Full narration of the briefing. Click a section heading to jump the player to that moment.
Last episode was deliberately discouraging, so let us balance it. You have more leverage in this negotiation than in almost any mature software category, for a straightforward reason: the market is young, the vendors are competing hard for enterprise reference customers, and switching is more credible here than it is in a database or an ERP. I am Claire, Daniel is with me, and these are the five things that actually move the outcome, roughly in order of how much they are worth.
The strongest is genuine portability, and the word genuine is carrying weight. Multiple capable model families exist, and your application can call more than one. But as we said last time, the cost is revalidation, not integration. So portability becomes real leverage at the point you have actually run your evaluation suite against an alternative model and know what happens.
Do that work before the negotiation and you can describe it in the room as a fact. Claim it without doing it and you will be asked one question you cannot answer.
The second is the route, and it is the most underused lever we see. The same models are available through the large cloud platforms as well as direct. If you already hold a substantial cloud commitment, consumption bought through that platform may draw down money you have already promised to spend, which changes the real cost to you even at an identical headline rate. It also gives you two commercial conversations instead of one.
Model both routes properly before you choose, and let each know the other exists.
Third, and this is where we would spend the negotiation. Ask for structure rather than a bigger discount. A ramped commitment that starts at your confident floor. Rollover of unused commitment rather than forfeiture.
Term wide measurement instead of annual. The better of your rate and the published rate. Every one of those is worth more over the term than a couple of points on the headline, and they are often easier to win, because they cost the vendor less than a rate concession and they solve a problem you genuinely have.
Fourth, be honest about what you are worth beyond the money. In a market competing for enterprise credibility, a named reference in a regulated industry, a public case study, or a willingness to talk to other buyers has real value to a vendor. That is tradeable. It is also the one concession we would encourage you to price carefully rather than give away, because it is easy to offer in a good mood at the end of a negotiation and it is worth more than most buyers realise.
Fifth, timing, which applies to every vendor including this one. Sales organisations have periods, targets and quarter ends, and a deal that closes inside one is worth more to them than the same deal a fortnight later. We do not have public filings to point at here the way we do with the mature vendors, so treat this as ordinary commercial reality rather than a documented pattern. But do find out when their period ends before you decide when to be ready to sign, because it costs you nothing to ask.
And one thing not to rely on. Do not build your position on a threat to leave that you have not tested, because in this category the vendor's engineers understand your revalidation problem better than most account teams understand anything. An untested switching threat is the weakest card in the deck here, and playing it badly damages the rest of your position. Test it or do not use it.
Next time, Tom and I close the series on the part that decides what you actually pay, which happens after signature.
Redress Compliance works on contingency: our fee is 25 percent of what we save you. Nothing saved, nothing paid. Independent, buyer side only, never vendor funded. Want Redress to contact you? Reach out and we respond the same day.
Talk to a Anthropic negotiator