Estimated reading time: 13 minutes
Under the Procurement Act 2023, social value data is now published and assessed – yet most organisations have never scrutinised it the way they scrutinise financial data.
Five common, avoidable errors can inflate sets of reported numbers, and each can become a business risk the at audit time, or when a procurement challenge or board question puts those figures under the spotlight.
The fix isn’t more effort from your team; it’s a smarter system that checks data as it’s entered. Here are the five errors, why they matter now, and how to tell where you stand.
Social value measurement has matured rapidly across the UK. Frameworks are in place, reporting is standard practice, and most organisations now collect data on employment outcomes, volunteering, local spend, and community engagement. But data quality has not kept pace with data collection. When practitioners look closely at social value data, the same five patterns of error appear across sectors, from construction to technology to real estate to government. Recognising them is the first step to fixing them.
Most organisations assume their social value data is roughly right. It was entered by someone who was there, it went into a system, and it came out in a report. But when practitioners and consultants look closely at data sets, the same five patterns come up again and again.
“Social value data quality is not up to scratch, not in the way you’d see with financial data, sustainability data, or even health and safety data across other industries,” says Danielle Aberg, Head of Social Value at Thrive. “Good data doesn’t happen by accident.”
None of these errors are caused by bad intent.
“It’s rarely with the wrong intent,” says Michelle Cole, Principal Social Sustainability Consultant at Thrive. “People go out and deliver real social value. But their data capture, having the right data aligned with the right metric, complete and evidenced, that’s where it falls down, an awful lot.”
Here are five errors that could be in your data right now, and why they matter more than they used to.
1. FTE overclaimed at start rather than accrued
FTE is “often misclaimed or miscalculated,” says Danni. Michelle calls it “a significant factor” across data sets she has reviewed in depth. It is the most common error across social value data.
The pattern: an organisation hires someone through a social value programme and records them as 1.0 FTE on the day they start. But FTE should be accrued over time based on how long the person has actually been employed, not claimed as a lump sum on day one.
Under the Impact Evaluation Standard, FTE is calculated from elapsed time since the employment start date, cross-referenced against the recording period and any other FTE already claimed against that individual. Someone who started three weeks ago has generated roughly 0.06 FTE, not 1.0.
Michelle explains that the two key issues are FTE being “claimed at start rather than accrued over time” and people not applying FTE correctly as a cumulative measure. The difference is significant, and it multiplies: if a proxy value is applied to an overclaimed FTE figure, the final reported number is inflated at two levels.
Most organisations making this error don’t know they’re making it. The metric name says “FTE” and they enter the number they think is correct.
2. Wrong metric selected from a similar name
Social value frameworks contain dozens of metrics, many with similar names. As Danni describes it, people read one word in a metric name and “put data against it without understanding the full metric definition.” A practitioner looking for the right place to record an activity sees something that looks right, enters data against it, and moves on.
The result is data recorded against a metric with different qualifying criteria, different evidence requirements, and a different proxy value from the one that should have been used.
Michelle recalls one organisation that took this to an extreme, using a single metric as a “holding ground, dumping hours, weeks, money, and FTE into a single field regardless of the metric’s actual unit.” The data was completely distorted.
Choosing the wrong metric means applying the wrong proxy value. The reported number may look plausible on screen, but it was derived from the wrong calculation. As Danni notes:
“We can’t rely on someone saying they’ve done something and just believing that it’s not an inflated number. We create a business risk when we put out numbers we can’t stand behind.”
In a system that accepts whatever is entered, the error passes through silently. It only surfaces when someone compares the data against the framework’s actual definitions, and by that point it may be embedded across months of reporting.
3. Missing evidence
This is the most understandable error on the list. Organisations deliver genuine social impact, but the data to prove it is not captured at the point of delivery.
Danni describes seeing claims submitted “without any evidence attached to that at all.” No supporting documentation, no context, no proof. The activity happened, but nobody recorded the evidence alongside it.
Michelle explains why this happens so consistently:
“As a social value manager, you’re so focused on delivering the impact that the data can become secondary. You’ve delivered this great thing, but you can’t prove it.”
She adds that social value managers “are people people, not data people,” and that the gap between delivering impact and capturing evidence is a systems problem, not a character flaw.
The absence of evidence does not mean the activity didn’t happen. But under the Procurement Act 2023, social value KPIs must now be published and assessed. In real estate, investors and GRESB assessors increasingly expect auditable evidence alongside reported numbers. As Michelle puts it: “People tick-box and collect the minimum, hoping no one circles back to it, because they know they’d have a problem. That’s the shift: it will be looked at, and it will need to be proved.”
