Anchoring a salary negotiation: the strongest result in the field

Anchoring is the best-measured phenomenon in negotiation research. A preregistered meta-analysis of 374 effects found a general first-mover advantage of g = 0.42, and a correlation of r = .62 between the size of the first offer and the final agreementPetrowsky et al. 2025 · strong evidence
Preregistered meta-analysis: 374 effects, 90 studies, N = 16,334. — where the conversation starts is, to a remarkable degree, where it ends. But the same analysis prices the aggression: ambitious first offers produce more impasses (g = −0.42) and leave the other side feeling worse about the deal (g = −0.40), with anger driving bothPetrowsky et al. 2025 · strong evidence
Two preregistered experiments (N = 2,121) traced impasses and recipient dissatisfaction to anger.. A job offer is a repeated relationship with the person across the table. Anchor — and decide in advance how much goodwill you are willing to spend.

Only anchor informed

“Always go first” is wrong. Under information asymmetry the first-mover advantage reverses: negotiators who opened without knowing the market value anchored catastrophically against informed counterparts, and both sides did better when the uninformed party did not openMaaravi & Levy 2017 · moderate evidence
The advantage reverses when you don’t know the range — the case for researching the band first.. The rule this converts to: anchor first only if you have researched the band. If you haven’t, your move is a question, not a number.

Precise numbers — with the reason attached

Precise first offers anchor harder than round ones, because a precise number reads as more plausibleLoschelder, Stuppi & Trötschel 2014 · moderate evidence
The “€14,875?!” paper — precision works through plausibility.. But precision has a boundary: against expert counterparts, extreme precision backfires — an inverted-U where too many digits read as incompetence — and the backfire disappears when the number comes with a rationaleLoschelder et al. 2016 · moderate evidence
Five experiments, 1,320 participants. The backfire is expert-only; a stated reason removes it.. A compensation partner is the expert in this scenario, so the escape hatch matters: say the number, then the reason. Overly precise offers also make counterparts less willing to negotiate with you at all, because they read precision as inflexibilityLee et al. 2018 · moderate evidence
“Too precise to pursue” — six studies plus 11,203 real-estate listings..

If a single number feels exposed, use a bolstering range: ranges with your target at the bottom and a more ambitious number on top drew better counteroffers than equivalent point offers — the economic gain without the relational cost of an aggressive point demandAmes & Mason 2015 · moderate evidence
“Tandem anchoring”, five studies. Target at the bottom of the range, never the top..

The pause after their number

Silent pauses of at least three seconds predict value creation in negotiation, by prompting more deliberative thinkingCurhan et al. 2022 · moderate evidence
Silence promotes reflection and reduces fixed-pie assumptions. — with a caveat we ship alongside the technique: when status differences are salient, silence created value for the higher-status party but not the lower-status oneCurhan et al. 2022 · moderate evidence
A job candidate is usually the lower-status party — silence is composure training, not a guaranteed lever.. A candidate is usually the lower-status party, so we train the three-second pause as composure — the ability not to concede against yourself while the silence hangs — rather than as a profit lever.

This page covers the anchor. The rest of the arc — trading beyond a fixed band, concession discipline, the pressure round — carries the same rule everywhere it goes: where the evidence is practitioner method rather than measured finding, the practice says so out loud.