third of the year was doing almost nothing. January and February together carried under four per cent of what the store took, against nearly half for November and December. Part of that is the category, since alcohol is front-loaded into Christmas and January brings a national month off drinking. The rest was a calendar with nothing to say to people who had stopped for a month and would start again in February.

Bounce rate was capping how often they could send. At nearly seven in a thousand it was not going to get the account blocked, but it drags on inbox placement, and placement decides whether an extra send earns anything.

27%

Orders from returning customers

Three quarters of all sales had to be bought in from somewhere else.

74 days

Median wait to a second order

Longer than the gap between the occasions this range sells into.

3.5%

Share of yearly revenue in Jan and Feb

Two months contributing almost nothing, against half the year in November and December.

0.69%

Bounce rate

Sets a ceiling on how often a seasonal brand can afford to send.

What we changed

Deliverability came first, because everything after it depended on being able to send more often without paying for it in placement. Suppression rules, list hygiene, and sending to people who had a reason to hear from the brand.

The segmentation underneath the calendar was rebuilt on purchase data and product history rather than on who happened to be subscribed. A single promotion went out in several versions, split by what somebody had bought and how recently they had engaged. Sends were followed up rather than repeated: people who did not open got the message again, and people who clicked without buying got a different one.

The automated layer was rebuilt across the lifecycle. Welcome and the four abandonment sequences were all delivering inside the first fortnight. Each later got an offer variant and a separate peak-season version, so somebody joining in mid-November met the brand through what it was selling that week rather than through an evergreen introduction. The post-purchase sequence was rebuilt around the second order instead of the first.

A VIP tier went in during that same fortnight, defined on order count and recency, with a calendar of its own: early access before each sale opened, first sight of new formats, separate windows at Black Friday and Boxing Day.

A churn-risk sequence came later, triggering at sixty days without an order. In this category that is roughly one missed occasion, which is a different conversation to the one you have with somebody six months gone.

0.69% → 0.18%

Bounce rate

What made it possible to send more often without losing inbox placement.

+84%

Campaign order rate

Same list and same products, with segments built on what people had bought.

8×

VIP revenue per recipient

Against the programme average, across nineteen months.

+200%

Lapsed customers returning, per month

Customers who had been gone long enough that most brands write them off.

Where it landed

The pattern of sales changed rather than one quarter going well. Half of all orders came from returning customers, more of them reached a fourth purchase, the quiet months started trading, and the month-to-month swing came down by a third.

+93% v +37%

Email growth against the business

Over twelve months, same seasons and same range, with only one of those ours.

33% → 74%

Email-attributed share of store revenue

Klaviyo-attributed revenue went from a third of the business to nearly three quarters.

+72%

Customers reaching a fourth order

Growth at the top of the customer base, where the margin sits.

+376%

Campaign revenue per month

The three months that followed still ran at twice the old rate.

+199%

Campaign orders per month

Held across twelve months rather than one strong quarter.

11 of 11

Months beating the year before

Every comparable month of the first year, by a median of 95%.

−32%

Month-to-month revenue swing

Across nineteen months, which makes a seasonal brand easier to plan around.

We Will Run This On Your Account

Not a benchmark report. Your own history: what your email actually earned against what the rest of the business did over the same period, how long your customers take to come back, and where the programme is underperforming its own past. The findings are yours to keep either way.

Book Your Ascent Call →