This matters most at the moment the numbers leave your organisation. Evidence captured at the point of delivery is what turns a reported figure into a defensible one when it lands in a public ESG report, a stakeholder update or a GRESB submission – and it is what protects your organisation’s reputation if a published number is later questioned. Without it, the same report becomes a liability as soon as someone asks to see the workings.
4. Aggregated spend reported as a single total
SME spend, local spend, or supply chain contributions are reported as one figure rather than broken down by supplier, transaction, or location. A total of £1 million in local SME spend may look impressive in a report, but it tells the reader nothing about whether every transaction underneath genuinely met the criteria. Danni asks:
“If you’re giving me a total SME spend of a million pounds, how do I know that every single transaction under that was spent with an SME, and how much that was, and where that SME was based?”
She describes aggregation as “taking the easy route” but notes it would not survive audit.
Michelle sees the same pattern from the consultancy side. When she reviews user data in depth, aggregated figures are among the hardest to verify because the detail that would make them meaningful has been lost at the point of entry. The data looks clean because it is a single number. The problem is that a single number hides everything.
As reporting requirements tighten across both public and private sectors, the expectation will shift from totals to itemised, evidenced records. Danni’s observation applies across all five errors:
“We create a business risk when we put out numbers we can’t stand behind.”
There is an upside to getting this right. The same itemised, evidenced records that survive an audit are also what let you show exactly where the money went: which suppliers, which locations, and how much of it reached SMEs and VCSEs. That detail does more than protect the number – it gives you the evidence base to demonstrate your SME and VCSE spend and to make the case for growing it. A single total can never do that.
5. Frameworks become out of date
Measurement frameworks are not static. Metric definitions, units, and proxy values are updated as methodologies improve and government data sources are refreshed. Michelle describes a common pattern:
“When frameworks change, for example moving a metric’s unit from hours to weeks, organisations who don’t keep up with changes automatically overclaim because they’re still using the old unit.”
This is a silent error. The data looks correct against the version of the framework the organisation started with. It is only wrong against the current version. Nobody gets an alert. The gap grows with every update the organisation misses.
Danni reinforces the point: in social impact, “you don’t know what you don’t know.” An organisation that set up its reporting three years ago and has not reviewed its metric definitions since may be overclaiming on multiple metrics without any visible sign that something is wrong. The data looks consistent internally, which creates a false sense of confidence.
Frameworks with independent governance, like the Impact Evaluation Standard’s Steering Committee, publish updates and maintain version control. But the responsibility for applying those updates sits with the organisation. Michelle notes that this is partly a organisations’ responsibility issue and partly an education challenge. When the system enforcing data entry is not aligned to the current framework version, drift is inevitable.
Drift also undermines benchmarking. If a metric’s definition or unit has shifted underneath you, this year’s figures no longer compare cleanly with last year’s – or with a peer still working from the previous version. Benchmarking against inconsistent or incomplete data is worse than having no benchmark at all, because it looks authoritative while being wrong.
What to do about it
These five errors share a root cause. Social value data has not had the quality infrastructure that financial or compliance data takes for granted. The people entering the data are not the problem. The absence of systems that guide, check, and enforce at the point of entry is the problem.
This reframes what good data quality actually asks of your team. It is not more manual checking piled on people who are already stretched: the point of validating data as it is entered is that errors are caught while the person still has the evidence to hand, which is far less work than reconstructing it months later. Done well, it is a resource multiplier, not another job. It also changes social value’s standing inside the organisation. Data captured casually is what keeps social value a second-class “add-on”; data held to the same standard as financial or compliance data is what earns it a seat at the table.
Thinking about data quality systematically means looking at five dimensions:
- Completeness – is everything captured?
- Consistency – is the same activity measured the same way everywhere?
- Accuracy – would the data survive a reasonableness check?
- Timeliness – is data captured close to the activity?
- Auditability – could a third party trace a reported number back to its source?
Most organisations are strong on one or two of these and weak on the others. Knowing where you stand is the first step. Treating data quality as infrastructure rather than a one-off clean-up is what makes that position compound: each reporting cycle starts from a more defensible base than the last. It also insulates social value from shifting political and organisational priorities — once your data can withstand scrutiny, social value stops looking like a discretionary extra and starts looking like a governed part of how the organisation reports.
