Why Deals Die on Fridays

Executive Summary

• A diagnostic — the Tuesday Test — for telling apart the decisions you made from the decisions the deadline made for you
• Four familiar deal behaviors — the aggressive counter, the disappearance, the stalled committee, the over-promise — and the single condition underneath all four
• Why a competitive search reproduces, almost exactly, the two conditions research has identified as producing the largest measured stress response in people: you do not control the outcome, and your performance will be judged by name
• The operational finding, which is not about emotion — under deadline pressure, experienced practitioners switch from weighted scoring to fatal-flaw screening without noticing, applying a longlist tool at best and final
• Decision headroom: the same idea as circuit headroom, applied to the people and boards deciding, and the standing loads that narrow it
• Nine changes at the level of deal design, and why an EDO that answers predictably gets shortlisted over one that is occasionally brilliant

Opening: The Tuesday Test

Take the last deal you worked that mattered. Find the three decisions you would want back — the number you put on the table, the timeline you agreed to, the site you stopped defending, the email you sent at 5:40. Write down when each one was made.

Then run the test. For each decision, ask whether you would have made it the same way at nine o'clock on a Tuesday morning, with exactly the same information in front of you and no clock running.

A word on the title, because it is a claim and I want to be honest about how far it goes. I am not arguing anything about the day of the week. Friday is shorthand for the compressed window in front of a deadline, and deadlines in this business cluster: the board meets Tuesday, the fiscal year closes Wednesday, the company wants best and final by close of business. Read “Friday” as “hour sixty of a seventy-two-hour window” and the argument holds on whatever day it lands.

This is not a wellness paper. It is a paper about decision quality, which is measurable.

Section 1 — Four Ways a Deal Dies in the Last Week

Four behaviors. You have watched all four, and if you have been in this business ten years you have produced all four. None of these is a character defect, and diagnosing a colleague with one is a misuse of the model. Each is a reasonable output of a specific structural condition — an information gap, an unassigned decision right, an accountability structure that punishes candor.

Under pressure, people do one of four things: push harder, disengage, stall, or give ground. In a deal they arrive with names we already use.

The Aggressive Counter

What it looks like. An incentive counter materially above what the project economics support, produced fast. A routine diligence question reframed as an attack. The escalation email over the consultant's head to somebody's SVP.

What it costs. Terms that fail the clawback review in year three, and a reputation among consultants for being difficult to run a process with.

The structural gap behind it. The threat is real and the information needed to size a proportionate response does not exist. You cannot calibrate against a competitive set you are not allowed to see, so you calibrate against the fear. This is what a rational actor does with a genuine threat and no scoreboard.

The Disappearance

What it looks like. Calls that stop being returned after a difficult RFI. A site quietly withdrawn rather than explained. A company that ghosts three finalists rather than issuing declines.

What it costs. The most valuable artifact in this business, which is the debrief. When nobody says why, nobody learns anything, and the next search reproduces the last one exactly.

The structural gap behind it. No protocol for delivering bad news, plus an accurate read that saying “we cannot serve that load” carries career consequences. An organization that punishes candor should expect silence. It is getting what it built.

The Stalled Committee

What it looks like. A time-sensitive item tabled twice. A board that will not vote on an option agreement before the deadline. An RFI response that sits at ninety percent complete for nine days.

What it costs. This is the most expensive failure mode precisely because it is invisible. Nothing is decided, nobody is blamed, the project goes elsewhere, and the minutes record that discussion continued.

The structural gap behind it. Unassigned decision rights combined with asymmetric accountability. The downside of a wrong yes lands on a named person. The downside of no decision is distributed across a body. Stalling is the correct play under those rules, which means the rules are the problem.

The Over-Promise

What it looks like. An incentive package expanded to hold a project that was leaving regardless. A diligence window compressed on request. An environmental condition waved past. An infrastructure commitment made ahead of the letter that would have supported it.

What it costs. This one is generational. The commitments a community lives with for twenty years are disproportionately made in the fifteen minutes after a company said it was walking.

