Why You Did Not Lose That Energy Tender on Price

The loss was decided before the price envelope opened

The familiar version of this argument is that price is an excuse. That is true but not useful, because it leaves you with nothing to do on Monday. The stronger and more specific claim is this: in long cycle energy procurement the outcome is usually determined before the commercial envelope is opened, at the shaping of the specification, at prequalification, at a technical minimum threshold, or inside the buying group's own disagreement. The price score is the record of that earlier decision, not its cause.

Three features make energy worse than generic business to business selling. The published weightings run the other way, with quality routinely carrying three to five times the points that price does. The feedback that would correct the misdiagnosis now arrives by statute in UK public and utilities procurement, and is almost never systematised. And the sample is small while the cycle is long, so a wrong loss reason propagates through bid and no-bid decisions, pricing policy and proposal boilerplate for years before anything contradicts it.

The consequence for a supplier is a doom loop that looks entirely rational from the inside. You lose a tender. The score sheet shows a price deficit. You cut price on the next bid. You lose again, because the actual deficit was on a scored technical criterion nobody examined. You conclude the market is irrational or the incumbent was preferred, and you stop bidding. Kelly and colleagues, studying UK public sector suppliers in the Journal of Public Procurement in 2021, documented exactly these three responses to tendering dissatisfaction: non-response to future tenders, deliberately lower quality submissions, and damaged buyer relationships.

The honest caveat, and it matters: there is no published win and loss study specific to energy, utilities or engineering procurement. The research base for this discipline is almost entirely software and professional services. That absence is itself the argument for doing the work yourself rather than importing a benchmark, and it is why everything below is built on procurement documents, buyer research and method rather than on an energy specific loss statistic that does not exist.

المشروع 54The people who decide most energy tenders are technical evaluators and operational sponsors, not the procurement lead who sends the rejection letter.

Buyers and sellers are describing different events

Start with what proposal teams say about themselves, because it is the cleanest measurement of the bias. Loopio's 2026 RFP Response Trends and Benchmarks Report, covering more than 1,500 proposal teams and published on 26 March 2026, found price has been the primary self-reported reason for losing since 2021, tied with competition at 55 per cent, while only 13 per cent attribute a loss to their own proposal quality. External cause 55 per cent, internal cause 13 per cent. That is attribution bias with a number on it. The same report puts average win rate at 39 per cent against a 45 per cent historical average, with 166 responses per organisation per year and 75 per cent operating a formal go or no-go process, rising to 81 per cent among top performers.

Then compare what the CRM says with what the buyer says. Clozd's analysis of 1,000 closed-lost deals found the recorded loss reason aligned with the buyer's own account only about 15 per cent of the time, 44 per cent of recorded reasons were outcomes rather than reasons, and the competitor field was wrong in nearly seven out of ten deals. Corporate Visions reports that sellers and buyers cite different reasons 50 to 70 per cent of the time, that 53 per cent of deals marked lost were winnable but for a fixable sales process misstep, and that 12 per cent of deals marked lost had buyers still actively considering the vendor. Both sets are vendor research without published methodology, and should be quoted with attribution rather than presented as independent fact. No study contradicts them, which is not the same as corroboration.

The underlying psychology is older and much better established. Nisbett and Wilson, writing in Psychological Review in 1977, showed that people routinely report causes of their own judgements that are inferred after the fact rather than introspected, and do so with confidence. A debrief answer is a post-hoc account, not a readout of what happened. Anyone building a programme on the assumption that buyers know and can articulate why they chose is building on sand, and has been for fifty years.

Three more findings change how you read any single answer. Forrester's State of Business Buying 2024 found 86 per cent of business to business purchases stall during the process, more than 80 per cent of buyers are dissatisfied with the provider they ultimately chose, and the average purchase involves 13 people across two or more departments. Amy Hayes, VP and Research Director at Forrester, summarised it bluntly: a broken B2B buying process is creating mayhem for buyers and providers. If four in five buyers are unhappy with the winner, the loser's assumption that the winner was simply better value is often wrong.

