--- title: "System Reliability: The Distance Between Having No Outages and Being Reliable" description: "System reliability is proven not by the absence of breakdowns but by defined, consistently recorded measurement of availability, mean time between failures, and recovery duration. An unmeasured reliability claim is treated as unverifiable and priced against the contractual worst case, producing a discount in the revenue forecast and an enlarged contingent portion of the purchase price." url: https://www.beirek.com/en/blog/system-reliability-due-diligence canonical: https://www.beirek.com/en/blog/system-reliability-due-diligence published: 2026-07-10 modified: 2026-07-10 category: "Technology & Engineering" category_url: https://www.beirek.com/en/blog/category/technology-engineering language: en-US reading_time_minutes: 8 publisher: BEIREK LLC publisher_url: https://www.beirek.com license: "© BEIREK LLC — citation with attribution and link permitted" keywords: ["system reliability due diligence","availability MTBF MTTR measurement","incident record and downtime evidence","untested redundancy risk","key-person risk in technical operations"] topics: ["Operational reliability evidence in investment review","Valuation impact of unmeasured technical performance","Transaction structuring around availability commitments"] alternate_language_url: https://www.beirek.com/tr/blog/system-reliability-due-diligence --- # System Reliability: The Distance Between Having No Outages and Being Reliable > **In short:** System reliability is proven not by the absence of breakdowns but by defined, consistently recorded measurement of availability, mean time between failures, and recovery duration. An unmeasured reliability claim is treated as unverifiable and priced against the contractual worst case, producing a discount in the revenue forecast and an enlarged contingent portion of the purchase price. *In an investment review, system reliability is assessed not by the absence of failure but by how failure is recorded, defined, and institutionally absorbed. A reliability claim without a record is pulled toward the contractual worst case in the buyer's model, and that adjustment reaches valuation through revenue assumptions, indemnity scope, and the deferred portion of consideration.* --- When the technical session of an investment review opens, the first answer offered on system reliability takes the same shape with striking consistency: no serious outage has occurred, the line has run steadily, no material customer complaint has arisen. The answer is usually honest; the difficulty lies not in its accuracy but in its foundation, since what is being offered is recollection rather than record, and recollection is by definition unmeasured — the events that are remembered are the ones that were memorable. When the availability figure given by the production manager in that same session sits a few points away from the figure carried in the commercial team's presentation deck, there is rarely a document capable of explaining the gap, because the two numbers were generated from two different definitions that have never been placed side by side. The second pattern, which follows immediately, is more determinative. When an incident log is requested, what arrives in the data room is typically the maintenance contractor's invoice set, a folder of email correspondence, or photographs of the shift book — documents produced after the event and for another purpose entirely. Even where a downtime figure can be extracted from such material, the question of when the stoppage began, which portion of planned maintenance was deducted from the availability calculation, and how hours of partial-capacity operation were counted must all be reinterpreted retrospectively. What the reviewing party is looking for at this point is not a high number; it is the ability to show which definition produced the number, and that the same definition was in force two years earlier. Reliability is not a single figure but the resultant of three quantities that diverge from one another: the proportion of total time in which the system is available, the mean operating interval between two failures, and the time required to return to normal after a failure. A line that stops frequently but recovers within minutes and a line that stops once a year but waits three weeks can produce an identical availability rate; their contractual consequences, however, run in opposite directions, since the first presses on quality and scrap while the second presses on delivery commitments, liquidated damages, and the deductible threshold of business interruption cover. A company speaking through a single aggregated rate cannot show the counterparty the difference between these two profiles, and therefore cannot influence which of the two the counterparty will assume. The mechanism underlying this absence of measurement is not negligence but a shortcut that is entirely functional at a particular stage of growth. In a company's early period, the cheapest way to secure reliability is a single individual who recognises the machine by its sound; that person senses the fault before it arrives, intervenes without consulting a procedure, and reports the outcome to no one, since he is himself the intended recipient of any report. So long as this arrangement holds, it generates no cost and is in fact faster than a formal system, carrying no documentation burden. The problem with the shortcut lies not in the shortcut itself but in its persistence once conditions change: when lines multiply, when shifts are added, or when the company gives an availability undertaking in a contract, the same arrangement ceases to be a capability and becomes a risk tied to one person's calendar. The same mechanism operates more quietly on the redundancy side. A spare may exist for critical equipment, a second supply feed may have been installed, a changeover procedure may be written down; yet when the question is asked as to when that spare was last actually run under load, the answer is frequently that it has not been since commissioning. Untested redundancy is an assumption rather than a capability, and assumptions are tested precisely at the moment they are needed. At the review table the physical existence of redundancy is largely taken as given; what is in contention is whether redundancy has been tied to a drill calendar, whether the outcome of each drill was recorded, and whether a failed drill translated into a corrective action. The channel by which this picture reaches valuation is direct and more mechanical than is usually anticipated. In the absence of a verifiable reliability series, the buyer's model does not fill the gap with an optimistic assumption; it fills the gap with the commitment threshold contained in the company's own signed contracts — that is, with the worst case. The result is an upward adjustment of the downtime assumption in the revenue forecast, full modelling of liquidated damages and service-credit line items, and maintenance and repair expenditure taken from the upper industry band rather than the historical average. The compound effect of these three corrections frequently produces a larger swing in value than the multiple under negotiation, and it is difficult to resist in bargaining, because the data required to rebut the counterparty's assumption does not exist on the seller's side. A second channel runs through working capital and spare parts. An equipment inventory that has not been classified by criticality keeps invisible, until the review stage, the fact that a component capable of halting production depends on a single supplier or has been declared end-of-life by its manufacturer; once surfaced, that finding produces two consequences at once — a capital reserve to be set aside for the post-closing period and an inventory burden to be carried across the procurement lead time. Business interruption insurance is similarly sensitive to the quality of the historical incident record: where the record is weak, premium and deductible are