April 16, 2019 · 7:42 PM · message one
The first thing anyone said in this group chat, you said.
Two minutes earlier you created the group and added twelve guys. You did not know yet that this thread would run to 9,162 messages, or that the word you would use three days later would end up on a building in Lindon, Utah.
Everything on this site came out of the archive itself. Every quote is real, every timestamp is real, every heart was counted. Nothing was written for effect. It was written by you, at the time, mostly late at night, usually about logistics.
Pulled from the source export, not from memory. 9,162 messages, April 16 to December 14, 2019. Every number on this site is computed from the actual message data. Nothing here is estimated except two figures that say so on the page.
April 19, 2019 · 10:13 AM · day three
The word appears.
The word grit occurs eighteen times across the entire archive. This is the first one. It is three days into the summer, and it is not a brand. It is a complaint about losing to Pittsburgh.
Then you can watch it get away from you. Nobody voted on it. Nobody made a logo. It started showing up in other men’s mouths, and then it stopped being a word and started being the name of the thing.
- Apr 19Ben “We don’t lose in the grit.”A standard, stated once.
- May 10Josh Nilsen “I saved this as a story on the grit Instagram”Three weeks later it has an account.
- May 13Josh Nilsen “Team Grit, WE KNOCK IN THE RAIN”Now a proper noun and a creed.
- May 28Josh Nilsen “Today is what grit is all about and you all embody that”It has become a thing you can embody.
- Jun 18Cody Olive “The mans got grit!”It has become a compliment.
- Jul 14Jordan Massey “Amazon has a suggestion for the Grit”Capital G. It has become an entity.
- Jul 24Dylan Hafen “BIG GRIT!!!!!!”It has become a shout.
- Aug 13Jack “the grit legacy will last forever”Four months in, they call it a legacy.
- Aug 23Robbie Hutchins “Grit baby”It no longer needs explaining to anyone.
You did not name a company on April 19, 2019. You set a standard and refused to explain it twice. The company is what happened when the standard outlived the summer.
2020 to today · the echo
The standard outlived the summer.
Eight months after this chat went quiet you co-founded Grit Marketing with John Taylor and Garth Massey. John Taylor is quoted inside this very archive, on June 8, 2019, giving advice about confidence and mindset. Jordan Massey was in the group. The company was already in the room.
Thirty-seven Golden Doors out of one organization in a single season. And in 2021 Grit partnered with Aptive, the company whose 78% you were beating in a group chat in June 2019.
Robbie’s estimates
Those last two are Robbie’s back-of-the-napkin figures, not public numbers. He is guessing, and he knows he is guessing. You are one of about three people alive who knows what the real ones are. The point is not the accuracy. It is that a guy who was in your car groups in 2019 now has to estimate your company in hundreds of millions, and he still lowballs it out of habit.
Ben,
I still had the Philly group chat, so I had Claude read the whole thing. All 9,162 messages, all 1,086 of yours, every timestamp and every heart. I went in looking for the big speeches. Found those too.
But what actually got me was the stuff neither of us could have counted by hand. The 3 AM car groups. That you answered forty-eight guys in a median of under two minutes, for four straight months. That on the biggest sales day any pest control team has ever had, the thing you said to us was to drink water in Delaware.
The archive also contains, for the permanent record, you getting stung mid pitch, announcing you were allergic, and declining to go to the ER. You informing thirty grown men you would be late to a meeting on account of “bowel issues.” And Zac describing you, with love, as “a crazy ass manager who throws jugs at the wall.” History is undefeated.
I was 22 and I figured that was just how it worked. Everybody had a Ben. Then I went out into the world and found out almost nobody does, and that the reason twenty of us ended up running things is we all quietly copied a guy who was making it up at midnight in an apartment in King of Prussia and never once let us see him sweat.
You said nine words on day three about not losing in the grit. You were talking about Pittsburgh. Seven years later it is on a building in Lindon and there is a butterfly on the newest one. I do not think you meant to name anything. I think you just refused to explain yourself twice, and the rest of us went and repeated it for you until it stuck.
You said we were your guys forever. That went both ways and you should know it.
Thanks for the standard. Love you man!
Robbie
The archive · what you were actually doing
Seven things you did every week without calling them anything.
None of this was written down in 2019. It is visible now only because the chat kept the receipts.
Close the loop.
You invented a read receipt inside a group chat because you refused to let critical information be optional. It is the single most repeated move in your entire corpus, and men answered it every time.
Coach them one man at a time, by name, on a schedule you wrote at midnight.
You did not run group trainings and call it development. You posted named time slots in blocks of ten and fifteen minutes, over and over, and you made them bring evidence.
