Every search it takes to reach your market
Type a job title and a place. One search finds about fifty people, so this hands you the whole set that finds the rest.
Free, nothing to connect, and nothing is searched until you click through.
We measured roughly 900 live searches over two days. The number that surprised us was not how many a good query returns. It was how little the page number matters.
people from twelve pages of one query
person added by pages five to eleven, between them
people from twelve titles across twenty-five cities
overlap between a city slice and a first-name slice
So depth is not a cursor. The only thing that gets past the ceiling is cutting the population into pieces small enough that each one fits inside a single result set, and every independent way of cutting it multiplies the reach again. That is what the tool above builds.
A job title is enough. A place narrows it to that market and changes which searches make sense, because a country decides both the site to search and the names and cities worth cutting on.
Every line is a real search. The label on the right says which way that line cuts the population: by city, by first name, by the words people put in a headline that are not a job title.
Copy one, or open it in Google. Then do the next one, keep track of who you have already seen, and open the profiles worth reading.
Each of these was measured on the share of results that turned out to be the person described, and on how many of them the other cuts never returned.
A market carries people who write their job one way and people who write it another. They are different strings, so they are different people.
Measured at 100% on target and 100% of them new. A market with expatriates and multinationals holds both, and one search only ever finds one of them.
The most legible cut, and it stays linear far longer than expected: 300 searches returned 2,552 distinct people with no sign of flattening.
A partition of the same population that barely repeats the city one. Six titles across ten cities returned 532 people, the same titles across ten first names returned 578, and 38 appeared in both.
Asking the index from inside a city is a different question from demanding the city appear on the profile. Against a 10% noise floor, it returned 50 to 95% people the unlocated search did not.
What somebody writes in a headline beside their role. Generic, cheap, and 93 to 100% of what it returns is new.
Nothing is searched to build this plan, so it costs nothing and needs no account. The searches open in Google under your own browser, on your own terms. Read the privacy policy.
Twenty-four searches is twenty-four tabs, every result read, every duplicate spotted, and a note of who you already wrote to. Describe who you want to reach and your agent runs the whole plan, keeps a real job title on every person, and remembers where it stopped so next week returns people you have not seen.
Searching a public profile site through a general search engine rather than through its own search box, by restricting results to that site. The classic form is a site: restriction plus a quoted job title. It reaches profiles that are publicly visible, and it needs no account on the site being searched.
Because there is no depth to page into. Twelve pages of a single query returned 37 people in total, and pages five through eleven added one person between them. Two fetches of the same page differ about as much as two fetches of different pages, so the page number is barely a cursor at all.
No. The date filter is accepted and ignored: whatever window we asked for, the median result was 346 to 379 days old. Anybody selling recency on x-ray search is selling a parameter that does nothing.
No. The plan is built locally and every search opens in your own browser against a public search engine. What you are looking at is the same public result anyone gets by typing the query themselves.
Barely, and past a point it finds fewer. Adding an industry taxonomy group to a title cost 37 points of precision in our measurements, and a longer query is likelier to make the search engine quietly drop the site restriction, which returns a page of results from everywhere else. Cutting the population beats complicating the query.
More than most people expect, and it stays linear. Twelve job titles across twenty-five cities returned 2,552 distinct people at about 8.5 per search, still rising when we stopped.
Every guide to x-ray search tells you to write a better Boolean string and page through the results. We measured roughly 900 live searches over two days and found that the page number barely matters. One query saturates at 30 to 50 people however deep you go. Here is the ceiling, the only thing that gets past it, and the method.
Alexandre Sarfati
Seven ways to find B2B prospects on LinkedIn without paying for Sales Navigator, honest about where free search throttles you, and where a free finder picks up.
Alexandre Sarfati
Almost every guide on this question treats it as a privacy problem: how to peek at a profile without being seen. That is the small version. The useful version is a pipeline question, how far down the funnel you can get with nothing connected, and why doing the research that way protects the account you eventually send from.
Alexandre Sarfati
Click a button and let your favorite AI weigh in.
