It is the second Tuesday of the quarter, and you have forty minutes on the operating review agenda, slot six of nine, right after the CFO's cash update and right before someone from product talks about a launch that is slipping. The localization-AI program you fought to fund eighteen months ago is now real: the engines are wired in, the linguists are post-editing MT-first files, the quality gate is running, and the numbers exist. Your job today is not to sell the program. That fight is over. Your job is to report it, honestly, to a room of executives who approved it on a promise and now want to know whether the promise held. And here is the quiet trap waiting for you in slot six: the easiest, most flattering story you can tell is the speed story. Throughput is up ninety percent. Cost-per-word is down a third. You could put those two numbers on a slide, take the nods, and sit down in six minutes. It would feel like a win. It would also be the single most dangerous thing you could do to your own program, because the day a Critical error ships in a drug label or an indemnity clause, that six-minute speed-only readout becomes the evidence that you knew about the throughput and never mentioned the risk. This lesson is about the other kind of readout: the one that pairs the speed win with the quality-risk posture, on the same page, in the same breath, so that leadership sees both the return you generated and the exposure you are actively holding down. We are going to build the dual-axis leadership report from the ground up, decide what executives actually want to see versus what you want to show them, learn to present the quality record as proof rather than reassurance, and walk a full worked readout end to end. Get this right and you do not just survive slot six. You become the operational leader the room trusts, because you are the one who reported the risk before it reported itself.

The Two Axes, and Why You Report Both or Neither

Start with the shape of the story, because the shape is the strategy and everything else is detail hung on it. A leadership report on a localization-AI program is a dual-axis report, meaning it presents two distinct dimensions of the program's performance side by side rather than collapsing them into one headline. The first axis is the ROI axis: throughput, cost-per-word, capacity, time-to-market, the speed and money story. The second axis is the quality-risk axis: the critical-error rate, the terminology conformance, the gate's catch record, the exposure the program is holding down. Let me define the load-bearing terms before we build on them, because a leadership report lives or dies on whether the room understands the words. ROI (return on investment) is the net benefit the program produces relative to what it costs, and in a reporting context it means the realized return you can show against the promise you made. Quality risk is the probability and consequence of a defective translation reaching a customer, a regulator, or a court, and in localization the defect that matters most is not a typo but a fluent, confident mistranslation of something that carries liability. The critical-error rate is the frequency, per unit of volume, at which Critical errors (mistranslations severe enough to cause safety, legal, or financial harm) appear in evaluated output. And a dual-axis story is the discipline of never letting one of those axes be seen without the other.

The reason you report both or neither is not a matter of thoroughness or good manners. It is a matter of what a single axis does to the reader's decisions. Report only the ROI axis and you have taught the room that this program is a speed-and-cost machine, which means the next time budgets tighten, they will optimize it the way you optimize a speed-and-cost machine: cut the slowest, most expensive part, which is the human quality layer. You will have handed leadership the argument for gutting the exact control that keeps the program safe, and you will have done it with your own slide. Report only the quality-risk axis and you look like a cost center pleading for indulgence, a team that spends money on governance and asks the room to trust that it matters, with no return attached to justify the spend. Neither axis survives alone. The ROI axis without the quality axis is a liability disguised as a win. The quality axis without the ROI axis is a cost disguised as a virtue. Together, and only together, they tell the true story: this program buys you speed and volume that you can prove, at a level of catastrophic-error risk that you can also prove is being held down, and the two facts are not in tension, they are the same well-run operation seen from two sides.

Report both axes or neither. The ROI axis alone teaches leadership to optimize your program by cutting the quality layer. The quality axis alone makes you a cost center pleading for trust. Paired, they are the same well-run operation seen from two sides, and only the pair is true.

Why the Speed-Only Narrative Is a Time Bomb You Set Yourself

It is worth sitting with exactly how the speed-only readout detonates, because the pull toward it is strong and you need to feel the cost before you will resist it. Imagine you gave the flattering six-minute version for four consecutive quarters. Throughput up, cost down, nods all around, and not one slide mentioning error rates or the gate. Now, in the fifth quarter, a post-edited file ships with a flipped dosage that read perfectly and meant the opposite of the source, and it reaches a patient information leaflet in-market before anyone catches it. The incident review convenes. Someone pulls your four quarters of readouts, and what they find is a program that reported, every single quarter, only how fast and cheap it was, and never once reported the thing that just went wrong. You are no longer the leader who ran a program that had an incident, which every program eventually does. You are the leader who buried the risk while celebrating the speed, and that is a different and far worse position, because it converts a quality incident into a credibility incident, and credibility incidents end careers and programs together.

