Low-bid survey tendering: what the race to the bottom really costs

What Twickenham actually signalled

The UK geospatial surveying profession has launched The Geospatial Accord, a sector-wide initiative driven by the Geospatial Survey Association (GSA), formerly The Survey Association. It follows a stakeholder event at Allianz Stadium, Twickenham, that drew more than 100 leaders from surveying, construction, infrastructure, government, education and the technology supply chain. The room named the same set of problems the profession has discussed for years: a skills shortage, inconsistent standards, late involvement in projects, and sustained pressure to compete on price rather than value. Participants labelled this a “race to the bottom” and flagged it as one of the biggest risks to both the profession and the UK’s wider infrastructure programme.

GSA has committed funding, resources and expertise, set up a taskforce, and brought in O.agency to run communications. The stated aim is to reposition geospatial surveying “from ‘just another cost’ to an essential investment” that supports better decisions and reduced risk. As GSA CEO Rachel Tyrrell put it, “Geospatial surveying is often invisible because, when it is done well, everyone else can rely on it.”

That last point is the whole management problem in one sentence. The value of survey is the assurance it gives downstream decisions, and assurance is precisely what a price-only tender fails to buy. This is not a lobbying question for a function head. In our view it is a procurement and competence question we each control inside our own organisations, whether we buy survey services, sell them, or both.

Why “take the lowest compliant bid” quietly fails

The standard answer in most procurement functions is simple: screen bids for compliance, then award to the lowest compliant price. On paper it is defensible, auditable, and fast. In practice it fails for a reason that has nothing to do with the bidders’ honesty and everything to do with how the work is specified.

Compliance is binary only when the specification is complete. A survey specification that names a deliverable – “a topographic survey”, “a bathymetric dataset”, “an as-built point cloud” – without naming the uncertainty budget, the datum and realisation, the coordinate reference system and epoch, the feature-detection requirement, and the data-quality acceptance criteria, is not a specification. It is a wish. Every bidder is free to price the thinnest reading that still returns something called a survey, and the lowest bidder almost always does, because that is what the evaluation rewards.

Consider the difference between asking for “a bathymetric survey” and asking for a survey to IHO S-44 Order 1a with total vertical uncertainty and total horizontal uncertainty stated at the specified confidence level, a defined feature-detection size, and full-coverage multibeam. The first invites a single-beam line plan at wide spacing. The second constrains the method, the sensor suite, the run-line density, the calibration regime and the processing effort. One of these is a priced deliverable. The other is an assured one, and it costs more because it does more.

When the specification is thin, the lowest bid is not cheaper. It has simply moved cost and risk back across the table to the buyer – into re-survey mobilisation, into data rejected at acceptance, into programme delay, and worst of all into decisions taken on data that looked complete and was not. The cost did not disappear. It changed owner and changed timing, and it reappears at the point where it is most expensive to fix.

Where assurance is won or lost

Assurance is produced at four control points, and in our experience as a team we need to be explicit, at each, about who holds decision rights. Treating any of them as a formality is where the race to the bottom does its damage.

Specification authorship. Someone has to own the technical specification, and that person must be competent to write it to a standard – IHO S-44 for hydrographic work, IOGP geomatics guidance for offshore positioning and seabed survey, RICS measured-survey specifications and ISO 19157 for data-quality description onshore. In many organisations this document is assembled by procurement from a previous tender, with the technical content copied forward and the numbers left stale. If the author cannot defend every tolerance in the specification against the decision it supports, the specification is not yet fit to issue.

Pre-qualification. The purpose of a pre-qualification questionnaire is to confirm that a bidder can actually deliver what will be specified – quality-management certification to ISO 9001, demonstrable experience on comparable work, and named personnel whose competence is evidenced rather than asserted. This is where competence verification belongs, because it is far cheaper to exclude an unsuitable supplier before the tender than to discover the gap at data acceptance. The question of how we confirm that a party chief or processor is genuinely competent, particularly on contract or freelance crews, is not trivial, and the available routes to validating individual competence each carry different evidential weight.

Tender evaluation. If price carries all the weight, quality carries none, and bidders respond rationally by stripping quality out until they reach the floor. A method statement that is never scored is a method statement nobody writes properly. The evaluation is where we convert “value” from a slogan into a number.

Data acceptance. The final control point is the person who signs the deliverable off against the acceptance criteria. If those criteria are not in the contract – stated as measurable metrics, not adjectives – then acceptance becomes a negotiation under programme pressure, and programme pressure always wins. The surveyor who holds the acceptance decision needs the authority to reject non-conforming data and the backing to make it stick.