Gartner's survey of 632 buyers, fielded in August and September 2024, found 74 per cent of buying teams show unhealthy conflict during the decision, with groups running from five to sixteen people across as many as four functions, and consensus reaching groups 2.5 times more likely to report a high quality deal. Delainey Kirkwood of Gartner's Sales Practice put the range plainly: buying groups are more diverse than ever, ranging from five to 16 people across as many as four functions. The reason you are given is one participant's account of a contested internal argument you were not in.

Finally, the competitor you think beat you often did not. The JOLT Effect research, based on machine analysis of 2.5 million recorded sales conversations, attributes 40 to 60 per cent of lost deals to customer indecision rather than to a competitor or to price. In tendered energy work the analogue is the cancelled, rescoped or indefinitely deferred award, which most CRMs record as a competitive loss because there is no other field for it.

Small universe, statutory feedback, eight year lock-outs

The weightings are the first difference, and they are public. The Netherlands allocated 85 per cent of tender points at IJmuiden Ver to qualitative criteria including environmental and system integration solutions, international corporate social responsibility and circularity, with 15 per cent on the financial bid. Germany's 2023 offshore wind tender for 1,800 megawatts at centrally pre-examined sites allowed up to 60 points for bid value and up to 35 for qualitative criteria covering renewable content in turbine manufacture, apprentice employment, environmentally friendly foundation methods and long term power supply to third parties. The UK Cabinet Office's own guidance says the weighting of price and quality should reflect the characteristics of the service and potential outcomes should be tested with the market before the weighting is fixed, and explicitly warns against criteria design that produces near-identical quality scores across bidders, which is precisely the mechanism that makes price look decisive.

The legal frame has moved further in the same direction. The Procurement Act 2023, in force since 24 February 2025, replaced Most Economically Advantageous Tender with Most Advantageous Tender, removing the obligation to include price in every award assessment at all. National Grid now includes a social value weighting in 100 per cent of large scale contract awards, and requires 75 per cent of carbon intensive suppliers by spend to set science based targets. A supplier still telling itself the market buys on price is describing a procurement regime that has been superseded.

The second difference is the buying group. Forrester counts 13 people across two or more departments. Gartner counts five to sixteen across up to four functions with 74 per cent showing unhealthy conflict. An energy invitation to tender routinely adds a technical authority, HSE, quality assurance, project controls, legal, an operational end user, a joint venture partner and, in project financed work, a lender's technical adviser, several of whom score sections of your submission and none of whom you have met.

The third difference is market access, and it is where the invisible losses live. Under the Procurement Act 2023 regime, utilities frameworks may run up to eight years against four for non-utilities, and private utilities face no maximum term at all. Dynamic markets have replaced qualification systems, and a qualifying utilities dynamic market lets a utility issue a tender notice to members of the dynamic market rather than publishing openly. A supplier that is not a member never sees the opportunity, never records a loss, and concludes from its CRM that it has a healthy win rate.

The fourth difference is frequency and duration. IJmuiden Ver Alpha drew two bids and Beta drew two, with Eneco and Equinor withdrawing before the deadline. The Energy Transitions Commission puts the current baseline for offshore wind at roughly twelve years from site mapping to operation. Constructing Excellence found bid costs averaging 0.57 per cent of project value across 8 billion pounds of costed bids, split 0.48 per cent on losing bids and 0.65 per cent on winning bids. Philip Collard's summary of that finding is worth keeping in front of any bid committee: when consultants and contractors invest more in the work-winning process on a bid, they are more likely to win.

The fifth difference is the one nobody uses. Under the Procurement Act 2023 contracting authorities must issue an assessment summary to each bidder containing feedback on the scores of both the recipient's bid and the winning bid, ahead of an eight working day standstill period running from publication of the contract award notice. That is a free, structured, criterion by criterion, quantitative comparison against the winner, delivered by law, on every regulated loss. It is the single most under-used win and loss dataset in the sector. Note the asymmetry: private utilities may skip the mandatory standstill on direct awards and face no equivalent feedback obligation, so a programme has to cope with both a statutory data feed and a total blackout.