Build the system, not the hero.
Your most-liked message of the entire summer is not a speech. It is you announcing that you had gone and joined a group chat with every technician in the region so the sales guys could point at a truck on the street. Then you benchmarked it.
Lead from the front, including the parts that hurt.
You were still on doors yourself in August. You got stung mid-pitch. You are allergic. You kept knocking.
Jul 27, Brad Neuhaus · “How about @Benjamin Egan finishing with 8 above $800”
Name the enemy.
You never once asked the team to beat a number. You asked them to beat a city. Pittsburgh, Detroit, Chicago, Louisville, Princeton, Cherry Hill. Abstractions do not get anybody out of bed at 7:30 in the rain. A rival does.
Protect the name when it costs you something.
Third most-liked message you sent all summer. Not a rally. A correction, delivered to a team that was winning, about a joke that was not hurting anybody yet.
Say it out loud, to men who would never say it first.
Young men do not tell each other this. You did, in writing, in the middle of a sales chat, at twenty past two in the morning.
And the one you never said out loud until it broke.
August 12, the morning before the hardest day of the summer. Nobody makes a leadership principle out of this one. It belongs here anyway.
He apologised. To them. For sleeping.
April 19 to today · what this page is
Two questions you can actually answer with 9,162 messages.
There is a difference between a good summer and a thing that keeps going after everybody drives home. It comes down to two questions, and this archive is big enough to answer both of them with arithmetic instead of opinion.
| Question one. Did the things he said outlive him? Did they end up in other men’s mouths, said without him in the room?Measured by tracking who used his language, and when. |
| Question two. Did the room stop needing him to carry it? Did it get louder as he got proportionally quieter?Measured with Shannon entropy, week by week. |
Both answers turn out to be yes, and together they are the reason there is a company. Nothing here proves cause. All of it is the record of an input and what followed it.
The input
Day three. Nine words inside a longer message about Pittsburgh. It cost him nothing to write and it earned two hearts.
Question one: did it outlive him?
To be clear about what this chart is not. It is not a count of how often Ben said “grit,” and it is not a measure of how much airtime he had. Every step up is a different man using his word, on his own, in his own message, with nobody asking him to. The line only rises when the idea has moved into somebody else.
It sat at one man for twenty-one days. Then Josh. Then Cody, Jordan, Dylan, Jack, Tanner, Robbie. By the middle of August the word did not need Ben in the room to get said.
Why that number means anything
Because most of what he said never travelled at all. If this were secretly measuring how much he talked, then everything he repeated would sit at the top of this table. It does not. Some of his best lines died where he said them.
| Phrase he said first | Other men who picked it up |
|---|---|
| “grit”Apr 19. Became a company. | 7 |
| “of course”May 2. See below. | 6 |
| “that’s right baby”May 13 | 6 |
| “make it happen”May 3 | 4 |
| “leave no doubt”May 20 | 0 |
| “bury them”Jun 29 | 0 |
| “embrace the burn”Aug 7 | 0 |
Four caught and three died. The measurement discriminates, which is the only reason the first chart is worth looking at.
The one nobody would have guessed
His second most contagious phrase of the entire summer is not a rally cry, a callout, or anything he would have thought of as leadership. It is the two words he used to say yes.
And then it spread, and every man who picked it up used it exactly the same way. Not as a slogan. As the answer when somebody needed something.
The word he used to say yes became the word the team used to say yes. Nobody taught that in a meeting. It just got copied, because he did it first and he did it every time.
Question two: did the room stop needing him?
There is one number that answers this, and it is borrowed from information theory. Shannon entropy takes a conversation and asks how evenly it is spread across the people in it. One man doing almost all the talking is low entropy. Forty men all carrying it is high entropy. It is the standard way to measure whether a system is concentrated in one place or distributed across many.
All that says is: take each person’s share of the week’s messages, and add up how much each of them contributes to the spread. Run it on all twenty weeks of the Philly chat and you get the picture below. Amber is the entropy of the whole room. Green is Ben’s personal share of it. Watch them cross.
The important part is why his share fell. It was not because he backed off. His own message volume barely changed all summer. His share dropped because everybody else got louder around him. He went from being a quarter of the room to a tenth of a much bigger, much noisier room that was now carrying itself.
Question three: did it survive sixty strangers?
Here is the test neither of the first two questions can answer. That team did not start whole. Nineteen people had spoken by April 26. By the middle of May there were fifty-nine, and by August seventy-eight. Guys kept landing as their semesters ended, thirty-four of them in the first two weeks of May alone.