The speed-only narrative is a time bomb because it works right up until the moment it does the maximum possible damage. The quarters where nothing ships wrong, it earns you easy nods and builds a false record. The one quarter something does ship wrong, that same false record becomes the proof that you were not watching, or worse, that you were watching and chose not to say. The dual-axis report is the defusing. When you have reported the critical-error rate every quarter, including the quarters it was zero, the quarter an error slips through is a data point on a curve you have been transparently tracking all along, not a revelation you concealed. The difference between "we caught our first shipped Critical this quarter, here is the root cause and the fix, and here is the eleven Criticals the gate caught before delivery" and a stunned silence in an incident review is the difference between a program that survives its first bad quarter and one that does not. You report the quality axis in the good quarters precisely so you have the standing to report it in the bad one.

What Executives Actually Want to See (Not What You Want to Show)

Now turn the report around and look at it from the reader's chair, because the most common failure of a technical leader reporting to executives is reporting what is interesting to build rather than what is useful to decide. You are proud of the pipeline. You want to show the retrieval grounding, the term-enforcement logic, the clever routing. The executive does not want any of it, and every minute you spend on it is a minute you are not answering the three questions the room actually convened to answer. Executives read an operational report to make three decisions, and everything on your slide should serve one of them. First: is this working, against what we approved? They funded a promise with numbers in it, and the first thing they want is the realized number next to the promised number. Second: what is it exposing us to? They are accountable for the enterprise's risk, and a program that touches regulated, customer-facing, or contractual content is a risk surface they are personally answerable for. Third: what do you need from us? A report that does not end in a decision, a resource ask, or an explicit "no action needed" is a report that wasted their time.

Notice what is not on that list. Executives do not want the mechanism. They do not want to understand how MQM severity scoring works or what retrieval-augmented grounding is; they want to know that quality is measured rigorously and the number is good or bad. They do not want a tour of the tooling. They do not want your effort, the heroic quarter your team had, the late nights integrating the engine. Effort is not an outcome, and a room of executives has a finely tuned allergy to being asked to reward effort in place of results. They want outcomes stated in their own currency: money, risk, time, and the confidence interval around each. The discipline of an executive report is subtraction. You know forty things about your program. Thirty-six of them are irrelevant to the three decisions, and the skill is cutting them without grieving them, so that the four that matter land clean.

The Currency of the Room Is Not the Currency of the Shop

There is a translation happening in every good leadership report, and it is the same translation you spent this whole program learning to respect. In the shop, your currency is words-per-day, fuzzy-match rates, MQM error points, terminology query counts, segment-level provenance. In the room, the currency is cost, risk exposure, revenue timing, and headcount productivity. A leader who reports in shop currency to a room that thinks in enterprise currency fails exactly the way a fluent-but-wrong translation fails: the words are all correct and the meaning does not transfer. "Our MQM score improved from 92 to 96 on a hundred-point scale" is a true sentence that means nothing to a CFO, because she has no calibration for whether 96 is good, no sense of what a four-point gain is worth, and no way to connect it to a decision. Translate it: "the rate at which serious errors reach a customer fell by more than half, and here is what one such error would have cost us." Same fact, delivered in the currency of the room, now actionable.

This is not dumbing down; it is translating up. The executive is not less sophisticated than you, she is sophisticated about a different domain, and the entire value you add in slot six is being the bilingual party who carries the meaning across. The terminology conformance rate (the percentage of approved terms rendered correctly against the client termbase) is a linguist's metric. In the room, it becomes "brand and regulatory consistency across every market, measured, currently at 98.5%, which is what protects us from the kind of inconsistency that triggers a regulator's question." You do not hide the metric; you carry it across the bridge to the thing the executive is accountable for. Every number in a leadership report should arrive already translated into a consequence the reader owns.

The report is a translation job. Words-per-day, MQM points, and conformance rates are shop currency; the room thinks in cost, risk, revenue timing, and productivity. Carry every metric across the bridge to a consequence the reader is personally accountable for, or it does not transfer.