The structural gap behind it. This is the hardest of the four to see from the inside, because in the moment it is indistinguishable from responsiveness and good client service. The tell is informational. Ask whether anything new actually arrived before the position moved.

Section 2 — The Search Is Built Like a Stress Test

What Actually Produces a Stress Response, and What Doesn't

Not everything that feels hard produces a measurable physical response. The cleanest synthesis is Dickerson and Kemeny's 2004 meta-analysis in Psychological Bulletin, which pooled 208 laboratory studies and asked which task features actually moved cortisol. The answer was narrow and specific. Two conditions did the work: the outcome was not the subject's to determine, and the subject's performance would be judged by other people. Tasks carrying both produced large cortisol changes and, notably, the longest recovery times. Tasks carrying neither produced changes statistically indistinguishable from zero.

The honest caveat is that a laboratory is not a deal room. Treat this as a well-supported model of a mechanism rather than a measurement of your Tuesday. The mechanism is the useful part, because it means pressure is not a general property of difficulty. Hard work that you control and that nobody is grading does not do this. Two specific ingredients do.

What a Competitive Search Is Made Of

• The outcome is determined elsewhere. You are scored on a weighted matrix you have not seen, by a party you frequently cannot identify, against a competitive set you are not permitted to know. If you lose, you often never learn why.
• Performance is evaluated publicly and by name. The EDO director answers to a board, a mayor, and eventually a newspaper. The corporate real estate lead carries a recommendation into a capital committee. The consultant's judgment is graded by a client deciding whether to hire them again.
• The clock belongs to someone else. The timeline is set by another party, is usually unexplained, and moves.

The Chemistry Outlasts the Call

Two chemicals do the work, on very different clocks. Adrenaline is immediate — within seconds of the call turning, cleared within about fifteen minutes. Cortisol is the slow one. It peaks roughly ten to thirty minutes after the call has ended and commonly takes an hour or more to return to baseline, with the slowest recoveries following precisely the tasks that were both outside the subject's control and publicly judged.

Read that against a deal calendar. A call goes badly at 3:45. The revised position is drafted at 4:10 and sent at 4:30. Every one of those decisions is being made on a body that has not cleared, by someone who experiences it as simply having made up their mind.

Section 3 — What Degrades Is the Screening Method

The industry's existing answer to all of this is professionalism. Stay calm, don't show your hand, don't let them see you sweat. That is a rule about your face. It has no particular effect on what your judgment does, and the evidence for the two coming apart is better than the evidence for them traveling together.

Research on decision-making under time pressure describes three adaptations, escalating in consequence.

• Acceleration. The same process, run faster. Mild, and mostly harmless.
• Filtration. You work from a subset of what is actually in front of you, weighted toward avoiding the worst outcome rather than achieving the best one.
• Method change. You stop weighing factors against each other and start eliminating on one factor alone.

That third one has a name in our business, and it should stop an economic developer cold. Weighing attributes against each other — so that a strong power position can carry a mediocre labor position — is what a weighted scoring matrix does. Eliminating on a single attribute is what a fatal-flaw screen does. Both are legitimate tools. They belong at different stages of a search.

A fatal-flaw screen is how you get from two hundred sites to twelve. Run in the final week, it is how a site that would have won dies over a fixable discharge question, and how a community still genuinely in the running comes off the list because one number in one column looked bad at four o'clock on a Thursday.

Section 4 — Decision Headroom

On the utility side, headroom is the load a circuit can carry before service degrades. Nobody plans a hundred-megawatt load onto a circuit with forty megawatts of headroom and calls the resulting failure a character flaw of the circuit. Decision headroom is the same accounting applied to the people and boards deciding: the pressure they can absorb before decision quality degrades. Everyone has some. Nobody has unlimited.

It is a model, not a measurement. As a working definition: headroom is the range of pressure inside which you can still do arithmetic, hold two contradictory facts at the same time, and say the words “I don't know.” Above it you get speed and certainty you have not earned, which is the aggressive counter and the disappearance. Below it you get the stall, which is the committee and the over-promise.