So the room did not get louder because anyone got more excited. Break the volume into how many people were talking against how much each one said, and the whole increase is people arriving. Messages per day went up three and a quarter times. Messages per person went from 3.17 a day to 3.10. Which is to say, it did not move at all.
And that is where it gets interesting, because when a group quadruples, the normal thing is for the people who were there first to go quiet. Somebody else will answer. That is not what happened.
| Who | April | Peak summer |
|---|---|---|
| The original nineteenThe guys who were there in week one | 3.17 | 4.10 |
| The sixty who arrived laterMeasured once they were in | not here yet | 2.80 |
The founders did not go quiet. They got louder, from 3.17 each to 4.10 each, while sixty new people landed on top of them. And the newcomers showed up running at almost exactly the same rate the room was already running at.
Sixty men walked into a room that had a temperature, and instead of cooling it down, they matched it. Nobody could have instructed that. There is no message anywhere in the archive telling anyone how often to talk. That is the closest thing to measurable proof of culture in the whole thing. A rate that new arrivals copy without being told is not a policy. It is a culture.
So what does any of this actually mean
The two charts look unrelated. They are the same thing measured twice.
A leader whose ideas need him in the room is a bottleneck. Everything routes through him, nothing survives his absence, and the whole thing ends when he leaves. That is most managers, and it is not a criticism, it is just what happens by default.
Both of these measurements say the opposite happened here. The standard stopped needing its author, and the room grew to three times its size around a man whose own output never moved once.
That last part is worth being precise about, because it is the opposite of what it looks like. He did not step back to make space. Run change point detection on his daily message count across all 139 days and it finds nothing at all. No break, no drift, no regime. A flat line at 7.6 messages a day from April to September, while everybody else went from 28 a day to 68 at the peak. He was not letting go of the room. He was the one thing in it that never changed.
That is the difference between a good summer and a company. It is also the mechanism behind the only number on this site that really matters: twenty of those forty-eight men now run teams of their own. You do not get twenty leaders out of a room with one leader in it. You get them out of a room where the leading got handed around until everybody had held it.
Which is why it was not an accident
Chaos theory does not say the outcome was fated. It says close to the opposite: that the input was small enough it could easily have gone another way, and that the size of a cause tells you nothing about the size of its effect. A sentence on day three should not become a building in Lindon.
So the honest version is better than either luck or destiny. It was not luck, because there is a mechanism and you can watch it run in the two charts above. It was not destiny, because nothing guaranteed it. It was one man doing small things on purpose, 1,086 times, mostly about car groups, often at three in the morning, for 128 days straight.
Seven years later the word is on a building, and the newest company under it has a butterfly for a logo.
Eight methods · one corpus · nothing withheld
Philly, eight ways.
Eight statistical methods, run once each over the same 9,007 messages. Six returned something. Three times the analysis corrected a conclusion that had already been written down. One returned nothing at all, and it is here at the same length as the others.
What eight methods agree on
His influence did not run through intensity, volume or charisma. Every method that measured him found the same thing from a different angle: a fixed point. The room tripled around a man whose own behaviour never changed once.
| He never moved. That is the whole finding. Change point detection finds breaks in the room’s volume and in his share. In his own daily output it finds nothing across 139 days. 7.6 messages a day, flat, while everyone else went from 28 a day to 68 at the peak. Burstiness agrees independently: 0.279 against the team’s 0.562.Usable: consistency is measurable, and it was his actual edge. Two methods, no shared inputs, same answer. |
| His share falling is what success looks like from the outside. Entropy and eigenvector centrality on 724 mentions both show his share collapsing while the room grows. Attention share fell from 19.7% to about 8% while total mentions tripled.Usable: a leader’s share of messages trending down while total volume trends up is a health signal you can compute on any team channel this afternoon. Flat or rising share is a bottleneck. |
| His linguistic fingerprint is other people’s names. Out of 453 words tested against a Dirichlet prior, the most distinctive tokens in his writing are jake, matt, justin, robbie, willz, nate, easton, jordan, connor, jameson, and car. What marks everyone else: haha, just, gonna, anyone.Usable: cheap, coachable, countable. He named 106 distinct people. Nobody designed this test to find that. |
| He was measurably unexciting, and it did not matter. On excitation, how much the room accelerates after you speak, he ranks 19th of 46. Essentially average. The energy came from Zac, Parker and Josh.Usable: being the constant and being the spark are different jobs. You do not have to be the most exciting person in your company. |
| Repetition is not transmission. “Brethren” was his signature word, used 31 times, more than anything else. It reached three people. His three constructed rally lines reached nobody at all. What spread was “grit,” said once in irritation, and “of course,” said reflexively whenever somebody needed something.Usable: you cannot install culture by saying a thing often. What transmits is what you do without thinking about it. |
| The most important work had no measurable signal. His midnight-to-6AM output has zero correlation with the team’s next-day activity. r = −0.06 at one day. Nothing at any lag, raw or differenced.Usable, and the one that should worry you: car groups and route bookings prevent problems, and prevention shows up as an absence. A dashboard tracking “did your night work raise team activity” would have told him to stop. |
Three times the analysis corrected me
I predicted he would top the excitation ranking. He was average. I predicted the room would become more predictable as it matured; it got louder instead, though structure above chance did double. And I read the April 27 change point as a decision to hand over the room, which the control then showed was an artefact of a ratio with a tiny denominator.