Building the Dual-Axis Report, Panel by Panel

Now we build the artifact. A leadership report on a localization-AI program has four panels, and they go in this order for a reason: the order is the argument. Panel one establishes that the program is working on the axis they care most about protecting; panel two proves you are protecting it; panel three reconciles the promise to the reality; panel four tells them what to do. Build them out of order and you undermine your own case. Lead with cost savings before you have established the quality posture and you have opened with the speed-only narrative you are trying to escape. The sequence is a discipline, not a preference.

Panel One: The ROI Axis, Realized Against Promised

The first panel is the return, and the single discipline that makes it credible is that every number is stated realized-against-promised, never realized alone. A realized number alone is a boast; a realized number against the promise you made is an account. "Throughput reached 4,600 words per linguist per day" is a boast the room cannot evaluate. "Throughput reached 4,600 against a conservative plan of 3,500 and an upside of 5,000, so we are running ahead of base and below ceiling" is an account, and it does three things at once: it shows the gain, it reminds the room you promised conservatively, and it signals there is still headroom. Report the throughput, the cost-per-word, the total production saving in dollars, the capacity gain (how much additional volume or how many additional languages the same headcount now serves), and, if the business tracks it, the time-to-market acceleration in revenue terms. Each one carries its promised figure beside it. The panel's job is to let a skeptical reader confirm, in thirty seconds, that the ROI axis is delivering roughly what was sold, and to catch, honestly, any line that is underdelivering, because a report that shows one line behind plan and explains it is trusted on the lines that are ahead.

Panel Two: The Quality-Risk Axis, as a Posture Not a Promise

The second panel is the one that separates a leadership report from a sales pitch, and it is the panel most localization leaders either skip or soften into meaninglessness. It reports the quality-risk posture, and the word posture is deliberate: you are not promising perfection, you are reporting the stance the program holds against catastrophic error. The centerpiece is the critical-error rate, and you report it two ways that matter enormously. First, the escaped critical-error rate: Criticals that reached delivery, per unit of volume, ideally zero and reported as zero when it is zero, because a reported zero is a claim you are staking your credibility on and it carries weight. Second, and this is the move most reports miss, the caught critical-error rate: the Criticals the gate stopped before delivery. That second number is not an admission that your engine is dangerous; it is the proof that your control is working. Every Critical the gate caught is a shipped incident prevented, a lawsuit or recall that did not happen, and reporting the catch count turns the quality gate from an invisible cost into a visible, quantified defense. Alongside the two error rates, report the terminology conformance rate, the volume of content that went through the severity-scored gate versus bypassed it, and the share of high-liability content that was correctly routed to full human or full post-editing rather than the cheap workflow. Together these say: here is how much we are protecting, here is how well the protection is working, here is what got through.

Report the caught critical-error rate, not just the escaped one. Every Critical the gate stops before delivery is a recall or a lawsuit that did not happen. The catch count converts your quality gate from an invisible cost line into a visible, quantified defense the room can value.

Panel Three: The Reconciliation to the Original Business Case

The third panel closes the loop that the business case opened, and it is the panel that earns you the right to be believed next time you ask for money. When the program was funded, you made specific promises: a throughput range, a cost saving, a risk-reduction benefit modeled as probability times cost avoided, a J-curve with a dated crossover where results dip before they rise. The reconciliation panel puts the promise and the reality side by side and tells the truth about the gap in both directions. Where you beat the plan, say so plainly and quietly, because overclaiming a win is how you lose the room's trust on the next line. Where you missed, name it, explain it, and state what you are doing about it. If the J-curve crossover came a quarter later than modeled because the ramp took longer, that is not a failure to hide, it is a modeling assumption that resolved, and the leader who reports "we crossed into net-positive in Q3 rather than the Q2 we projected, because linguist ramp to full throughput took a quarter longer than the benchmark suggested" is more trusted than the leader whose every projection magically came true. Executives have funded enough programs to know that none of them land exactly on plan, and the one that claims it did is the one they stop believing.