What Narrows It

Headroom narrows with sleep debt, with the fourth consecutive week of travel, with a personal crisis running alongside the deal. It narrows most reliably with standing load that never comes off — and this industry has a signature one.

What the Codename Costs

Consider Project Cardinal. For fourteen months the EDO director cannot tell their staff, their board chair, their spouse, or the county engineer what the project is. A public-records request is pending. A reporter has a partial. The obligation is real and I am not arguing against confidentiality — it is a legitimate condition of the work and the reasons for it are good ones.

It is also, mechanically, a textbook standing load: constant vigilance, no debrief, nobody to talk to, and a known and serious consequence for error. A director carrying three confidential projects has less absorptive capacity for the fourth thing that goes wrong, and the fourth thing is usually the Friday counter. Three responses help, none of them expensive.

• Tier the confidentiality deliberately, in writing, at project start. Default-maximal secrecy is a choice made by not choosing. Most projects have an inner ring that genuinely needs full knowledge and a wider ring that needs a sanitized version, and writing that down converts a diffuse burden into a defined one.
• Name one person inside the wall. A board chair, a general counsel, a state partner — somebody with whom the project can actually be discussed. Isolation is the active ingredient in this particular load, and a single legitimate interlocutor removes most of it.
• Put a sunset on the confidentiality and hold a debrief after it. NDAs routinely outlive the projects they cover and nobody rereads them, so the debrief never happens and the organization never learns. Diary the date at execution.

Section 5 — Designing the Deal Around It

Everything above points at the same intervention, and it is not an individual one. Telling an EDO director on hour sixty to breathe is not a strategy and is faintly insulting. The intervention sits at the level of deal design, and it works for a specific reason: not controlling the outcome and not knowing what comes next are the two active ingredients, and both are things a process can be built to reduce. Nine changes, in three groups.

Decide While You Still Have Headroom

Pre-commit the limits in writing. A board-approved incentive ceiling, a wage and capital floor, and a written statement of the conditions under which the community declines — adopted before a live project exists. A walk-away drafted on a quiet Thursday in February is a different document from one drafted at six o'clock on the day of.

Assign decision rights at kickoff. Who can say yes to what, up to what number, without convening anyone. Most stalling is an unassigned-authority problem wearing the costume of prudence.

Require the underlying document before the commitment. The load-serving letter before the megawatt figure. The title work before the assemblage promise. A commitment that outruns its supporting document is usually an over-promise in professional clothing.

Change the Clock

Treat a deadline you did not set as information, not as physics. Ask what it attaches to: a board date, a fiscal close, a lease expiration, a financing condition — or nothing. A meaningful share of “best and final by Friday” resolves to the last category, and the ones attached to a real constraint survive the question without difficulty.

Never accept a materially new term inside the final seventy-two hours without a stated pause. “We'll come back to you Monday morning” costs nothing when the counterparty's deadline is genuine and tells you a great deal when it is not.

Make the Process Legible

Publish the sequence. A one-page schedule of what happens when, who decides, and when an answer is expected removes most of the uncertainty from the other side of the table at essentially no cost.

Issue declines with a reason. The debrief is the highest-return, lowest-cost practice available to anyone in this industry, and the disappearance is the entire reason it does not happen. Write the paragraph. Send it.

Log the decision at the moment. One line, in real time: what we decided, what we knew when we decided it, and what would change it. It costs thirty seconds and it is the only way to run the Tuesday Test later with any honesty, because memory reliably reconstructs decisions as more deliberate than they were.

The Same Work Is a Competitive Position

Read that list again from the other side of the table. Every item on it removes uncertainty for your counterparty, and uncertainty is one of the two ingredients doing the damage. Which explains something the industry already observes and explains badly: the EDO that answers within twenty-four hours, every time, gets included in searches it does not technically qualify for.