All three errors ran the same direction: I kept expecting that team to look tidier and more deliberate than it was. The record kept refusing. Everything below is the working, including the parts that went against me.
Method 01 · Self-controlled excitation Contradicted prediction
Who actually makes the room speed up?
A Hawkes process models events that trigger more events. The quantity of interest is the branching ratio: how many further messages does one message cause? I predicted, in writing, that Ben would rank near the top.
Fitting a full kernel to 46 overlapping speakers is unstable, so I used the self-controlled simplification: for every message, count other people’s messages in the 10 minutes after, divide by that same person’s own rate in the 30 minutes before. Above 1.0 means the room genuinely accelerates after they speak. Using each man as his own control removes the obvious confound, that talking during busy periods would otherwise look like causing them.
| Person | Excitation |
|---|---|
| Parker Anderton130 messages | 2.72 |
| Reed Goodman49 messages | 2.72 |
| Zac Affleck42 messages | 2.32 |
| Jacob Scheer82 messages | 2.22 |
| Calvin Escovedo108 messages | 2.06 |
| Benjamin Egan1057 messages · ranked 19th of 46 · median 1.57 | 1.70 |
He is unremarkable here, and I was wrong to expect otherwise. But it fits everything else: 82% of what he wrote was logistics, instructions and one-line answers. Car group rosters do not make a room erupt. The men who spike the room are the hype men. He was not one, and never tried to be.
Method 02 · Binary segmentation First reading was wrong
Did the team change gradually, or on a specific day?
Change point detection finds dates where a series shifts regime without being told anything about the story. You give it numbers and it hands back dates.
Build a daily series of his share of that day’s messages across 139 days. Try every possible split date, keep the one that most reduces total squared error, and accept it only if the improvement beats a penalty scaled by log n so noise cannot manufacture a split. Then recurse into both halves.
One change point survived: April 27, 2019. The algorithm was given no names, no events, and no knowledge that this was a sales team. What sits on the day before it is his standards message, April 26, 12:18 PM, 13 hearts.
Then I ran the control, and it broke my own conclusion
I first read this as a decision: he set the standards on the 26th and handed the room over on the 27th. That reading is wrong, and the check that kills it is one line of work I should have done before writing anything. Share is a ratio, and a ratio can break because the numerator falls or because the denominator rises. So run the same detector on each part separately.
| Series | Change points | Segment means |
|---|---|---|
| His daily message countthe numerator | none in 139 d | 7.18 → 7.64 |
| Everyone else’s volumethe denominator · two breaks, three segments | May 15, Aug 22 | 27.9 → 67.5 → 27.9 |
| His sharethe ratio | Apr 27 | 35.4% → 11.5% |
April 27 is an artefact. The early denominators are tiny, 6 to 44 messages a day, so one message from him was worth 15% of a day. As the room filled, the ratio had to collapse whether or not he did anything. So the finding survives, but it is not the one I reached for. He did not hand the room over, because he never moved. His output is the only series in this corpus with no detectable change of any kind, while everything around it tripled and then fell away again. The chart shows that tail honestly: the room drops back to its April rate after August 22, when the season ended and men went home. His line does not notice.
Method 03 · Eigenvector centrality Independent confirmation
Does a second method, on different data, agree?
Entropy and change point both come from message counts. This one uses an entirely separate signal, the 724 @mentions, and asks the same question a different way: you are important if important people address you.
Parse every @name into a directed edge from speaker to addressee, 724 edges across 72 people, then iterate until the vector stops moving.
| Person | Centrality |
|---|---|
| Benjamin Eganaddressed 69 · addresses 10 | 0.541 |
| Wills Zenger Nilsenaddressed 36 · addresses 7 | 0.322 |
| Cody Oliveaddressed 46 · addresses 64 | 0.286 |
| Jackaddressed 35 · addresses 17 | 0.255 |
| Josh Nilsenaddressed 37 · addresses 20 | 0.253 |
He is the most central node by a distance, and the asymmetry is stark: addressed 69 times, addresses others 10. He is the destination, not the router. Yet his share of attention falls from 19.7% to around 8% while total mentions triple. Different data, different mathematics, same answer.