Panel Four: The Decision, the Ask, or the Explicit All-Clear

The fourth panel is the one that makes the meeting worth holding, and it is the one most technical reports forget entirely, ending instead on a summary slide that asks the room for nothing and therefore gives them no reason to have convened. Every leadership report must end in one of three things. A decision you need them to make: approve the scale-up to three more languages, sign off on the engine change, ratify the revised risk appetite. A resource ask, stated with what it buys: two more evaluators to hold the critical-error rate as volume doubles, and here is the expected effect on the escaped-Critical number if you fund it and if you do not. Or an explicit all-clear: no action needed, the program is on track, this is a status confirmation, and you have just given the room the gift of a decision they can make in zero seconds, which they will remember fondly. What you must never do is end ambiguously, because an ambiguous ending forces the room to guess what you want, and a room that has to guess defaults to the cheapest guess, which is "do nothing and maybe trim the budget." Name the ask, or name the all-clear, but name something.

Presenting the Quality Record as Proof, Not Reassurance

Here is the deepest move in the entire lesson, and it is the one that turns a good report into an unassailable one. The quality-risk axis has a credibility problem that the ROI axis does not: cost savings are self-evidently real because they show up in a budget line the CFO can see, but "our quality is good" is a claim, and claims from the team being evaluated are the weakest possible evidence. If your quality panel amounts to "trust us, the quality is high," you have offered reassurance, and reassurance is exactly what leadership has learned to discount, because every team reassures. The move is to replace reassurance with proof, and the proof already exists in your pipeline. It is the quality record: the per-segment, auditable trail your program produces as a matter of course, capturing the MT source, the post-edit, the terminology decisions, and the severity-scored evaluation for every file that shipped. That record is not a byproduct. In a leadership report, it is the single most powerful asset you have, because it converts your quality claim from an assertion into a verifiable fact.

The distinction between reassurance and proof is the distinction between "our critical-error rate is low" and "here is the severity-scored evaluation record for the quarter's regulated-content volume, showing zero escaped Criticals across 1.4 million words, eleven Criticals caught and logged at the gate with their root causes, and 98.5% terminology conformance, and it is the same record we would hand a client auditor or an ISO certifier tomorrow without changing a line." The first is a number the room can doubt. The second is an artifact the room can, in principle, verify, and the mere fact that it is verifiable is what makes it believed. You are not asking them to trust you; you are showing them the thing an independent auditor would check, and telling them it would survive the check. This is the same logic that runs through the entire program: a defensible quality record is what turns "looks fine to me" into proof, and at the leadership level, that same record is what turns "the quality is good" into a fact leadership can put their name behind.

Reassurance is a claim from the team being evaluated, and leadership discounts it by default. Proof is the auditable quality record you already produce. Do not tell the room the quality is good; show them the artifact an ISO certifier or a client auditor would inspect, and tell them it would pass unchanged.

Tying the Record to the Standards Leadership Half-Remembers

There is a specific and underused lever in presenting the quality record, which is that it lets you cash in the standards without lecturing on them. Leadership does not want an explanation of ISO 18587 (the post-editing standard, whose revision expands its scope to cover AI and LLM output and requires the post-editor to hold full professional-translator competence) or ISO 5060 (the 2024 standard that formalizes the Critical, Major, and Minor error severities your gate uses to score output). They will glaze over the moment you explain the mechanism. But they understand the word conformant, because conformance to a recognized standard is enterprise currency: it is what lets legal sleep, what a client's procurement team asks for, what a regulator respects. So you do not teach the standard, you cash it: "the quality record is maintained to be conformant with ISO 18587 and scored against the ISO 5060 severity model, which means if a client's auditor or a certification body asked to see our quality evidence, we would hand them this record as-is." One sentence. It tells the room the program is not just internally disciplined but externally defensible, and it does so in a currency they already value without a single minute spent on the mechanics. The standards are not the subject of the report; they are a credential you spend in one sentence.

The Worked Leadership Readout, End to End

Now assemble the whole thing on one concrete operation, so you see the machine turn. The setup: a mid-size operation, roughly 10 million source words a year, that stood up a governed MT-first program eighteen months ago. You have the operating-review slot. Here is the readout, panel by panel, in the currency of the room, with the discipline of everything above baked in. The numbers are illustrative and round for clarity; the structure is what transfers to your own operation.

