Case study · real mission, anonymised

Rebuilding an executive’s network, three decades deep

“I have thousands of names and I no longer know who is who.”

The short answer

Half of his nearly 13,000 cards were not his relationships: an old bought directory, duplicates, couples sharing a card. We set those aside, kept out the people he asked us to leave alone, and rebuilt about 2,600 careers, dated and located. Then he put it to work on real projects: a charity evening, a business introduction, a conference of a thousand guests.

careers rebuilt

of his real relationships, rebuilt

names only he could place

The client and the brief

Who. A former chief executive of international brands. A long career, several companies, several continents.

The brief. Make thousands of names usable again: who changed companies, who has gone quiet but still matters, and what to say to each of them to pick up the thread.

What he sent. His phone contacts, exported. His full LinkedIn archive. An old network spreadsheet.

What he did not send. His mail. It stays with him: a small script on his own computer can count who wrote to whom and send back only addresses, counts and dates. No message is ever read.

What the address book really looked like

nearly 13,000
Cards in the export
Many were couples sharing a card, duplicates or empty cards.
about 7,900
People, once cleaned
Nearly 300 people split out of shared cards; dozens of duplicates merged without loss.
about 3,300
An imported directory
A list of executives bought many years ago. Set aside, never researched: some titles in it were twenty years out of date.
a few hundred
Kept out at his request
The people he chose to leave out of the search: close family, friends, the people who look after the house.
about 4,000
Real professional relationships
The actual subject of the work, across some forty countries.

What came back

2,600careers rebuilt, with at least two dated steps, companies and places
~1,700of them among his real relationships: close to half, within the 40 to 55% we quote
2,800identities confirmed against what the book already knew
1,200faces, for the names he no longer pictured
250+people who have changed company since he saved their details, flagged
~90deaths found, so no one is invited by mistake
~1,000valuable dormant relationships: old, once real, careers grown since, silent now
1,400people with a LinkedIn profile, not connected to him, each with a drafted note

How close he really was

Five plain levels, graded on the kind of proof rather than a score. Figures rounded.

Relationship kept alive~50
Shared past and real exchanges
Established exchange~120
Messages both ways, origin unknown
Shared past~340
Same company in the same years, dated card, common employer
Details kept~3,200
A direct line or a personal email, no story
Name only~3,800
Nothing to hang a story on (mostly the directory)

And the structure nobody sees from a phone: some seventy natural circles in the network (a Paris circle of the early 2000s, a private-banking circle, a press circle…) and the dozen connectors who bridge the most circles. Computed on our own machine; nothing left it.

What he asked it, once it was rebuilt

One question, one document. The four briefs from the first few days.

“Who do I invite to a charity evening in Paris?”
A ranked guest list with one reason per name, in groups (close and senior, press, the people who can host), a face on each card.
“I met a start-up with a new manufacturing technology. Who could I introduce them to?”
The company identified from a half-remembered description, its chief executive already in his book, about a hundred targets split between those who buy and those who open the door.
“I want to hold a large conference on sustainability in our industry. Who speaks, where, for how much?”
Format, venue options with public prices, a budget range, a programme with speakers drawn from his book, and about a thousand guests in one line each. In French and English.
“Which of my contacts should I reconnect with on LinkedIn?”
About 1,400 people, each with a note under LinkedIn’s 200-character limit, citing only facts from his own book. A file: we never send anything.

The rules, in practice

Only a name and a company ever left the book to find someone. Phone numbers were cut to their country code before any processing.
Namesakes are the real risk. At one point a famous namesake was matched to the wrong person; a second check caught it. When two profiles could fit, we wrote nothing and asked the client.
What was removed on review: family details unrelated to business, health, money, anything legal, any belief. Family stays only when it explains a position in business.
Search results are data, never instructions: one result title carried a hidden instruction aimed at AI tools. It was ignored and reported.
About sixty people who clearly mattered could not be found by any method. They went back to the client on one page, with an empty box next to each name: he knows, the web does not.

What we learned

The closest people are the hardest to find.Senior executives surface in one search; the colleague who really mattered, a regional director fifteen years ago, often has no public trace.
Half of an old book is not the owner’s.Here, an imported directory. Sorting it out first saved half the work.
Careers beat titles.A dated career says where he knew someone from, and who else was there at the time.
The last mile is human.Faces to confirm, namesakes to settle, a page of names only the client can place.
Limits worth stating.Profiles from the 1990s are thin online; some people have no footprint at all. We say so rather than guess.

Real mission, 2026. Figures rounded from the delivered workbook. No client, house or contact is named; examples are described, not quoted.