Up to ten short questions after onboarding, one at a time, that teach matching what a thin profile is missing
The first ten messages are short questions Z sends after someone finishes onboarding. One question per message, by email or SMS, with numbered quick replies.
They are not a second onboarding interview. They are the handful of facts matching needs and onboarding did not collect: the brands a person already loves, the categories they will never touch, the lowest offer worth bringing them, where their audience is, or how many talent an agency represents.
There are banks of questions for talent, brands and agencies, and the series stops after ten messages at most.
A profile that just finished onboarding is thin. It has a name, some links, and whatever the person happened to say. Matching can still work, but it has to guess: a talent's price is an estimate from public sources, and nobody knows which brands they actually like or refuse.
Every guess weakens a score. The first ten messages replace guesses with answers, a few at a time, in the place the person already reads Z. Each answer is saved on the account straight away, most of them where matching reads them, so the next match is better without anyone on the team typing anything in.
An email arrives with the subject One quick question, or an SMS from Z. SMS is used when the person has a phone that agreed to SMS; otherwise it is email. It is one line, with numbered options when the question has them:
Which brands do you already use or love?
A few names is perfect, like Nike, Glossier, Celsius.
The person can reply with a number, a few names, or a price. Several numbers work too: 1 3 picks two options. A very short line is read as well as a long one.
When the answer is understood, Z sends back one short thank you that says what happens next, such as "Got it. I'll filter out anything below that." There is no form and no page to open.
Every email carries Reply stop and I won't ask again at the bottom. The first SMS in a series carries Reply STOP anytime to stop texts.
Once the series is switched on, Z looks for questions that are due about once an hour and sends those that fit the rules below.
Before a question is chosen, Z checks what is already known. Anything the person already told onboarding, an offer floor that is already set, a brand list, an earlier answer to the same question, all of it counts.
A person who filled in a lot during onboarding may get only a few questions, or none at all. When every question in the bank is answered, the series ends early instead of asking something for the sake of filling ten slots.
Every answer is saved on the account itself, not in a separate quiz.
| What the person answers | Where it lands | What changes |
|---|---|---|
| Brands they use or love | Brand signals on the account | Matching leans toward those brands and their category |
| Anything they would never promote | Blocked categories, or blocked brands | Those never reach them, however well the rest fits |
| The lowest offer they would take | Their offer floor, and the price Z filters on | Offers below it are flagged instead of sent as if they fit |
| The kinds of deals they want | Their deal preferences | Z remembers it; matching does not score on it yet |
| Where most of their audience is | Their profile | Audience fit in matching gets a real answer |
| Dates they are unavailable | Their availability | Z remembers it; matching does not score on it yet |
| A brand's budget per creator | The brand's deal preferences | Z remembers it; matching does not filter on it yet |
| The talent a brand wants | The brand's deal preferences | Z remembers it; it does not start a search by itself |
| A brand's competitors | The competitor set the brand joins | Talent tied to those brands get flagged |
| An agency's roster size | The agency's roster details | Z remembers it for the roster import |
| Who approves deals at an agency | The agency's approval contact | Z remembers who it is |
Brands also answer what kind of talent they would never use, which matching then treats as a hard no, and how fast they decide, which Z remembers. An agency answers which deals its talent want most, which Z remembers too.
An email reply is read in the same conversation the question came from, and only if it is a reply to the open question, within two weeks of it. An SMS reply is read within three days of the question, because an SMS has no thread to tie it to what was asked.
Most replies are read by simple rules, at no cost: a number, a name, $500, "none", "not sure". When the rules cannot read a reply, one cheap model call tries. If that also cannot read it, the reply is left alone and goes on to the normal chat, so a reply that reads as an instruction still reaches Z as a message.
A reply is only accepted on the channel the question went out on. A person who was asked by email cannot answer that question by SMS, and the other way around.
The series stops for good when:
The state is per account and can be seen on the First 10 page, including the reason it stopped.
The First 10 page in the admin sidebar is the window into the series.
Nothing in the series runs until the team switches it on, so shipping the feature never starts messaging people. That switch is the off switch for the scheduled sender: while it is off, only a person pressing Send N due now on this page can put a question out. Switching it off again stops the scheduled sends. The series has only this one switch; the level shown on the Automation page does not turn it on or off. A person can always end their own series with one word.
If the scheduled sender and a press of Send N due now happen at the same moment, each person still gets the question once: whichever run reaches an account first sends it, and the other skips that account.