Responding on the Consumption Side: Eight Moves and the Ramped Commit

Transcript

Full narration of the briefing. Click a section heading to jump the player to that moment.

Forecast, then the alternative 0:00

Session five: how to respond on the consumption side, in eight moves. I am Tom, Claire is with me, and this closes the Google AI Negotiation Series. Move one: forecast in Google's meters. Tokens by model, actions, seconds of video, images, tool calls, and grounding requests, from your own billing exports and pilot telemetry, for three years.

Compare it with the account team's forecast and expect a large gap. Move two: establish the alternative. A second model provider running a real workload. It is the only leverage Google respects on tokens, and the only thing that makes a rate hold negotiable.

Fix, cap, size below 0:48

Move three: fix the unit rates. Token rates held past December thirty first, twenty twenty six; successor models at equivalent tier priced no higher; ninety days notice on deprecation with a migration credit; and media and agent unit rates in the order form. Move four: cap what you cannot forecast. Monthly ceilings on media, avatars, and agent consumption, with alert thresholds and an approval gate.

A ceiling you never reach costs nothing. Move five: size any plan below the forecast. Flexible Savings Plan or committed plan at no more than eighty percent of your own number, the balance pay as you go, and the symmetric downside modeled before signing.

Size the commit from both sides 1:30

Move six is the one that matters most: size the commit from both sides together. One ramped floor covering Workspace, Gemini Enterprise, Vertex, and infrastructure, from your combined forecast, with shortfall rollover, coterminous end dates, the committed use discount scope confirmed in writing, and Marketplace drawdown to the permitted cap. On the illustration that is the difference between a thirty six million dollar flat annual floor sized to Google's model and twenty seven million ramped and sized to yours. Everything else in this series is what makes that number honest; this is the number.

Year two terms and the calendar 2:11

Move seven: secure the terms that decide year two. No training on customer data. Zero data retention for prompts and outputs where available. The generative AI indemnity confirmed for the products in scope.

Data residency by region. And support as a fixed amount rather than a percentage of the commit, because a percentage grows with every line you add. Move eight: sign on your calendar. Google's quarters close in March, June, September, and December; the proposal, the credits, and the executive escalation cluster there.

Set the timeline from your renewal date, and hold it.

The mistakes that cost the most 2:49

The mistakes that cost the most, across both sides. Accepting the Workspace AI uplift on every seat. Signing Gemini Enterprise for the whole workforce on a demo. Building the three year token budget on introductory rates and discovering the standard rates on the first January invoice.

Leaving avatar, video, and agent consumption uncapped because nobody could forecast them. Committing to a Gemini Enterprise plan sized to Google's forecast and paying more than pay as you go. Negotiating Workspace and Cloud in separate rooms. And treating the Google proposal as the only proposal, because without a second provider on a real workload, every discount is a favor.

How Redress helps 3:34

How Redress helps. We run the exercise from the buyer's side across both sides of the Google relationship: the Workspace edition and seat reconciliation, the Gemini Enterprise pilot design and staging, the consumption forecast in Google's meters, the plan and commit model, the benchmark on seats, tokens, and media rates, the contract terms that decide year two, and the negotiation itself on your calendar. Our fees are fixed, and every dollar removed from Google's number belongs to you. The full research note is free under this video, and the whole series lives at redresscompliance dot com slash google ai negotiation series.