Neutral interviewer, four to ten weeks, questions sent in advance

Whoever owned the bid must not conduct the interview, and neither must anyone whose bonus touches the deal. The response rate evidence is consistent: Corporate Visions reports 20 to 25 per cent response on wins and 10 to 15 per cent on losses when a neutral party interviews, rising above 70 per cent when outreach timing is right, and Clozd reports 15 to 30 per cent participation for interviews against 3 to 5 per cent for surveys. Ryan Sorley of DoubleCheck Research put the seller's problem precisely: it's just really easy for the buyer to say price. It kind of lets them off the hook. In descending order of data quality, use an independent researcher, then an internal person with no commercial line such as a technical or quality director, then the managing director explicitly framed as not a sales conversation. Never send two people. Never let sales sit in to listen.

Timing has two fixed points. Obtain the assessment summary inside the standstill window, as of right, because it is data rather than a favour. Then run the research interview four to ten weeks after award: early enough for reliable recall, late enough that the winner has mobilised and you are visibly not preparing a challenge. Interviewing inside the standstill reads as pre-litigation and closes the buyer down. Interview the wins as well, because a programme that only interviews losses cannot distinguish a cause from a constant, and wins respond at roughly twice the rate.

Ask for the right people inside the buyer, in this order: the technical evaluator who scored the sections where you lost points, the end user or operations sponsor who will live with the asset, the person who wrote or shaped the specification months before the invitation to tender, and procurement last, for process and arithmetic. Procurement administers the score. It rarely sets it.

Work the questions backwards from the buyer's process rather than forwards from your bid. How did this requirement originate, and who wrote the technical specification? Who was on the evaluation panel, by function, and who had a veto? What removed suppliers at prequalification? Were there pass or fail thresholds, and did any bidder fail one? Looking at our scores against the winner's, take me through the criterion where the gap was widest, and what did the winning submission contain there that ours did not? What did we assert that you did not believe, and what would have made you believe it? Where did you have to guess what we meant? What was the internal disagreement about, and who resolved it? And the question that converts a price story into a testable claim: if our price had matched the winner's exactly, would the award have changed?

What not to ask matters as much. Do not ask whether you lost on price, because it is leading and hands the buyer the easy exit. Let price arise unprompted, and raise it only at the end if it has not. Do not ask for the winner's price, commercial terms or submission content: in regulated procurement much of it is commercially confidential, and asking signals that you do not understand the rules. Do not ask whether the process was fair or whether criteria were applied consistently, which sounds like groundwork for a challenge and is a legal matter on a separate track with separate people. Do not defend the bid, because the moment you explain, the interview becomes a negotiation and the data stops. Do not record unless agreed in writing, and with NDA bound senior buyers assume the answer is no.

Four things make senior, NDA bound buyers say yes. Send the question set in advance, which senior technical people almost always prefer. Offer explicit Chatham House terms, no attribution to individual or organisation, notes only, findings aggregated. Cap it at 30 minutes and honour the cap, which matches the 15 to 30 minute benchmark for productive interviews. And offer reciprocity: an anonymised market feedback summary back to them, which in a small buyer universe is genuinely valuable and is the single most effective lever on participation. Route the request engineer to engineer rather than through the commercial relationship. If all you can get is a three line email on the widest score gap, take it. It beats a missing interview.

Code where the loss happened before you record why

Use a fixed taxonomy from day one, applied to every bid, and record the stage before the reason. Never saw the opportunity, because you were not on the framework, dynamic market or approved vendor list. Declined to bid. Eliminated at prequalification. Failed a technical minimum or pass-fail gate. Lost on scored quality. Lost on scored price. Lost on evaluated cost despite a competitive headline price, through terms, risk allocation or lifecycle. No award, cancelled or rescoped. Incumbency or relationship displacement. Only one of those nine is a price loss. In most energy portfolios the first four combined outnumber it, and all four are invisible to a CRM because they never reached a closed-lost stage with a reason field.