calculated not from realised risk but from risk that cannot be demonstrated. The third and often costliest channel is visible in the structure of consideration. In a file where reliability cannot be measured, negotiation concentrates less on price itself than on which portion of the price is paid in cash at closing: representations and warranties expand to cover technical performance, the escrow proportion rises, an independent technical verification report is imposed as a condition precedent, and a portion of consideration is shifted into an earn-out tied to an availability threshold to be achieved over a defined post-closing period. From the seller's perspective, the meaning of this is that the risk of measuring a previously unmeasured quantity remains entirely with the seller; the buyer prefers not to pay for what he has not measured. The continuity dimension enters valuation through key-person risk. Where reliability is found to reside in an individual rather than a role, that individual's retention ceases to be a human resources matter and becomes a transaction condition; a binding employment undertaking, a non-compete covenant, and frequently the linkage of part of the consideration to that person's continued service all come onto the table. The substantive issue here is not whether an individual is replaceable but whether the knowledge exists in transferable form: the ability of two different technicians on two different shifts to resolve the same fault in the same sequence evidences institutional capacity, whereas a resolution possible only through a single telephone call evidences that the capacity sits with the person rather than the company. The intervention that neutralises this tendency is system design rather than awareness, and it typically separates into five components: first, a single incident record created at the moment the event occurs rather than the moment it is reported, with mandatory start time, end time, and root-cause fields; second, written and frozen definitions of availability, planned maintenance, and partial capacity, matched line for line against the definitions used in customer contracts; third, an asset register in which equipment is classified by criticality and single points of failure are explicitly flagged; fourth, a drill regime under which redundancy and changeover procedures are actually exercised on a calendar; and fifth, the assignment of the whole domain to one named owner holding both a budget line and stop authority. BEIREK's intervention in this area rests on treating reliability, from the construction phase onward, as a recording discipline rather than an output: commissioning test results are fixed as the starting point of an incident record kept in identical format from the first operating day, so that when the company sits at a review table two years later it brings a continuous data series rather than one assembled retrospectively. Because the definitional gap between the availability undertaking given in contracts and the quantity the operation actually measures typically surfaces in the first months of operation rather than in negotiation, reconciliation of those two texts is run as a distinct control step, and this step prevents liquidated damages exposure from accumulating in an interval no one is watching. The second line of intervention concerns ownership and rhythm. Reviewing reliability indicators in a monthly session where deviations are explained by the accountable role rather than the founder; making the drill calendar, the critical spares register, and the end-of-life inventory standing agenda items of that session; and closing each corrective action against a date and a name — this is itself the mechanism that transfers person-held knowledge into the institution. Once the arrangement is in place, the same information feeds both the operation's daily decisions and a future due diligence question from a single source; no separate preparation exercise is required, because preparation has become the ordinary rhythm of the business. The question actually being asked at the review table is never whether the system has broken down; it is whether what restored the system was an institution or a person. Having experienced no outage is not a performance indicator but another name for not yet having been tested, whereas a failure that occurred, was recorded, had its root cause identified, and was prevented from recurring constitutes stronger evidence for most buyers than an unblemished history — because what it demonstrates is not the durability of the machine but the learning capacity of the organisation. ## Key Points - Reliability is the resultant of three distinct quantities — availability rate, mean interval between failures, and recovery duration — each of which carries different contractual consequences. - When incident records are created at the moment of reporting rather than the moment of failure, the record ceases to be a measurement instrument and becomes a retrospective narrative. - Untested redundancy is an assumption rather than a capability; the review table asks not whether redundancy exists but when it was last actually carried under load. - When the availability definition written into customer contracts diverges from the definition the operation measures, liquidated damages exposure accumulates in an interval no one is monitoring. - Reliability that rests on one person's knowledge is priced by the buyer as key-person risk and shifts consideration from cash at closing toward contingent structures. ## Questions ### How is system reliability measured in a due diligence process? The review looks not at a single availability rate but at three distinct quantities: availability across total time, mean operating interval between failures, and recovery duration after a failure. All three must originate in a record created at the time of the event, under a definition that did not change across the period. Series reconstructed retrospectively from invoices or email correspondence are not treated as verifiable and are priced in the counterparty's model against the contractual threshold. ### Is a facility that has never experienced a serious failure considered reliable? The absence of failure reads as evidence of not having been tested rather than as evidence of performance. Where no material event has occurred, no observation has accumulated regarding the system's behaviour under fault conditions; whether redundancy actually functions, what recovery duration truly is, and what intervention capacity the organisation holds all remain undetermined. In that situation drill records, commissioning test results, and the critical equipment register become the only available verification ground. ### Through which channels does unmeasured reliability reduce valuation? Three channels typically operate together. On the revenue side, the downtime assumption is pulled toward the commitment threshold written into contracts and liquidated damages items are modelled in full. On the cost side, maintenance and spare parts requirements are calculated from the upper band and a post-closing capital reserve is set aside. In transaction structure, representation and warranty scope widens, escrow rises, and part of the consideration shifts into an availability-linked earn-out. ### How does reliability resting on a single individual affect the transaction? Where knowledge sits with a person, that person's retention ceases to be a human resources matter and becomes a transaction condition: a binding employment undertaking, a non-compete covenant, and frequently the linkage of part of the consideration to continued service. The distinguishing test is whether the same fault can be resolved in the same sequence on two different shifts; if resolution requires a single telephone call, capacity is treated as residing with the individual rather than the company, and the discount is calculated on that assumption. --- Source: https://www.beirek.com/en/blog/system-reliability-due-diligence Publisher: BEIREK LLC — https://www.beirek.com