The scenario: negotiating with someone who reads the same data you do
In new-offer mode, a candidate is negotiating with a hiring manager or a fellow recruiter who has access to the exact same salary benchmarking tools the candidate uses in their own job, which means a vague or padded number gets noticed immediately, and the negotiation rewards precision over a confident guess.
In internal-raise mode, an employee who closed a surge of open requisitions on a tight timeline, or built a program that measurably improved offer-acceptance rates, is asking their own manager for more, and the conversation tests whether that result gets tied to a specific number rather than a general claim about a busy quarter.
What gets covered
New-offer mode has the AI on the other end respond the way a hiring manager with real market data would, questioning a number that doesn't match the actual band rather than accepting it because it sounds reasonable.
Internal-raise mode has the manager ask for the specific requisition count, timeline, or metric behind the cited result, since a raise case inside HR needs to hold up to the same scrutiny a recruiter would apply to anyone else's ask. Either mode also tests whether a candidate holds a stated number once the other side pushes back once, rather than caving immediately to a counter.
Scoring
What actually gets scored here
Calibration floors and caps matter directly in this context, since a candidate who sounds confident while citing a number that doesn't match real market data won't score higher than one who states a precise, defensible figure, even though confidence alone might work in a field with less access to comparison data.
Frequently asked questions
Does it help to practice against someone who might actually know the market data?
Yes. The AI on the other side responds the way a hiring manager with real compensation data would, pushing back on a number that doesn't hold up rather than accepting a confident guess.
What if my hiring-surge result doesn't have a clean number attached?
It still helps to state whatever specifics you have, the requisition count or the timeline, since a vague claim scores worse than a concrete one even under generous calibration.
Is internal-raise mode useful for a program I built, not just a hiring push?
Yes. A program tied to a measurable result, like an improved offer-acceptance rate, works the same way as a hiring-volume story, as long as the specific metric behind it gets named.
Does sounding confident help if my number doesn't match the actual market?
No. Calibration floors and caps mean a confident tone alone can't outscore a precise, defensible figure, which matters more in this field given how easily a mismatched number gets noticed.
Can I practice negotiating a title or level change instead of just a salary number?
Yes. Both modes work for stating a specific ask out loud, whether that's a salary figure, a title, or a level, and the AI on the other side responds to whatever position is actually stated.
Related pages
Rehearse the comp conversation before it's real
Practice negotiating your own HR or recruiting offer, or asking for a raise after a hiring surge, with an AI that pushes back on the data.