For eighteen years at Roche I was on the inside of enterprise internal communications —
not writing about it, running it. I made the business case for a campaign platform,
introduced it, and scaled it until it was carrying more than twenty million
messages a year for over six hundred teams, at under a fifth of what the
incumbent tools cost. I trained and coached more than a hundred markets on it. And
because it was pharma, GDPR and pharmacovigilance obligations were architected in from
day one rather than retrofitted after an audit.
So I know exactly what this job is. The all-hands email. The reorganisation
announcement. The benefits change that half the company will misread unless the second
paragraph is exactly right. Not campaigns — correspondence, to people
who already work here and cannot unsubscribe from their employer.
And I know what the tools get wrong, because I spent nearly two decades working around
it. They are marketing tools. Every assumption inside them was built for reaching
strangers who might buy something: upload the contact list, score the engagement,
segment the audience, find out who didn't open it. Aimed at customers, those are
ordinary features. Aimed at your own colleagues, the last one stops being a metric.
“Can you send me the list of who hasn't opened it yet?” is a reasonable question in
marketing and a disciplinary one in a company.
I have been asked that question. I understand why people ask it — they are usually
anxious, not malicious, and they want to know the message landed. But the honest answer
is that a tool which can produce that list will eventually be made to produce it,
by someone more senior than whoever said no the first time. The protection has to be
in the architecture, not in the policy.
So I did what I have always done — in 1997 it was a CorelDraw script engraving circular
text onto copper trays, and the instinct never changed: find the step no human should
repeat by hand, then build the thing that removes it. I built my own.
It sent from the sender's real mailbox, using the company's own accounts, and it worked
well enough that people asked me for it. Then I left, and it stayed behind — wired to
an account that was no longer mine. I could not take it with me, and the people using
it could not keep it running.
SlipperySign Mail is that tool, rebuilt properly. Same core idea — the mail leaves
your mailbox, not ours — but this time built as a real product, on infrastructure
I own, with the privacy guarantees written into the database schema instead of into a
promise on a website.
Most product people learned systems thinking from a course. I built the systems first,
and did the job titles afterwards — sysadmin, webmaster, developer, full-stack lead,
project manager, business analyst, product owner. Eight titles, one underlying
capability: see the system inside a scattered mess of pieces, propose the version worth
building, then build it so the next person can extend it without friction.