Section 6 — Three Scenarios

The three scenarios below are constructed models, not client engagements. They are assembled from the patterns described in this paper, with figures set at realistic values for organizations of this size. Every number in them is hypothetical. The shapes are not — I have watched each of these dynamics play out, in some form, on real deals.

Scenario 1: The Counter Nobody Could Explain

Profile: County EDO, six staff. Late-stage competition for a $180 million manufacturing project, two finalists, decision expected inside a week. Board-approved incentive framework with no stated ceiling.

Discovery

• The company's consultant called at four o'clock on a Thursday to say the competing location had materially improved its position, and asked for a response by close of business Friday. No detail on the competing offer was provided, and none was requested.
• The revised package, assembled that evening and approved by three board members by phone the next morning, extended the abatement term by six years and added an infrastructure commitment that had not been costed.
• The analysis supporting the original offer was not rerun. No new information about the project, the competitor, or the company's economics had arrived between the original offer and the revision.
• The organization had no written walk-away position and had never discussed one.

The project landed. A post-award review two years later found the package exceeded the organization's own internal return threshold by a wide margin, with the uncosted infrastructure commitment absorbed by the county. The board subsequently adopted a written ceiling, a required twenty-four-hour pause before any material revision to a standing offer, and a rule that no revision proceeds without a one-paragraph statement of the new information that prompted it.

Key Lesson: The cheapest control in incentive negotiation is a written requirement to name the new information before the position moves. Most aggressive counters and most over-promises cannot survive that one sentence.

Scenario 2: The Silence After the Site Visit

Profile: Regional EDO, four counties, nine staff. Competes for logistics and light manufacturing projects and participates in eight to twelve RFIs a year.

Discovery

• Over three years the organization had advanced to site visit on five projects and lost all five. It had received a substantive debrief on one.
• In four of the five, contact from the organization's side dropped off sharply after a diligence request it could not fully answer — in two cases a rail question, in two a wastewater discharge question.
• Staff described the pattern consistently once someone asked: the request arrived, the honest answer was bad, the reply was deferred while a better answer was sought, the better answer never came, and the reply was never sent.
• The organization's own decline practice mirrored it exactly. It had never sent a written decline to a developer or a landowner.

The organization adopted a forty-eight-hour response standard that explicitly permitted incomplete answers, in a required format: what we know, what we do not, and when we will know it. Over the following eighteen months it received substantive debriefs on three of four losses. Two identified the same discharge constraint, which became the basis of a capital request that had never previously had evidence behind it.

Key Lesson: A fast incomplete answer outperforms a slow complete one in nearly every case. The disappearance is why a fact that obvious is so rarely operationalized.

Scenario 3: The Committee That Didn't Vote

Profile: City redevelopment commission, seven members, meets monthly. Considering an option agreement on a 140-acre assemblage with an eleven-week window before the seller's stated expiration.

Discovery

• The item appeared on three consecutive agendas and was tabled each time, on requests for additional information that staff had already supplied in the packet.
• No individual member had authority to bind the commission to anything, and no subcommittee had been delegated authority to act between meetings.
• Two members had taken public criticism eighteen months earlier over an unrelated land transaction. Neither raised it in connection with this item.
• The seller did not extend. The assemblage was bought by a private party and is currently held with no announced use.

The commission adopted a standing rule that any item carrying an external expiration date must be voted at the meeting where it first appears, with the available options being approve, decline, or delegate to a named subcommittee with a stated authority limit — but not table. Both changes cost nothing.

Key Lesson: The stalled committee is an accountability design problem rather than a competence problem. Removing “table” as an option on time-limited items converts an invisible decision into a visible one, which is the whole objective.

Conclusion

Site selection is an information business that has spent twenty years getting considerably better at information and approximately no time at all on the conditions under which that information gets used. We have improved the data, the mapping, the utility analysis, and the diligence. We have not touched the part where a person who slept five hours makes a twenty-year commitment at 5:40 on a Friday because somebody they have never met used the word “final.”

Run the Tuesday Test on your last deal. Then run it on the one you are in now, while there is still time for the answer to matter.