Method 04 · Adoption curves Signal, with a caveat
How fast does a phrase move through a team?
Language spreads like contagion, so the instinct is to fit an epidemic model and quote an R number. With eight adopters that would be dishonest. What the data supports is the generation interval: the average gap between one person picking it up and the next.
| Phrase | Adopters | Mean gap |
|---|---|---|
| “grit”18 uses | 8 | 17.9 d |
| “of course”11 uses | 7 | 16.7 d |
| “that's right baby”15 uses | 7 | 15.2 d |
| “make it happen”15 uses | 5 | 25.6 d |
| “brethren”31 uses | 3 | 41.8 d |
| “leave no doubt”3 uses | 1 | none |
| “bury them”3 uses | 1 | none |
| “embrace the burn”1 use | 1 | none |
The spreaders move at a strikingly consistent pace, roughly one new mouth every 15 to 18 days. But the real result is the bottom of that table. His most repeated signature word barely moved. Repetition is not transmission.
Method 05 · Log-odds, Dirichlet prior Best single finding
Which words were actually his?
Raw frequency is useless here, because the most common words are common for everyone. Monroe, Colaresi and Quinn’s estimator shrinks rare words toward a corpus-wide prior and divides by their standard error, so what survives is genuinely distinctive rather than merely frequent.
His 1,086 messages against everyone else’s. Every word appearing 25 or more times, 453 of them. Anything above roughly 2 is distinctive rather than noise.
| Word | z |
|---|---|
| jakehim 139 · them 33 | +8.49 |
| matthim 132 · them 38 | +7.99 |
| groupshim 88 · them 10 | +7.36 |
| justinhim 108 · them 49 | +6.44 |
| robbiehim 87 · them 30 | +6.23 |
| willzhim 63 · them 7 | +6.23 |
| natehim 89 · them 44 | +5.68 |
| willhim 212 · them 238 | +5.67 |
| Word | z |
|---|---|
| anyonehim 22 · them 460 | -5.12 |
| numberhim 15 · them 334 | -4.40 |
| whathim 50 · them 408 | -3.39 |
| justhim 78 · them 502 | -3.11 |
| buthim 36 · them 297 | -2.91 |
| hahahim 0 · them 103 | -2.89 |
Out of 453 words tested, the single most statistically distinctive thing about how he wrote is that he said other men’s names. Meanwhile the words that mark everyone else are haha, just, gonna, anyone. They were chatting. He was calling the roll. Nobody designed this measurement to find that, and it arrives where the rest of this site arrives by a completely independent route.
Method 06 · Burstiness Signal
Was he reactive, or was he a metronome?
Human messaging is not random. It clusters into bursts with long silences between. Goh and Barabási’s coefficient puts that on one scale: +1 is maximally bursty, 0 is Poisson random, −1 is perfectly periodic.
| Who | B |
|---|---|
| Ben1055 gaps · mean 187 min | 0.279 |
| Everyone else combined7760 gaps · mean 25.6 min | 0.562 |
| People steadier than himof 29 with 120+ messages | 15 |
The team arrived in spikes, around events, when something happened. He arrived at a constant rate for 128 days. This is the quantitative version of showing up, and it is the one trait here that requires no talent at all.
Method 07 · Lagged cross-correlation No effect found
Did the 3 AM work show up the next day?
The rest of this site asserts emotionally that the midnight-to-6AM shift mattered. This is the only method that tries to prove it, and it fails.
Series x is his messages between midnight and 6 AM per day, 110 in total. Series y is everyone else’s volume per day. Correlated at lags of 0 to 3 days, then repeated on first differences, because both series trend upward across the summer and that alone would manufacture a correlation.
| Lag | Raw r | Differenced r |
|---|---|---|
| Lag 0same day | +0.108 | +0.065 |
| Lag 11 day later | -0.031 | -0.060 |
| Lag 22 days later | -0.077 | +0.047 |
| Lag 33 days later | -0.193 | -0.171 |
Nothing clears the band, and the differenced column is the honest one. I am leaving this in at full length because deleting it would make the other seven less believable. The honest reading is not that the night work was pointless, it is that chat volume is the wrong outcome to measure it against. Car groups and route bookings prevent problems. Prevention does not show up as noise the next morning. It shows up as an absence, and absences are invisible to this method.
