You’ve been a prospect for 0:000:00.You’re a lead now. It took 0:000:00.You were my prospect. Not any more.
I’m Alexander Andersson, a CRM and lifecycle marketing consultant. This page runs a lifecycle program on you, in the open: profile, rules, messages, an off switch. Click around and watch the panel.
Free, on Google Meet. You leave with what I’d fix first.
CRM work from Tibber, Bauer and IKEA. Every number wears a label saying where it came from.
For fifteen-plus years I’ve run CRM from the inside — the person the numbers land on. Head of CRM at IKEA Ingka Group across 30+ markets. Head of CRM at Bauer Media Group across 20+ brands in six markets. At Tibber — energy-tech, four markets — the referral program and Spark, the AI email tool I built on Braze.
Now I work independently, from Malmö, for a small number of clients at a time across Sweden, Denmark and Norway. I don’t hand over decks. I build the program, wire it into your stack and leave it running. When AI helps, I build the tool myself.
— Alex
Three ways in: a fractional Head of CRM, a lifecycle marketing program or an AI marketing audit.
Senior CRM leadership without the hire. I sit in your team, own the program and ship it.
One defined program — journeys, triggers, measurement — designed, built and live in your stack.
Where AI pays off in your CRM and marketing — and what to build first.
Five stages, five rules, a propensity score, one win-back and an off switch. The same mechanics I’ve built in Braze and the rest of the stack — pointed at a website visitor, with the wiring left visible. It follows you across the site: every page you open is a signal, carried from page to page in this tab’s memory.
anderssonp(book) = 1 / (1 + e−logit)
logit = −3.20 + Σ weight × signal
cap 0.92 · = 1.00 after booking · = 0 when tracking is offHow this score works, weight by weight →| signal | weight | you | log-odds |
|---|---|---|---|
| receipts_openedreceipts you opened · 0–4 | +0.55 | 0 | +0.00 |
| programs_viewedprograms you looked inside: the card, or its own page · 0–3 | +0.85 | 0 | +0.00 |
| pages_readpages beyond the home page, 15 s or more on screen · cap 4 | +0.35 | 0 | +0.00 |
| opened_matchopened the program marked as your match | +0.40 | 0 | +0.00 |
| book_intentclicks toward booking (a book button, a time), email or phone · cap 2 | +0.60 | 0 | +0.00 |
| read_minminutes reading: a card open, or a page on screen · cap 3 | +0.30 | 0 | +0.00 |
| active_minminutes on screen, input in the last 30 s · cap 5 | +0.20 | 0 | +0.00 |
| scroll_depthfurthest point on screen, home page · 0–1 | +0.90 | 0 | +0.00 |
| came_backreturns after 10 s or more away · cap 2 | +0.35 | 0 | +0.00 |
| copied_textcopied text · the length is logged, never the text | +0.30 | 0 | +0.00 |
| idle_minminutes without input, after a 30 s grace · cap 4 | −0.25 | 0 | +0.00 |
| exit_intentheaded for the tab bar, or pressed leave | −0.60 | 0 | +0.00 |
| intercept | −3.20 | ||
| logit | −3.20 |
A logistic model with hand-set weights: the same shape as a real propensity score — signals in, log-odds, a probability out. Nothing from the analytics feeds back into it, so there’s nothing to train it on. The real thing learns its weights on Databricks, XGBoost and MLflow — that’s what I build with.
If you know what leaks, we go straight to it. If the whole program needs work, we go through it, onboarding to win-back, and I tell you where I’d start. — Alex
Or book by email: [email protected]
BOOKED · IT’S IN MY CALENDAR
Nothing in your inbox within a few minutes, or need another time? Write to [email protected].
YOUR RECEIPT · KEPT IN THIS TAB ONLY · CLOSE IT AND IT’S GONE