A small research note on cold outreach. With data. And a French expression

A few weeks ago, I posted a rant on LinkedIn about cold outreach emails: the kind that start with “Hi Isabelle,” as if we were old friends, go on for three paragraphs of self-congratulation, and end with an exclamation mark. 

The rant referenced a French expression I am particularly fond of: “on n’a pas gardé les vaches ensemble”. Loosely translated as: “we did not herd cattle side by side in a field”. We have no shared history. You do not know me, and I do not know you — so perhaps, just perhaps, you might want to dial down the familiarity.

The post received more attention than I anticipated, which, I must admit, was reassuring. It confirmed that I am not alone in my reaction. So, rather than simply complain, I decided to conduct a small cold outreach analysis and spent ten days collecting data.”

The methodology, such as it is

Between 11 and 20 May 2026, I read every unsolicited email that landed in my Junk folder. 357 in total, averaging roughly 35 per day. I separated newsletters and digests from genuine cold outreach, which left 291 emails in the core dataset. These are the ones claiming, implicitly or explicitly, to have a reason to contact me specifically.

In the interest of full disclosure, I enlisted my friend Claude to assist with the methodology and the data analysis. It was not an entirely successful endeavour, and I will write another post explaining the vagaries of working with AI. Setting that small contretemps aside for a moment, Claude and I scored each email on eight dimensions: greeting type, name accuracy, tone, quality of personalisation, email length, spelling and grammar, exclamation mark usage (a particular favourite of mine), and vocabulary. All senders have been anonymised, as the focus here is not on individuals, but on patterns.

The patterns are, as it turns out, quite something.

The email greeting

Forty-four per cent of cold outreach emails opened with “Hi.” A further 2.7% began with “Hey.” As I have noted before, these are not professional greetings between strangers. They are the greetings of people who have already decided that warmth, or at least mutual respect, is a given.

What I find most striking, however, is that 45% of these emails omitted a greeting entirely and proceeded directly to the pitch. No salutation, no acknowledgement — simply the product, the conference, or whatever the offer happened to be. Let’s just pause for a moment: 45% of cold outreach emails that claim to be personalised did not address me at all.

Of the emails that did use my name, not one opened with “Good morning” or included any professional acknowledgement.

One sender addressed me by my last name: Dear Ms. Duarte. Points for formality, but points immediately deducted for omitting the accent. The correct spelling is Duarté. It has appeared on my LinkedIn profile, my email signature, and my website for years.

Of the five emails that used my last name in any form, not one included the correct accent. The é was never attempted. I appreciate that accents can be challenging for native English speakers, but these emails are, after all, attempting to convey familiarity and care.

Whether the name appeared at all

Almost 39% of all cold outreach emails, or 113 out of 291, did not use my first or last name at all. It was absent from the greeting, the body, and the opening. In essence, these were emails addressed to nobody in particular.

I should note that what I classified as “incorrect” was broader than simple misspellings: it included instances where my name was replaced entirely by something else — “Hi GTMT,” or “Hi There” in place of “Hi Isabelle.” Five emails fell into this category. What genuinely surprised me was finding this so rarely; I was quite certain it would be more prevalent. The more telling finding is the one above: 39% of emails never attempted my name at all.

A further 11.7% placed “Isabelle” in the subject line — not as a greeting, but as a tactic to increase open rates. This is not personalisation; it is the typographical equivalent of waving, and it is equally irritating.

The illusion of personalisation

Claude and I classified personalisation on four levels: none, fake, surface, and genuine.

“None” means the email contained no reference to me, my company, or anything suggesting the sender had spent more than three seconds on research. 36.8% of cold outreach fell in this category.

“Fake” refers to emails that appeared personalised but were clearly templated — merely a company name inserted into a generic pitch. This accounted for 39.5% of emails and was the largest single category. 

“Surface” means a passing reference to my industry or job title, without any real engagement with what I might actually care about. That was 19.6% of emails.

“Genuine” means the sender had visibly done some real homework — it either referenced something specific about my company, a recent piece of content, a known position, or a real business challenge. That accounted for a paltry 4.1%, or just 12 emails out of 291.

Those twelve emails shared another characteristic: they were shorter, averaging 115 words. They also contained no spelling mistakes. My favourite detail is that not one of them used a single exclamation mark. 

I do not believe that correlation is accidental.

The fake RE: and FW: problem

6.5% of cold outreach emails – nineteen in total – used a RE: or FW: prefix in the subject line. In every instance, there was no prior email, thread, or history. It was simply a manufactured visual cue intended to suggest I had already participated in a conversation that had never occurred.

This takes the cattle idiom to another level. Familiarity is not just assumed; in many cases, it is actually fabricated.

Punctuation as personality

There were 146 exclamation marks across 291 cold outreach emails. 71.5% of cold outreach emails used none at all — which makes the ones that did use them all the more noticeable.

The word “please” appeared 138 times in the dataset. The word “transform” and its variants appeared 150 times. The word “sorry” appeared zero times. Not once, across 291 cold outreach emails. The emails were marginally more invested in being courteous than in describing themselves as transformative, but they were not at all interested in apologising for the intrusion.