Readout, Panel One: The ROI Axis

"The program is delivering on the return we promised, running ahead of our conservative base and inside our ceiling. Throughput on eligible content reached 4,600 words per linguist per day, against the 3,500 we modeled as base and the 5,000 we set as upside, so we are running hot but not at the optimistic edge. Production cost on the eligible 7 million words fell from the human baseline to full post-editing rates, delivering a realized production saving of 540,000 dollars this year, within a whisker of the 560,000 we projected. That freed capacity let us localize into two additional markets with no new headcount, which the business values at roughly 380,000 dollars of incremental revenue, and pulled our average time-to-market for a new-market launch forward by about six weeks. One line ran behind: our light-post-editing throughput on marketing content lagged the plan because the content was messier than sampled, and we are re-sampling before we extend that workflow. Net, the ROI axis is delivering, and I will show you exactly what it is costing us on the risk side to deliver it safely."

Readout, Panel Two: The Quality-Risk Axis

"Here is the posture that speed sits on top of, because the speed is only worth having if this holds. Escaped critical errors, meaning Criticals that reached delivery, this quarter: zero, across 1.4 million words of regulated and high-liability content. That zero is a number I am staking on our evaluation record, not a hope. Caught critical errors, meaning Criticals the gate stopped before they shipped: eleven, each logged with its root cause, and every one of those eleven is a shipped incident that did not happen, a flipped dosage or an inverted obligation that our control caught and raw machine output would have delivered. Terminology conformance across all markets held at 98.5%, which is the brand and regulatory consistency that keeps us out of a certain class of client and regulator questions. And the routing discipline held: 100% of content we classified as high-liability went to full human or full post-editing, none of it slipped into the cheap workflow. The gate is doing exactly the job we funded it to do, and the eleven catches are the visible proof of the risk we are holding down."

Readout, Panel Three: The Reconciliation

"Against the business case you approved: we promised a J-curve crossover into net-positive by Q2 of this year, and we actually crossed in Q3, one quarter later, because linguist ramp to full throughput took a quarter longer than the industry benchmark we modeled on. That is the one place reality ran behind the plan, and I would rather you hear it from me than find it in the numbers. Everything else landed inside the modeled ranges: the production saving, the capacity gain, and the risk-reduction benefit we modeled as expected liability avoided. On that risk-reduction line specifically, we modeled the governed program as lowering our expected annual liability from a shipped regulated-content Critical from roughly 300,000 dollars to roughly 75,000, and the eleven caught Criticals this quarter are the live evidence that the mechanism producing that reduction is actually running, not just assumed. The case is tracking, one quarter behind on crossover, and honest about it."

Readout, Panel Four: The Ask

"One decision, and here is what it buys. We want to double regulated-content volume next year as two more product lines go multilingual. At current evaluator headcount, doubling that volume would stretch the gate, and my honest projection is that the escaped-Critical rate, which is zero today, would come under pressure by the second half. I am asking for two additional qualified evaluators. Funded, we hold the escaped-Critical rate at or near zero through the volume increase, and I will report that number to you every quarter as the proof. Unfunded, we still ship the volume, but I am telling you now, on the record, that I cannot promise the same posture, and this readout is where I said so. That is the decision: two evaluators to hold the quality posture through the growth, or accept a modeled increase in escaped-Critical risk. Everything else in the program is on track and needs nothing from you."

Read that readout back and notice what it does. It never once asks the room to trust the team; it shows realized numbers against promised ones, a quality posture backed by a catch count and an auditable record, an honest miss on the crossover date, and an ask framed as a decision with both outcomes named. It leads with the return but never lets the return stand alone. It reports the zero escaped Criticals as a staked claim and the eleven caught ones as visible proof. And it ends by putting the quality posture, not the cost saving, at the center of the resource ask, which tells the room that the leader running this program understands that the speed is only ever worth having if the risk stays held down. That is the dual-axis story delivered. That is slot six survived, and more than survived: turned into the reason they trust you with the next program.

The Cadence and the Standing Record Behind the Readout

One layer deeper, because a single good readout is a performance and what you actually want is a system. The dual-axis report should not be a thing you assemble in a panic the night before the operating review; it should be the surfacing of a standing record you maintain continuously, so that the quarterly readout is a summary of numbers that already exist rather than a construction project. Instrument the pipeline so that throughput, cost-per-word, the escaped and caught critical-error rates, and terminology conformance are captured automatically as the work flows, and so that the quality record accretes segment by segment as files ship. When that instrumentation is in place, the readout writes itself, because every number on it is a rollup of a metric you have been tracking all quarter, and, crucially, you can answer the follow-up question. Executives test a report by pushing on one number, and the leader who can drill from "eleven caught Criticals" to "here are the eleven, here is the file each was in, here is the root cause of each" is the leader whose report is believed, while the one who can only repeat the summary number is the one whose report is doubted. The standing record is what lets you drill.