There is a continuity question running through all four points: what happens when the competent person is unavailable. If only one individual can write the specification or sign off the data, the assurance has a single point of failure. The answer is a documented acceptance procedure and at least two people competent to apply it, so that the integrity decision does not evaporate when someone is on leave or offshore. Where survey is procured as part of a larger contracted scope, the question of who carries the assurance across organisational boundaries has to be settled before mobilisation, not after a dispute.

What a quality-weighted tender costs and what it buys

Moving from price-only to a quality-weighted award is not free, and we should budget for it honestly rather than pretend it is costless virtue.

It costs evaluation effort. Scoring method statements and personnel against a published rubric takes competent assessor time – typically a technical lead for one to two days per significant tender, plus moderation. It costs specification effort up front, because the tolerances and acceptance criteria have to be settled before the invitation goes out rather than argued afterwards. It can narrow the field, because suppliers who were winning on price alone will score poorly and may decline to bid.

Against that, the purchase is specific. A meaningful quality-price split – one that weights quality substantially rather than treating it as a tie-breaker – buys bidders who price the real scope, method statements we can hold them to, named personnel we have checked, and a far lower incidence of data rejected at acceptance. The saving is in the costs we no longer carry: the re-mobilisation when a dataset fails coverage, the rework when the datum was wrong, the dispute when “survey” meant different things to each party, and the downstream engineering change when a decision was taken on inadequate data. Those costs routinely dwarf the tender-price difference between the lowest and the second-lowest bid.

There is a market structure point here that we think is routinely underrated. Price-only competition systematically favours whoever can carry the thinnest overhead, not whoever carries the best assurance, and over time it hollows out the capability needed on the harder jobs. The question of whether scale is an advantage or a liability in the survey market is live precisely because procurement behaviour decides which firms survive. If an organisation relies on a healthy supplier base for its difficult work, a procurement policy that rewards only the floor price erodes the very thing it depends on.

Building a procurement process that holds

The Accord is a collective effort, but the changes that matter are the ones we can each make inside our own remit this quarter. The following are concrete and testable.

1. Fix the specification before anything else. No survey goes to tender without a named accuracy standard and the parameters that make it real: IHO S-44 order (or RICS band / IOGP equivalent as appropriate), TVU and THU at the specified confidence level, feature-detection size, datum and realisation, CRS and epoch, coverage requirement, and the QA/QC deliverables we expect returned. If a parameter is missing, the tender is not ready. We hold the line on this even under schedule pressure, because every gap here becomes a dispute later.

2. Set a quality-price weighting and publish the rubric. Adopt a meaningful minimum quality weighting for standard work, and a higher one for safety-critical or decision-critical surveys, and publish the scoring criteria with the invitation. Score the method statement, the personnel, the QA plan and the risk assessment against defined descriptors, not against a general impression.

3. Make competence evidential, not asserted. Require named key personnel with CVs mapped to a recognised competence framework – IMCA competence guidance for offshore survey roles, RICS or equivalent for land – and make substitution of key personnel a change that needs approval rather than a free option after award. Verify the claims that can be verified.

4. Put acceptance criteria in the contract. State the data-quality metrics using ISO 19157 descriptors, define the rejection thresholds, and name the role that holds the acceptance decision. Ensure at least two people are competent to apply the procedure so that sign-off does not stall when one is unavailable.

5. Measure the right things. Track, per contract and in aggregate: the proportion of awards made on quality-price versus price-only; the first-time data acceptance rate; the re-survey or re-mobilisation cost as a proportion of contract value; and the number of specification clarifications raised during tender. A rising clarification count is a direct signal that our specifications are leaving room for the race to the bottom.

Watch the early signals that it is going wrong. Bids clustering tightly at the price floor mean the field is competing on how little to do, not how well. Falling continuity of named personnel between bid and mobilisation means competence is being swapped out after award. A drift toward accepting “near enough” at data acceptance under programme pressure means the acceptance criteria are not holding. Any of these tells us the discipline has slipped before the bad dataset arrives.

None of this requires a new regulator, and the Accord is explicit that it is not one. It requires us to treat the specification, the evaluation weighting and the acceptance decision as engineering controls rather than procurement formalities – because that is exactly what they are, and that is where survey assurance is bought or sold.

---

Based on: UK geospatial sector pushes back against ‘race to the bottom’