Just after midnight on 15 September 2023, the contracts between the United Auto Workers and the three Detroit carmakers expired, and the union did something it had never done: it struck all three at once. Not everywhere. One plant each: GM's Wentzville Assembly in Missouri, Ford's Michigan Assembly in Wayne, and the Stellantis Jeep complex in Toledo, Ohio. About 13,000 workers walked out, from a membership of 146,000 at the three companies.
The union called it a "stand-up strike", and it grew in steps. On 22 September, 38 parts distribution centres at GM and Stellantis joined; Ford was spared for having moved. On 29 September, Ford's Chicago Assembly and GM's Lansing Delta Township. On 6 October, nothing: GM had agreed to bring its battery plants under the national agreement, and union president Shawn Fain said a planned walkout at GM's Arlington plant was called off. On 11 October, a Wednesday evening, with no warning, 8,700 workers left Ford's Kentucky Truck Plant, Ford's most profitable factory. Then Stellantis' Sterling Heights plant on 23 October, Arlington on 24 October, GM's Spring Hill on 28 October. At the peak, about 46,000 workers were out.
Ford settled first, on 25 October. Stellantis followed on 28 October, GM on 30 October. The pattern: 25 per cent general wage increases over a contract running to 30 April 2028, 11 per cent of it on ratification, cost-of-living adjustments restored, three years instead of eight to reach the top rate, big raises for temporary workers, a $5,000 ratification bonus, and a contractual right to strike over plant closures. Members ratified in November: 69.3 per cent at Ford, 70 per cent at Stellantis, 54.7 per cent at GM.
Six weeks, three counterparts, one union. Four moves in that sequence are worth taking apart.
1. Escalate in steps, and keep the next step visible
The obvious way to strike three companies is to strike them. All 146,000 members, day one. The union did the opposite, and the reason is arithmetic. Its strike fund held about 825 million dollars and paid 500 dollars a week per striker. By one estimate reported at the time, everyone out at once would have drained it in about twelve weeks. Thirteen thousand out could be sustained for far longer, and nobody across the table knew how long.
The step-by-step design did three things at once. It kept the union's cost low. It kept the companies' cost uncertain, because every plant that was not yet struck was a plant that could be struck next week. And it kept the union's options open: every Friday it could escalate, hold, or reward, depending on what had happened at the table that week.
Notice that the escalation always came with a reason. Parts centres at GM and Stellantis, about 5,600 workers, because those two had rejected the union's job-security, profit-sharing and temp-conversion proposals while Ford had made what Fain called serious progress. Kentucky Truck because Ford, after two weeks of saying it would improve its offer, arrived at a meeting in Dearborn with the same offer as before. According to a Ford executive quoted by AP, Fain's reply was "You just lost Kentucky Truck Plant." The meeting lasted about fifteen minutes.
At your own table, this is the difference between "take it or leave it" and "here is what happens next if nothing changes". The second keeps you in the room. A big, single, final move spends all your leverage in one go and leaves you nothing to answer with if it does not work.
2. Three counterparts, one table: the first to settle sets the pattern
Bargaining with all three companies at once was the riskier half of the design and the half that paid off most. Each company could see what the others were being offered, and each could see what happened to those who moved and those who did not. Ford was spared in week two for progress. GM bought a week of calm on 6 October by conceding on battery plants. Stellantis, which had not moved, lost its most profitable plant on 23 October.
The rule that made it work was simple and public: whoever settles first gets the strike lifted and sets the terms for the other two. Ford's deal on 25 October became the template; AP described the Stellantis deal three days later as following the model Ford set, and GM's followed on 30 October. From the moment Ford signed, the other two were no longer negotiating against the union. They were negotiating against a document that already existed, with their own plants still shut.
When you negotiate with several counterparts at once, the first signature is your strongest argument with the rest.
You will not often have three counterparts at once. But you will have two suppliers quoting for the same scope, or two employers interested in the same candidate. The move is the same: let each know there are others, be visibly fair in how you treat them, and reward the one that moves first, out loud.
3. Public deadlines bind both sides
The union ran the strike on a clock everyone could read. Friday livestreams. Noon deadlines. "Serious progress by Friday or more plants go out." Public deadlines do two things to the other side. They put a price on waiting, so the comfortable option of doing nothing this week stops being free. And they bring in third parties: suppliers, dealers, local politicians, the press, all of whom now know when the next blow lands and start asking the company what it plans to do about it.
They also bind the side that sets them. On 22 September the union had to follow through, and did. A deadline you set and then let slide is worse than none, because the counterpart learns that your next one is also negotiable. Note too what the union did in week four: it broke its own rhythm. After three Fridays, the companies had started planning around Friday. Kentucky Truck went out on a Wednesday. The lesson is not "be unpredictable". It is that a deadline is a tool, and a tool the other side has fully priced in has stopped working.
At your own table: set deadlines you are certain you can enforce, say them once, in writing, and keep them. If you cannot enforce it, do not set it.
4. What the union did not get
The opening demands included a 46 per cent raise, a 32-hour week for 40 hours of pay, the return of defined-benefit pensions for workers hired after 2007, and retiree health care. The final contracts delivered none of those four. The raise came in at 25 per cent, which, with cost-of-living adjustments, AP put at about 33 per cent compounded for top-rate workers. The 32-hour week, the pensions and the retiree health care were dropped; 401(k) contributions were raised instead.
That is normal. An opening position is supposed to be above where you land, and the gap here was managed well: the big aspirational items were traded for things the companies could actually sign. But the gap had a cost, and you can read it in the ratification. At GM, 54.7 per cent voted yes. Ford's Kentucky Truck Plant, the one that had been pulled out as a surprise, voted the deal down by 54.5 per cent, and Toledo Jeep, one of the three original plants, rejected it as well. The people who carried the strike remembered what had been promised.
Two things to take from that. First, an ambitious opening works, but everyone behind you hears it as a goal, not as a tactic, and you will have to explain the distance later. Second, when you close, name what you did not get, before the other side or your own people do it for you.
At your own table
- Escalate in steps you can afford, each with a stated reason, and keep the next step visible.
- With several counterparts, say the rule out loud: the first to move gets the best treatment, and sets the terms for the rest.
- Set only deadlines you can enforce, and enforce them. Once the other side has priced one in, change the rhythm.
- Open high, but keep a list of what you traded away, and own it when you close.
The hard part is not the design. It is holding the line in the room when the counterpart says the offer is final and the fifteen minutes are up. On Pacct you can rehearse that moment against an AI counterpart that pushes back, and the debrief afterwards tells you whether you escalated, folded or stalled.