Cadence matters as much as content. Report the same axes, in the same shape, every period, because the value of a metric to an executive is largely in its trend, and a trend requires consistency. If you report throughput and cost this quarter, critical-error rate next quarter, and conformance the quarter after, you have given the room three disconnected snapshots and no trajectory. Report all of it, every quarter, in the same order, and the numbers start to tell a story over time: the throughput climbing as the J-curve resolves, the escaped-Critical rate holding at zero quarter after quarter, the caught-Critical count as a running tally of incidents prevented. That consistency is also your protection, because it is what makes the bad quarter legible. When you have reported the escaped-Critical rate as zero for five straight quarters and it ticks to one in the sixth, the room reads it correctly, as a rare event on a well-tracked curve, precisely because you built the curve. The cadence is not bureaucracy. It is the mechanism that makes your good record protective and your bad quarter survivable, and it is the difference between a leader who reports and a leader who is trusted.

A great readout is the surfacing of a standing record, not a construction project the night before. Instrument the pipeline so the numbers already exist, report the same axes in the same shape every period, and the trend, not the snapshot, becomes the story that makes your good record protective and your bad quarter survivable.

Key Takeaways

• A leadership report on a localization-AI program is a dual-axis report: the ROI axis (throughput, cost-per-word, capacity, time-to-market) and the quality-risk axis (escaped and caught critical-error rate, terminology conformance, routing discipline). You report both or neither, because the ROI axis alone teaches leadership to optimize the program by cutting the quality layer, and the quality axis alone makes you a cost center pleading for trust.
• The speed-only narrative is a time bomb you set yourself: it earns easy nods every quarter nothing ships wrong, and the one quarter a Critical escapes, your record of reporting only speed becomes the proof you buried the risk, converting a quality incident into a career-ending credibility incident. You report the quality axis in the good quarters to earn the standing to report it in the bad one.
• Executives read the report to make exactly three decisions: is it working against what we approved, what is it exposing us to, and what do you need from us. They do not want the mechanism, the tooling tour, or your team's effort; they want outcomes in enterprise currency (cost, risk, revenue timing, productivity), which means every shop-currency metric (MQM points, conformance rate, words-per-day) must be carried across the bridge to a consequence the reader is personally accountable for.
• Build the report in four panels, in order, because the order is the argument: panel one is the ROI axis stated realized-against-promised, never realized alone; panel two is the quality-risk posture, centered on both the escaped critical-error rate (ideally a staked zero) and the caught rate (the gate's catches, each a prevented incident); panel three reconciles promise to reality and names the misses; panel four ends in a decision, a resource ask with both outcomes named, or an explicit all-clear.
• Report the caught critical-error rate, not just the escaped one. Every Critical the gate stopped before delivery is a recall or lawsuit that did not happen, and the catch count converts your quality gate from an invisible cost line into a visible, quantified defense the room can put a value on.
• Replace reassurance with proof: "our quality is good" is a discountable claim from the team being evaluated, but the auditable per-segment quality record you already produce is verifiable evidence. Show the room the artifact an ISO certifier or client auditor would inspect, tell them it would pass unchanged, and cash the standards (ISO 18587, ISO 5060) in one sentence as a conformance credential rather than teaching the mechanism.
• In the worked readout, the ROI panel runs numbers ahead of a conservative base and inside a ceiling and honestly flags the one line behind plan; the quality panel reports zero escaped and eleven caught Criticals across 1.4M regulated words with 98.5% conformance and 100% high-liability routing; the reconciliation admits a J-curve crossover that landed one quarter late; and the ask puts two evaluators against holding the escaped-Critical rate at zero through a volume doubling, with both funded and unfunded outcomes stated on the record.
• Make the readout the surfacing of a standing, instrumented record rather than a panic build, so you can drill from any summary number to its underlying detail when an executive pushes, and report the same axes in the same shape every period so the trend (not the snapshot) tells the story, because that consistency is what makes your good record protective and your inevitable bad quarter survivable.