Seven metrics carry the programme. Score gap to the winner by criterion, taken from the assessment summary and tracked across bids, because the repeating deficit is the finding. Points per pound, meaning how many quality points you would have needed at your submitted price to beat the winner, which converts we lost on price into arithmetic anyone can check. Concordance rate, the percentage of losses where your internal reason matched the buyer's, which is the health check on every other sales number you hold. Loss interview coverage, split between assessment summaries obtained and interviews completed. Days from award to interview, which drifts upward silently. Pre-tender contact depth, meaning months of contact before the tender issued and whether you had met the technical evaluator, cross-tabulated with win rate. And bid cost as a percentage of contract value, split win and loss, against the 0.57, 0.48 and 0.65 per cent benchmarks.

The highest leverage change is bid and no-bid discipline, not proposal writing. Three quarters of proposal teams run a go or no-go and 81 per cent of top performers do. In energy the gates should be evidential rather than sentimental: no bid without pre-tender contact with the technical evaluator, no bid where the specification carries a competitor's fingerprints, no bid where you fail a stated minimum threshold and cannot remediate before submission, no bid on a framework whose term exceeds your credible capability roadmap. Every gate should be traceable to a coded loss.

Next comes market access, which is a different budget from selling. Losses that happened before evaluation are a framework and dynamic market membership problem, an accreditation roadmap problem and a two year capability investment problem. No amount of proposal quality fixes a tender you never saw. Then proposal evidence, at the specific scored questions where you repeatedly underperform, and the diagnosis there has three branches: you cannot prove it, you can prove it but did not, or you cannot do it. Only the third justifies investment. The first two are evidence library and writing work.

Last, and most commonly confused, is pricing structure versus price level. Losses on evaluated cost are the ones misreported as losses on price. The levers that move evaluated cost without touching your headline number are payment milestone profile, retention, liquidated damages exposure, warranty term, spares and consumables bundling, operations and maintenance or lifecycle cost inclusion, currency and indexation terms, risk allocation at interfaces, and delivery schedule certainty. Enrique Gonzalez Huitron of Nautech de Mexico described the incentive problem exactly: when a team is measured on how much they negotiated off the list price, they optimize for that, adding that too many organizations stop at price comparisons and overlook integration costs, amortization, disruption risk, and working-capital drag.

The rule that follows from all of it: do not change the price level on win and loss data alone. With a self-reported price loss rate around 55 per cent and buyer to seller concordance around 15 per cent, a price cut driven by stated loss reasons will destroy margin without changing a single outcome. Redesign the commercial envelope first, and only reprice when the coded data shows a genuine scored price deficit that quality points cannot close.

Census, not sample, pooled over three years

The statistical framing is the wrong framing, and saying so out loud is the most useful thing a practitioner can hear. With twelve losses you are not sampling a population, you are enumerating one. A census has no sampling error. The question is not whether your n is large enough to generalise, it is whether your coverage is complete enough to be honest.

The qualitative research supports this directly. Guest, Bunce and Johnson, writing in Field Methods in 2006 across 60 in-depth interviews, found data saturation within the first twelve interviews with basic thematic elements appearing by six. Kelly and colleagues, studying precisely this population of suppliers to public sector tendering, reached theoretical saturation after twenty interviews. Commercial benchmarks land in the same place, at 20 to 30 interviews per segment.

So the operating rules are these. Interview all twelve losses and all the wins. Pool across 24 to 36 months, which at twelve losses a year gives roughly 36 cases, above every saturation threshold above. Long cycles are an argument for a longer window, not for abandoning the method. Report counts, never percentages, below ten cases: four of our last eleven losses failed the same HSE management sub-criterion is defensible, whereas 36 per cent of losses on eleven cases is not, and will be attacked by the first person who dislikes the conclusion.

Code against the fixed taxonomy from the first case, because retrospective recoding of free text is where small programmes die. Anchor every interview to the assessment summary, so the score gap is the quantitative spine and the interview is the explanation. In a small sample that pairing substitutes for statistical power: you are not inferring a distribution, you are reading the actual scores and asking why. Treat the whole thing as a longitudinal case series, where the unit of learning is the recurring mechanism across bids rather than the distribution of reasons within a year.