Perhaps the most ironic entry was an email offering advice on running better email campaigns. It contained multiple exclamation marks, seemingly oblivious to the irony.

The language of cold outreach

I also catalogued the jargon, and the word counts tell a a story I did not expect.

The most-used term was not what I anticipated. “Transform” and its variants — transformation, transformative, transforming — appeared 150 times across 71 emails. It is a word that does a great deal of work in cold outreach: it promises fundamental change while committing to nothing specific. Close behind it: “solution” and its variants (92 times in 62 emails), and “innovative” and “innovation” (77 times in 52 emails). The word “innovative” is obviously self-awarded and comes with no external evidence. This might be why it is so popular.

“Ecosystem” appeared 28 times across 20 emails — borrowed, as I have hinted at before, from biology to make software partnerships sound more alive. One email used it four times in a single paragraph. Other notable entries included: “scalable” (15 emails), “streamline” (14 emails), “proactive” (13 emails), and “thought leadership” (12 emails). Not once did any of the thought leadership references contain actual thinking. The phrase appeared to function as a synonym for “a slot at our conference.”

A note on one term I expected to find more often: “game-changing” appeared just twice. I had assumed it would be endemic. In fact, the more insidious terms are the ones that sound reasonable in isolation – transform, solution, innovative – but lose all meaning when every email uses them simultaneously.

Then there is the matter of call scheduling. Fourteen emails – just under 5% – included a link to book a call directly. On reflection, I am genuinely torn about this practice. I appreciate that it is intended to reduce friction, and I understand the logic. In practice, I find it utterly presumptuous: the subliminal message is that your time is sufficiently valuable to warrant a booking system, but mine is not sufficiently valuable to warrant a personalised invitation. To add insult to injury, one email used the phrase “Schedule a call with me!” — with an exclamation mark. Less of a call to action and more of a shout to action. I don’t know about you, but I personally don’t like being shouted at.

The entry that most encapsulates the spirit of the entire dataset was an email inviting me to dinner at Nobu. The subject line promised an exclusive evening, the body addressed me as though we were old friends, and the email included a Calendly link. We had, to my knowledge, never spoken. I mention this not because dinner at Nobu is objectionable but because the combination of assumed intimacy, celebrity venue, and a self-service booking link perfectly captures the contradiction of modern cold outreach: the assumption of a personal relationship, delivered at industrial scale.

Length of these emails

Forty-five per cent of cold outreach emails exceeded 300 words, with an average length of 390 words. The longest extended to over 4,500 words, which, for the record, is quite a bit longer than this blog post.

The twelve genuinely personalised emails averaged 115 words. The emails that made the greatest effort to be relevant to their intended recipient were the shortest. Conversely, those with the least to say were the longest.

The errors

Twenty-eight and a half per cent of cold outreach emails contained at least one spelling, grammar, or punctuation error. The most common was the use of ALL CAPS mid-sentence, present in 22% of emails. “REGISTER NOW”. “CLAIM YOUR FREE PASS”. “LIVE TOMORROW”. Capitalisation is often deployed as a substitute for compelling copy.

Grammar errors were less frequent but more memorable: “We are bring together 300+ senior finance leaders” appeared in the first sentence of a conference sponsorship pitch. “Tech companies such as like Paddle, Vanta and Clay” was found in line one. This particular example made me laugh, as I could not decide whether it was a tautology or a deliberate attempt at informal language.

Perhaps the finest example in the entire dataset was one email that addressed me as “Hi GTMT.” The mail merge tag had inserted the email alias rather than a human name. Personalisation, at the very moment it was meant to demonstrate its value, failed completely.

One final observation, and one that is particularly relevant given that I am based in Europe: an unsubscribe option appeared in only 68% of cold outreach emails, meaning that nearly a third of senders appeared to consider the GDPR requirement optional. Make of that what you will.

What the data actually says

Cold outreach is hard. I do understand that, and I sympathise. Building pipeline from a cold start requires patience, resilience, and a tolerance for rejection that I find admirable.

However, this data suggests the industry has mistaken volume for value. The 291 cold outreach emails that landed in my junk folder did not fail because they were cold; they failed because they were careless with language, research, and basic courtesy. The simple act of reading one’s own email before sending would prevent many of these errors.

The twelve emails that came closest to eliciting a reply were short, accurate, and included a genuine observation about something I might care about. They understood what Soldo does and had a specific reason to believe their offering was relevant to me. They addressed me — not a trend, a job title, or an unresolved merge field.

I am not asking for much, and I don’t believe I am setting a high bar.

We did not herd cattle together. However, if you do your homework, introduce yourself properly, and write in clear English, I might just read beyond the first line.

Methodology

This analysis is based on 357 emails received between the 11th and 20th of May 2026. 291 are classified as cold outreach (excluding newsletters and digests). All senders have been anonymised. Body text was extracted from raw .eml files via programmatic analysis; HTML-only emails were stripped of CSS and style blocks before processing. Exclamation marks were counted from clean body text with CSS !important declarations excluded. This is a personal research note, not a peer-reviewed study — though it is more carefully sourced than most of the emails it describes.