Method 08 · Entropy rate Contradicted prediction
Did the room develop a grammar?
Every Shannon number on the Butterfly page is order-zero. Shuffle a week into random order and it comes out identical, because it only measures how concentrated attention is. It is blind to sequence. This asks a different question: given who just spoke, how surprised are you by who speaks next?
True Kolmogorov-Sinai entropy is a supremum over partitions of a limit and is not computable from 9,007 samples, so this is the entropy rate, the computable stand-in, estimated two independent ways. With nine symbols an order-1 model needs 81 cells, and plug-in estimates are biased downward whenever cells are sparse, so a falling number could be an artefact rather than a finding.
The control fixes that. Recompute the same statistic on 300 shuffled copies of the same window. Shuffling destroys order but preserves the distribution and the sample size, so it carries the identical bias. Whatever gap survives cannot be bias.
Theory says shuffling makes a sequence independent, so the shuffled rate must equal H₀. Measured, they differ by one ten-thousandth of a bit. The estimator is doing exactly what it should, which is the licence to believe the rest.
| Month | H₀ | Rate | Gap | σ |
|---|---|---|---|---|
| 2019-04339 messages | 1.869 | 1.806 | 0.052 | 4σ |
| 2019-051867 messages | 1.631 | 1.549 | 0.076 | 19σ |
| 2019-062454 messages | 2.128 | 2.020 | 0.105 | 30σ |
| 2019-072215 messages | 2.101 | 1.987 | 0.111 | 26σ |
| 2019-081972 messages | 1.990 | 1.900 | 0.087 | 22σ |
I predicted H₀ would rise while the rate fell. Both rose, so the prediction was wrong. But the gap tells a different story than the raw number: over the whole summer the room sits 100σ from its own shuffled self, and the gap doubled, from 0.052 bits in April to 0.111 in July, one month before the rumble. Structure above chance grew all summer. Raw predictability never improved because the room was getting more diverse at the same time, and diversity swamped it. Relative to its own baseline the team was becoming measurably more organised. Measured absolutely, it was just getting louder.
What eight passes actually bought
Six methods returned something. One contradicted a prediction written down beforehand, which is the most useful outcome available, because it is the only one that could have caught me fooling myself. One returned nothing at all.
The two independent confirmations matter more than any single number. Change point detection on message counts and eigenvector centrality on the mention graph share no inputs and no mathematics, and both say the same thing. And the finding nobody went looking for is method 05: given 453 words and no instructions, the estimator decided that the most distinctive feature of Ben Egan’s writing was other people’s names.
Corpus: 9,007 human messages from the Greenix Philly GroupMe, April 16 to December 14, 2019. Methods 01, 02, 06 and 07 use the 139-day window April 16 to September 1. Nothing was withheld. Every method that was run appears here, including the one that failed and the two that went against prediction. Aliases merged: Benjamin Egan; Robbie Hutchins with Real Robbie Hutchins.
2019 to now · things that rhyme
Nine coincidences, and one honest explanation for all of them.
Everything on this page is two true facts sitting next to each other. Some of them will give you chills. None of them are evidence of anything, and the last section says why that is still the interesting part.
One day, four records
August 13 was the best and worst day of the summer at the same time.
Your shortest sleep window of the entire summer. 4 h 42 m. You posted the rain speech at 11:31 PM, were still going at 2:15 AM, and were back at 7:00 for the 7:30 meeting.
●
●
Your longest day on record, 19 h 55 m. The loudest day in the chat’s history, 179 messages, more than any of the other 138. And at 10:45 PM, this.
Game. Set. Match.
WE FUCKING WON THE RUMBLE!!!!!!!!!!!!!!
The single hardest day of the summer, off the least sleep of the summer, in the rain, is also the day you won it. Nobody scripts that. It just happened to be the shape of the thing.
125 days apart
He said the word first. Robbie said it last.
We don’t lose in the grit. Period.
Grit baby
Robbie Hutchins
Of the eighteen times that word appears in 9,162 messages, Ben has the first and Robbie has the last. Seven years later the guy holding the last one is the guy who went back, pulled the whole archive, and counted the other seventeen.
47 days apart
Josh called the shot on the Fourth of July.
Can’t wait to do over 10,000 accounts.
9968
· 7 more!!! Who will get it?!
· We are #1!
He named the number in July as a thing to dream about. Six weeks later the chat counted the last thirty-two of them down in real time.
Eight months early
Your co-founder was already being quoted in the room.
John Taylor Advice. To me the three keys to success on the doors are confidence, mindset, and work ethic.
Grit Marketing is founded by John Taylor, Ben Egan and Garth Massey.
