Full narration of the briefing. Click a section heading to jump the player to that moment.
The commit contract is usually the biggest number in the Google relationship, and it is negotiated in exactly the wrong order by most buyers: discount first, relief later, exit never. This episode runs it in the right order. Because the discount is the headline, but the clauses are the deal. A deep discount on a commit you cannot safely carry is a loan you took against your own forecast.
Size the commit below the forecast. Your measured run rate from the billing export is the floor. Real committed growth, projects with budgets and dates, can join it. Aspirational growth stays out: if Google believes in your AI future, let Google fund it with credits, not you with commitment.
The ramp follows the same rule: year one at your migration reality, not the account team's optimism. A ramp you beat is leverage at every review. A ramp you miss is a shortfall conversation you scheduled yourself. If Google wants a bigger year one, let them buy it: onboarding credits, funded migration sprints, or a discount step that triggers at the higher tier.
Commit to 80, celebrate at 110, and let the renewal reprice the growth you actually delivered.
Negotiate the relief before the number. The default shortfall clause is pay the difference. The alternatives all exist in signed deals: rollover of unused commit into an extended term, carryforward into the next period, a one time commitment reduction right when the business changes. Google never offers them first, and they cost the most to ask for after signature.
And widen the drawdown while you are there: direct marketplace purchases count in full, channel private offers up to a quarter of the commitment since June 2025. Every third party line you route through the marketplace, Databricks, Datadog, the security stack, is shortfall insurance you were already going to spend. Then list the marketplace candidates in the contract conversation itself, so nobody later argues about what qualifies.
Stack every discount layer deliberately. The private rate you negotiate and the committed use discounts you buy from the rate card are separate layers, and they compound: estates running both save roughly double either alone. Sign the commit, then still buy the CUDs, and re-base any commitment written before the July 2025 overhaul, because the numbers changed meaning. Then attack the quiet line: support.
It bills as a percentage calculated on list before your discounts, so ask for support on net, or a cap, or price the partner route, where partner led premium support runs at a flat 4 percent. On a large estate the support line alone can fund everything else you asked for. And take the support quote apart before you accept it: the tier you actually need, the response times you actually use, and whether a lower tier plus a partner escalation path covers the same risk for a third of the line.
Contain the AI lines. Seats scoped to named populations, never all employees by default. The pay as you go edition for the long tail instead of committed seats nobody opens. A measured pilot before any committed volume, because first proposals run 50 to 100 percent above real burn.
And two clauses that matter more every quarter: model flexibility, the right to route to successor and cheaper models at equal or better unit economics, because the price list changes faster than your contract. And spend caps on every agent workload from day one, so the meter can never outrun the budget while nobody is watching.
Write the exit while they still want you. Price holds for the term. A termination right for chronic SLA failure, because credits are the sole remedy on the standard paper. Wind down assistance and a data retrieval window if you leave.
All of it costs almost nothing at signature and almost everything after. The egress story helps you here: exit migration fees have been waived since 2024 and the EU Data Act is regulating switching costs away. The lock in argument is dying. Your leverage peaks the week before you sign.
Spend it on the exits. None of these clauses will ever feel urgent again after signature. That is exactly why they belong in this week, not next quarter.
One last point. At Redress Compliance we negotiate Google commits on pure contingency: 25 percent of what we save you. Nothing saved, nothing paid. Next episode: the contract terms that outlast every discount, from fee change rights to sovereignty.
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 Google negotiator