And it is not only a defensive play: when you can trust the numbers, you can see which interventions genuinely work, direct funding towards them, and give colleagues the confidence to champion the results internally. The organisations that address data quality now, before an audit, a procurement challenge, or a board question forces the issue, will be in a stronger position for every reporting cycle ahead.
Data quality is decided in delivery, not just at entry
One pattern sits underneath several of these errors, and it matters most for the teams delivering social value on contracts: data quality is decided across the whole life of a contract, not only at the moment a number is entered. Commitments are made in a bid, and then, too often, nobody owns them once delivery starts.
Research by WorldCC and Ironclad puts average contract value leakage at around 11% after signature; social value, which attracts far less scrutiny than cost or legal terms, tends to leak faster.
As Danielle Aberg puts it:
“The weighting gets social value into the procurement. What happens after determines whether it creates any worth.”
On one five-year contract she worked on, Danni notes, commitments made at bid stage were not raised again until the final review — by which point the evidence, and the value, were long gone.
The organisations that hold onto their social value, and the data that proves it, treat delivery with the same discipline as cost and risk.
The single strongest predictor, Michelle Cole notes, is simply whether a commitment has named owners on both sides — client and supplier — before delivery starts.
Practical habits follow: make social value a standing agenda item early in contract reviews, before time pressure pushes it down the list; tie data capture to the reporting systems the delivery team already uses rather than running a parallel exercise; and when circumstances change, document the change rather than quietly discounting it. Each of these lightens the load on small social value teams, because the data is captured in the flow of delivery rather than reconstructed at the end – and because it is captured as delivery happens, it shows which interventions are actually working on a given contract or site in time to do something about it, not months later when the opportunity has passed.
This also changes how a stretched social value lead wins support internally. The most effective route is rarely an appeal to ethics; it is translating social value into the commercial and reputational terms the rest of the organisation already cares about, and recruiting allies in procurement, legal, and programme teams who can make the case in rooms you are not in. Good data is what powers that argument: it turns “we think this is slipping” into named owners, dated decisions, and numbers on a page – the difference between being heard and being overlooked when priorities compete.
We have published a practical guide, The Data You Can Defend, which covers each of these dimensions in detail and includes a self-assessment tool you can use to rate your organisation across all five. The guide is available in four editions, each tailored to a specific sector:
- Construction and Infrastructure edition
- Technology and Professional Services edition
- Central Government and ALBs edition
- Real Estate edition
Download the edition most relevant to you for the full framework, including the self-assessment grid and practical guidance on what good looks like.
Q&A: Common Social Value Data Errors
Common enough that practitioners and consultants see the same five patterns across sectors, organisation sizes, and levels of social value maturity. Thrive does not publish specific error rates, but the types described above appear consistently when data sets are reviewed in depth. Most organisations will recognise at least two.
Does this mean our reported numbers are wrong?
Not necessarily. These are patterns that could be present in any organisation’s data, particularly where social value data has not been subject to the same systematic checks as financial or compliance data. The point is not that your numbers are wrong. It is that without a way to check them at the point of entry, you may not know either way.
To some extent, but it is expensive and unreliable. Retrospective correction means going back to data that was entered weeks or months ago by someone who may no longer remember the context. The more effective approach is to catch errors at the point of entry, when the person entering the data still has the evidence to hand and can correct immediately.
The cost of poor social value data stays invisible until it isn’t. A figure you cannot defend becomes a problem at the worst possible moment — during an audit, a procurement challenge under the Procurement Act 2023, or a board or investor question — and, as above, retrospective correction is expensive and unreliable. Building quality in at the point of entry is comparatively cheap and pays back every reporting cycle rather than once. Thrive does not publish a single figure for the cost of poor data, because it depends on an organisation’s size and exposure, but the pattern is consistent: those that invest before the issue is forced are in a far stronger position than those that wait.
Validation is automated, real-time checking at the point of data entry: does this record meet the framework’s rules? Verification is an independent review of whether the methodology was applied correctly across a data set. Audit readiness is the stage at which data and processes could withstand formal third-party scrutiny. These three steps form a progression, and validation is the foundation the other two depend on.
Any framework is affected if it relies on users to interpret metric definitions, select the right category, and enter complete data without system-enforced checks. The risk is lower in frameworks that build enforcement into the data entry process, that derive proxy values from publicly accessible sources, and that are governed by an independent body responsible for methodology updates. The Impact Evaluation Standard, which is governed by an independent Steering Committee and aligns over 1,000 Validation rules to its metric library, was designed with this problem in mind.
Alan Boyce
Senior Researcher More articles by Alan >