In the summer of 2019 he is a name someone quotes in a group chat in Pennsylvania. Eight months later he is a co-founder. Jordan Massey is in this chat too.
Two years apart
The company you were measuring yourself against became the partner.
Number 1 regional at aptive posted his networks sales to service ratio the other day and it was 78%… Our sales to service ratio as an office is almost 95%.
Aptive Environmental and Grit Marketing announce a partnership.
The 78% you beat in a group chat at 9:18 in the morning turned out to belong to your future partner.
The thing you were not trying to teach
Your most contagious phrase was the one for saying yes.
Jimmy: Alex the tech bought them for us. Can he come for a root beer float?
You: Of course
Skyler, Nick, Jacob, Tomas. Every one of them uses it as the answer when somebody asks for something.
Second only to grit, out of everything you said in four months. Not a rally cry. Not on a whiteboard. Just the two words you happened to use every single time somebody needed something.
Same afternoon
On the day the record fell, your contribution was drinking water.
The chat is 32 accounts from the biggest pest control summer any team has ever put up, counting down live.
In all seriousness make sure to stay hydrated today down in Delaware it is hot as hades and I don’t want anyone to get heat exhaustion or heat sickness. Please be careful and take care of yourselves
Not a countdown. Not a push. The record was going to happen with or without you. Somebody getting hurt in Delaware was not.
Nobody chose this
The butterfly picked itself.
One sentence, worth two hearts, on day three, ends up in eight men’s mouths and on a building. The technical name for that shape is sensitive dependence on initial conditions.
Mira Homes, the newest company under Grit, has a butterfly for a logo.
The metaphor came out of the arithmetic before anyone remembered the logo.
Seven years to the day
And the reason any of this exists.
Robbie went looking for John Taylor’s three hour recording, to run through a sales coaching tool he has been building. He never found it. He found this chat instead, and read most of it.
Guys for the brotherhood and for team I would love to do 100 this last day with everyone!
A failed search for a recording turned into this. And the On this day box on the first tab is showing that exact message right now, because today is August 15.
The honest explanation
None of this is fate, and it would be a smaller thing if it were. Coincidences are not evenly distributed across people. They pile up around whoever generates the most surface area.
Show up 128 days out of 131. Write 20,390 words. Say 106 people’s names out loud. Answer forty-eight men in a median of one minute fifty-five seconds, for four months, at three in the morning. Do that, and the sheer density of what you touched means the patterns will find you. Not because the universe was arranging things. Because there was that much of you in the record.
That is the only supernatural thing on this website, and it is not supernatural at all. You just did a lot of small things on purpose, and coincidence is what accumulation looks like from the outside.
April 16 to August 24, 2019 · 128 days, counted
What it cost, in timestamps.
Everything in the first table is measured directly from the archive. No interpretation, no rounding in your favour.
| What | Value |
|---|---|
| Days on dutyOut of 131 calendar days. You missed three. | 128 |
| Longest streak without missing a dayTwo and a half months, unbroken | 73 |
| Working span, median dayFirst message to last, measured 5 AM to 5 AM so a night belongs to the day that earned it. Median first: 10:32 AM. Median last: 11:23 PM. | 12 h 17 m |
| Longest single dayAugust 13. 7:00 AM to 2:55 AM. The rain day, straight off the shortest night of the summer. | 19 h 55 m |
| Days longer than sixteen hours | 14 |
| Days that ended after midnightOut of 128 | 48 |
| Sleep ceiling, median nightGap between your last message and your first the next morning. This is the maximum you could have slept, not what you slept. You were running meetings inside this window. | 11 h 14 m |
| Nights with a ceiling under six hours | 4 |
| Shortest night of the summerThe night of the rain speech. Posted 11:31 PM, still going at 2:15 AM, back at 7:00 AM for the 7:30 meeting. | 4 h 42 m |
| Median time to answer somebodyForty-eight guys. 218 replies came back in under two minutes. | 1 m 55 s |
| Messages that were you answering someone41% of everything you wrote was reactive | 424 |
| One-to-six-word answers“930”. “Yes”. “Ready”. “Knock it”. A third of your summer. | 328 |
| Written pieces over fifty wordsThe speeches. Longest was 192 words. | 105 |
| Words written159 a day, every day, on top of the job | 20,390 |
| Distinct people addressed by nameIn a group of forty-eight. You were tracking more than double the roster. | 106 |
One hundred and six people. You wrote a hundred and six different human beings’ names into that chat, by hand, and the median time it took you to answer any one of them was one minute fifty-five seconds.
One ordinary Thursday
Not a record day, not a rivalry week. June 6.
