I Wrote This to My Team in 1998. The World Built It Later.
Some weeks ago I went back into my old files and pulled out a set of internal notes I had written to my colleagues at 3P — the executive search firm I was running in Mumbai in the late 1990s. These were not manifestos. They were working instructions: to Chetan, to Anjana, to Sajida, to Tasneem, to "All Consultants." Notes about scanning job advertisements, coding resumes, despatching floppies, computing professional fees.
I had not read them in twenty-five years. And reading them in 2026 was a strange experience — because in among the floppy-disk logistics and the courier rates, there are paragraphs that, today, describe products worth billions of dollars.
I am not claiming I built those products. I did not. I was a recruitment man with a few PCs, a LAN, and a head-hunter's instinct. But I saw the shape of the thing — and I wrote it down, with a date on it. So let me set the two columns side by side, honestly, and let you be the judge.
1. The career graph — written 2 July 1998
What I wrote to my team:
I instructed that we must "keep track of each & every executive for his entire working career — no matter how often he moves from one company to another." When an executive's salary appeared again in next year's annual report, we were not to erase the old figure — we were to move it into an Archive file against his name, so that "after 5/6 years, we are able to draw a graph" of his trajectory. And I added the corollary: when an executive resigns and leaves, a vacancy is created, and "knowing who has left which company is a good lead for our marketing consultants to follow-up."
What the world built:
That is a career graph. A longitudinal record of every professional's moves over an entire working life — exactly the asset LinkedIn began assembling in 2003, five years later. The "vacancy as a lead" idea is now an entire category called job-change signal intelligence, sold by firms like ZoomInfo. I was describing the data architecture for talent intelligence in a memo about raddiwala-shop annual reports.
2. The company–industry knowledge base — written 9 September 1999
What I wrote to my team:
I asked Anjana to build a single master list mapping every company in India — I estimated 500,000 Ltd and Pvt Ltd companies — to its industry category. The goal: the moment a consultant clicked a company name inside a candidate's resume, the correct industry would auto-fill. My instruction to her was a principle: "Focus has to be on PREVENTION of a data-entry error rather than CURE." And I told her not to attempt the mega-list in one go, but to enlarge it incrementally as we cleared the backlog.
What the world built:
That is entity resolution and master-data management — and the incremental approach is what is now called data enrichment, the business of Clearbit, Crunchbase and ZoomInfo. "Prevention over cure" is a data-quality slogan people put on conference slides today. I put it in a note to a data-entry colleague in 1999.
3. The matching engine — and the wall I ran into — written 2 July 1998
What I wrote to my team:
I proposed treating each job advertisement as a search query run against our resume database — what I called "PRO-ACTIVE MARKETING." And then I hit a wall and named it precisely: software roles had 116 "super-specialties," and the permutations ran "into BILLIONS," so "it is simply impossible to manually enter the exact combination." Our engineer Cyril's answer at the time was keyword-string search.
What the world built:
The job-ad-as-query is the candidate-matching engine at the heart of every modern hiring platform. But here is the part I want to be honest about, because it is the part I am most proud of: the wall I described was real, and it did not come down with the tools of 1999. The combinatorial-tagging problem I named in that note was only actually solved fifteen-plus years later — by vector embeddings around 2013, and then properly by large language models. I did not have the technology to solve it. But I correctly identified the problem that would take two decades and an entirely new kind of mathematics to crack. Naming the right wall is its own form of foresight.
4. The death of the print ad — written 3 September 1999
What I wrote to my team:
Persuading companies to advertise jobs on our website instead of in newspapers, I built the pitch around one feature — a "View-Counter" that told the advertiser "how many jobseekers clicked/downloaded their advt." My argument was that a print advertisement "will remain on an executive's table for 3/4 days — max one week — till the next issue arrives," and then it is dead. Ours could be measured.
What the world built:
Measurability versus the dead print page is, in one sentence, the value proposition on which Google built an advertising empire. I was making that exact argument to Indian corporates in 1999, to sell them a free job-posting.
5. The video interview — written 8 August 1998
What I wrote to my team:
I argued that if we could install video-conferencing for under ₹2 lakhs, "our American/overseas clients can interview our candidates over VIDEO, live" — and that "no amount of print-advt can get you as much mileage."
What the world built:
Remote video interviewing — HireVue and its peers — which the rest of the world only adopted at scale in 2020, when a pandemic forced it. I costed it out in 1998.
The honest column
Now let me do the thing that most people writing a post like this refuse to do — separate the genuine foresight from the merely good practice. Because if I claim everything, I deserve to be believed about nothing.
A great deal in these notes was simply sound operating discipline, not prophecy. Insisting on structured resumes rather than typed ones; building a lifetime relationship with every client; never being arrogant to the smallest inquiry; archiving rather than deleting. A sharp operator in 1999 could arrive at all of that. I will not dress it up as vision.
The parts I will stake a claim on are narrower and, I think, sturdier: the longitudinal career graph, the vacancy-as-signal, and the explicit naming of the combinatorial-matching wall that only neural methods could later breach. Those three were not standard practice. Those three, I saw early.
I did not have the engineers, the capital, or the silicon to build any of it. What I had was a habit — the habit of writing the idea down, with a date, and handing it to the person who would have to execute it. Twenty-five years later, the dates are the whole point.
— Hemen Parekh
No comments:
Post a Comment