The first number on the table sets the range for everything that follows. Here is the buyer side guide to anchoring well in an Anthropic negotiation, who should move first, and how to set an anchor you can actually defend.
The first number spoken in a negotiation does more work than any number that follows it. It becomes the reference point both sides measure against, it sets the boundaries of the range the deal will land in, and it quietly shapes what each party comes to think is reasonable. This is anchoring, and it is one of the most studied effects in negotiation because it is so reliable. In an Anthropic deal the anchor is usually the proposed commit level or the headline rate, and whoever sets it well gains an advantage that is hard to claw back later. Most buyers let Anthropic anchor by default, accepting the account team's opening figure as the natural starting point and then negotiating down from it. That is a weaker position than it needs to be, and understanding anchoring is how you fix it.
The reason anchors are so powerful is that people adjust away from them insufficiently. Once a figure is on the table, every subsequent move is framed as a step from that figure rather than an independent assessment of what the deal is worth, and the final outcome tends to land closer to the anchor than to where it would have settled if the anchor had been different. This is not a failing of weak negotiators, it affects experienced ones too, which is why it is a structural feature of the conversation rather than a trick you can simply see through. In a vendor negotiation the practical consequence is direct: if Anthropic opens with a commit or a rate and you spend the conversation arguing it down, the deal will tend to settle near their number, discounted a little, rather than near the number your own analysis says is fair. The opening figure has done its work before you have made your first real argument.
The conventional advice is to let the other side name a number first so you learn their position. In a vendor negotiation that advice is often backwards, because the vendor's opening number is not informative, it is strategic, set high precisely to anchor you. Letting them go first hands them the anchor and gives you little real information in return, since you already know roughly where list pricing and standard discounting sit. The stronger move, when you have done the work to support it, is to anchor first yourself with a number grounded in your own analysis. This reframes the entire conversation around your figure, and forces the account team to argue up from your position rather than letting you argue down from theirs. The condition attached to that move is the whole game: you can only anchor first effectively if your anchor is defensible, because an anchor you cannot support collapses the moment it is challenged, and a collapsed anchor is worse than none at all.
A defensible anchor is one built on evidence the other side cannot easily dismiss. In an Anthropic negotiation that evidence comes from two places. The first is your own consumption model, the build up that turns your real usage into a forward spend figure you can explain line by line, so that when you propose a commit level you can show exactly how you arrived at it. The second is benchmarking, an understanding of what comparable enterprises actually pay Anthropic at similar scale, so your proposed rate is anchored to the market rather than to a number you wished for. An anchor supported by both is hard to wave away, because it is not a negotiating posture, it is a reasoned position with the work shown. The contrast matters: an aggressive number with nothing behind it invites the account team to discredit it and reset the conversation on their terms, while a slightly less aggressive number you can defend completely holds its ground and pulls the settlement toward it.
An anchor should be ambitious but not absurd. Set it too soft and you have given away ground before the conversation starts, since the settlement will drift up from your opening and you will have anchored yourself low. Set it so aggressive that it is plainly unrealistic and you lose credibility, the account team dismisses it, and you forfeit the anchoring advantage entirely. The right anchor sits at the optimistic edge of defensible: a number you can justify with your model and your benchmarks, positioned at the favorable end of what that evidence supports, with deliberate room built in to move toward the middle as the negotiation proceeds. That room is essential, because a negotiation needs the other side to feel they have moved you, and an anchor with no give leaves nowhere for the deal to go. You are not trying to name your walkaway as your opening. You are trying to set a reference point that is defensible, favorable, and leaves a path to a settlement that still lands well for you.
Sometimes the vendor anchors first regardless, and when they do the worst response is to engage with their number as if it were the legitimate starting point. Negotiating down from an inflated anchor concedes the frame, and the deal settles near their figure. The stronger response is to refuse the anchor and reset it with your own. You do this not by rejecting their number emotionally but by calmly setting your evidence based figure beside it and making clear that your position rests on analysis while theirs rests on a list. This neutralizes the anchor by replacing the reference point, shifting the conversation from how far below their number you can push to where between two stated positions the deal should fairly sit, with yours grounded in evidence and theirs not. The discipline is to do this early, before their anchor has had time to settle in as the assumed baseline, because the longer an unchallenged anchor sits on the table the more it shapes what both sides treat as reasonable.
Anchoring is not only about the headline number, it operates on every term that has a range, and a buyer who anchors only on price leaves value on the table everywhere else. The overage rate, the unused commitment treatment, the price protection across the term, the ramp schedule, the renewal cap: each of these has a first number too, and each can be anchored. The strongest negotiators open with a full position rather than a single figure, anchoring not just the commit level but the structure around it, so the entire shape of the deal is framed by their proposal. This matters because vendors often concede on a secondary term more readily than on the headline rate, and if you have not anchored that term you will accept whatever default they offer. Set an opening position on the whole package, defend each piece with the same evidence based discipline, and you anchor the deal rather than just the price, which is where the durable value in an Anthropic agreement actually lives.
An anchor is only as strong as the credibility of the person setting it, which is why anchoring and preparation are inseparable. When you propose a commit level and the account team senses you cannot explain how you got there, the anchor evaporates, because an unexplained number reads as a wish rather than a position. When you propose the same level and can walk through the consumption model that produced it, the rate benchmarks that frame it, and the assumptions behind each, the anchor holds, because there is nothing for the other side to dismiss. This is the quiet reason buyer side preparation matters so much in a negotiation that turns on anchoring: the work you do before the conversation is what gives your first number the weight to move the settlement. A buyer who arrives with a binder of analysis and a buyer who arrives with a hopeful figure may name the same opening number, but only the first one's anchor will hold under pressure, because only the first one can answer the challenge that always comes. Credibility is not a soft factor here, it is the mechanism by which an anchor exerts force.
Anchoring first is powerful when your number is defensible, but it is not always the right move, and knowing when to hold back is part of using the tactic well. If you genuinely lack the information to set a credible anchor, if you have not yet built the consumption model or gathered the benchmarks, then anchoring first risks setting a number you cannot defend, which is worse than letting the other side open. In that situation the better path is to delay the number entirely, to keep the early conversation on requirements and structure while you finish the analysis, and to anchor only once you can do it with confidence. There is also the case where the vendor's likely opening is so far from reality that letting them anchor and then resetting hard is cleaner than trying to preempt it. The judgment is situational, but the principle is constant: anchor first when you can do it credibly, hold back when you cannot, and never set a first number you would be unable to defend if challenged, because a collapsed anchor concedes more than silence ever would. The tactic serves the preparation, not the other way around.
Get a quote for a bounded engagement. Fixed fee or gainshare, no risk to you.
Get a QuoteWeekly intelligence on Anthropic pricing moves and the buyer side counters that work.