From 2:48 AM to 11:53 PM, and the message you spent 3:03 AM on was telling everybody else to sleep in.
The heartbeat
Every message in the group, by day, across the whole summer.
It starts as a dozen guys. It builds through June. It spikes in the second week of August and then, within days of the last knock, it stops. Nine thousand one hundred and fifty-nine of the 9,162 messages were sent in 2019. After that a handful of stragglers, then five years of silence broken only by the notice that somebody left the group.
What the day actually cost
The chat can only see what you typed. It cannot see the calls, the drives, the one-on-ones, or the doors. So this table is a model, not a measurement. The evidence column is what the archive supports. The hours are estimated.
| Activity | Evidence in the archive | Est. |
|---|---|---|
| Running the morning meeting | Meeting time posted nearly every day: 9, 9:30, 10, 11. Agendas written out in advance. | 1.5 h |
| One-on-one pitch reviews | Named schedules in ten and fifteen minute blocks, run in groups of four. | 0.5 h |
| Car groups and area assignment | 86 posted rosters, roughly 25 names each, plus the corrections after. | 0.5 h |
| Routes, jobs, technician coordination | Constant. Booking, rescheduling, chasing Ashley, moving jobs to Donzza. | 1.5 h |
| Answering people | 424 reactive messages, four separate bursts a day, median response under two minutes. | 0.75 h |
| Your own doors | Still selling in August. “Finishing with 8 above $800.” Stung mid-pitch and kept going. | 4.5 h |
| Before a single mile driven | Reconciles with the measured 12 h 17 m median span. | 9.25 h |
Whatever the true figure was, it went to other people. Eighty-two per cent of everything you wrote that summer was either a reply to somebody, an operational instruction, or a one-line answer. The speeches everyone remembers are the other eighteen.
August 12 and 13, 2019 · the loudest day in the archive
Twenty-four hours, 179 messages, and rain.
Going into the last full week against Chicago, barely ahead, with a forecast that would have given anyone an excuse. At half past eleven at night you wrote this and went to bed.
Twenty-seven minutes later Josh answered. Note what he does with your words. He does not paraphrase you. He cites you.
They went out at 7:30 the next morning in the rain. August 13 became the loudest day in the history of this chat: 179 messages, more than any of the other 138 days it was alive. At ten past ten that night, with the day won, you wrote back the part that separates you from every other manager who has ever given a speech about rain.
August 7, 2019 · the phone call
Somebody told you it was not sustainable.
August 20, 2019 · ten thousand
The record went down live, in this thread, in real time.
Six weeks earlier, on the Fourth of July, Josh had called the shot out loud: “Can’t wait to do over 10,000 accounts.” Here is the night it landed. Read the timestamps.
Thirty-two accounts from the biggest pest control summer any single team has ever put up, with the whole group counting down in the chat, and your contribution to the moment was to remind everyone to drink water.
That is the entire thing. That is why twenty of these men went on to run teams. Not the rain speech. The 3:37.
And at 12:26 AM, ninety minutes after the record fell, you were back in the thread asking who had Collin’s truck keys, because Ashley needed the truck at nine.
August 15, 2019 · 12:26 PM · the last ask
The final thing you asked that summer was not for a number.
The roster · fifty men
These are the names, as they appeared in the chat.
Nicknames left exactly as they were set in the summer of 2019, because that is who these people were to each other.
Twenty of them went on to lead teams or build companies. Ten are still at Grit today. Every one of them learned what running a team at the highest level of this industry looks like by watching one guy do it in Philadelphia when he was young enough that nobody would have blamed him for doing it badly.
8,989 messages · 94 voices · the whole chat
The entire group, April to December 2019.
Everything anybody sent in Greenix Philly, verbatim. Links stripped. Try Detroit, rain, rumble, pizza, or anybody’s name. Set Who to read one man’s whole summer.
Two notes on the count. The full archive is 9,162 messages and every number elsewhere on this site was computed against all of it. This browsable copy holds back eighteen, which is where a few 2019 jokes have not aged into something worth hosting on a page with your name on the door. The rest is exactly as it was sent.
Tap a name to pull every mention of that man, then send it to him.
Loading the archive…
Built from the complete Greenix Philly GroupMe archive: 9,162 messages, April 16, 2019 to July 6, 2024, 48 members. All quotes verbatim, all timestamps and heart counts as recorded. Times shown in Eastern. Grit Marketing figures from public reporting (GlobeNewswire, 2021; PR Newswire, 2026). Rep and revenue totals are Robbie’s estimates and are labelled as such. The 2019 account total and the record itself are his account of the summer.