I run a marketing company of about fifteen people spread across Morocco, the Gulf and the US. No HR department. No talent function. For most of the company's life, talent acquisition meant me, a Google Doc, and a job post I had assembled by reading four other job posts.
That last part is the confession. My job descriptions were a collage. I had never once sat down and asked what the document was actually doing. I assumed it was an advertisement. It is not. It is a filter, and it runs before you get a say in anything.
Talent Acquisition Starts Before Anyone Applies
Here is what I did not understand for years. By the time a CV arrives in your inbox, the most consequential decision in the process has already been made, and you did not make it. The candidate made it. They read your post and decided whether the role was for them.
Every word in that document is doing selection work whether you intended it or not.
Mine selected for the wrong thing. One post asked for "5+ years managing paid media campaigns at scale." I wrote it because it sounded serious. What it did was tell every good marketer with four years and unusual range that this was a role for a specialist who had done one thing for a long time. It also told a certain kind of applicant, the kind who counts years, that they qualified. Guess who applied.
The post said nothing about what the work felt like. That we work across time zones and write a great deal. That nobody above you decides what to do next. That a client's revenue depends on a judgment call you make on a Tuesday with incomplete information. All of that is the job. None of it was in the description.
What a Small Company Sees That a Large HR Function Cannot
A big HR function is built around a funnel. Its inputs are volume, time-to-hire, cost per hire. Reasonable numbers when you are filling forty roles a quarter.
At fifteen people, I fill maybe three roles a year. I have no funnel. I have three decisions, each of which is roughly 7% of the company. That changes the mathematics completely.
A wrong hire at two thousand people is a line item. A wrong hire at fifteen is felt by everyone. The work drops on real accounts, the people around them absorb the difference and resent it. I have watched one bad hire cause a good person to leave, a cost that lands in no HR spreadsheet.
So the incentive is inverted. A large function wants more qualified applicants. I want fewer. I want the post to actively repel almost everyone, including some people who could probably do the job, because the cost of processing a near-miss is enormous relative to my capacity and the cost of a wrong yes is existential.
The Rewrite
Two years ago I threw out the collage and wrote a post that tried to do only one thing: let the right person recognize themselves and let everyone else self-select out.
It opened by naming what was hard about the job rather than what was attractive. That the client roster spans several markets and languages, that you will be asked for an opinion before you feel ready to have one, that nobody will tell you your priorities on a Monday.
It removed every year-count requirement. All of them.
It said what we would not offer. No office culture in the traditional sense. Limited hand-holding. If you need a manager to structure your week, this will make you miserable, and I said so in that sentence.
Applications dropped by roughly two-thirds. The conversations changed so much I thought at first I had posted to a different channel. People arrived already knowing what the role was. Several told me they had applied because of the paragraph about what was hard.
What Actually Keeps People
The retention half is simpler, and almost nobody does it, because it is boring. People stay when the work is interesting and they can see that it mattered. That is nearly the whole thing.
In a small distributed company, there is no promotion ladder to dangle and no office to make people feel part of something. What we have is the ability to hand someone a real account and let the outcome be visibly theirs. When a client's organic traffic recovers, everyone knows whose call that was. That visibility is the compensation a fifteen-person company can afford, and it is worth more to the right person than a title.
That is the deal here, and it is the only lever a company this size genuinely controls.
The second thing is honesty about the state of the business. Our team knows what is going well and what is not. When a large account paused last year, I told everyone the week it happened rather than the month it became obvious. There is a version of leadership that treats staff as people to be protected from information. It does not survive contact with a small team. They already know. Pretending otherwise just tells them you will lie about other things.
The third is keeping the work current. Marketers leave when the job stops teaching them. Everyone at the digital marketing agency in Dubai side of the business is dealing with a search environment that has genuinely changed, where assistants now answer questions directly and cite sources, which Google has documented as generative results roll out. That is not a threat to a curious person. It is the reason they stay. The way buyers actually research a decision keeps shifting, and Google's own work on search behaviour keeps confirming that the ground moves. A marketer who is bored is a marketer who has stopped being trusted with hard things.
Everything else, the perks and the process and the engagement survey, is downstream of those. If the work is real and the information is honest, people stay. If it is not, no HR initiative will hold them.
I still write the job posts myself. I doubt I will ever hand that off. It is the only document in the company where a bad sentence costs me a person I will never meet.