Written Evidence submitted by Soil Benchmark (SH0055)



Soil Benchmark are a high-growth startup based in Northallerton, North Yorkshire, helping agronomists and farmers share best practice around improving soil health. Our platform helps agronomists and farmers to make better decisions by harnessing the untapped potential of the information that often sits under-utilised and siloed on individual farms – soil samples, cropping records, spraying logs.

We are proud to have been initially funded by grants from DEFRA and Ordnance Survey, and to work closely in partnership with ADAS, NIAB, the British Geological Survey, and Cranfield University.

Having launched in August 2022, our platform now supports the management of more than 93,000 acres of UK arable land.


Executive summary:

  1. How can the Government measure progress towards its goal of making all soils sustainably managed by 2030? What are the challenges in gathering data to measure soil health how can these barriers be overcome?

Countrywide data, regularly updated

  1. National soil surveys building on the historical datasets such as Countryside Surveys – with long term funding to create a ‘Soil Census’ every 5-10 years
  2. Utilising the data farmers already collect, through routine record-keeping, regulatory compliance, or simply generated by modern farm machinery – and particularly working with the companies who already have wide datasets recording this information – to cheaply and efficiently monitor, at a national scale, changes in soil health and how different management practices / subsidy schemes are influencing those changes


  1. What changes do we need to see in the wider food and agriculture sector to encourage better soil management and how can the Government support this transition?

More farmer-led research

  1. Supporting on-farm trials such as those run by Innovative Farmers – particularly by redirecting funding too often spent on ‘academic-led’ research
  2. Supporting more innovative approaches based on ‘hive-learning’ approaches – learning from the collective experience of thousands of farmers as expressed through the information they routinely collect




Full Submission:

Our submission mainly focuses on the first and fourth questions posed by the committee in its Call for Evidence:

  1. How can the Government measure progress towards its goal of making all soils sustainably managed by 2030? What are the challenges in gathering data to measure soil health how can these barriers be overcome?

Undoubtedly national level soil surveys will play a part – particularly if they can follow up from the previous Countryside Surveys for instance, using the same sampling locations. Most often data is useful as part of a timeseries, rather than in isolation, so it will be important to make sure that future national soil surveys link in with previous research. Ideally, long-term funding would be found to establish these soil surveys on a regular basis (perhaps every 10 years, similar to the census – a ‘Soil Census’?) – perhaps the committee could investigate what information the Treasury would require to sign off on such an undertaking?

However, we feel that these national level surveys, while useful in giving a high-level picture, are not nearly granular enough to monitor whether individual policies are improving soil health at the field level.

One solution would be to randomly select a portion of farmers for monitoring – for instance 1 in every 100 farmers collecting the SFI could have additional monitoring (or management practices, soil samples etc), paid for by DEFRA (ie when you join SFI you accept the chance that may be you). If these were chosen representatively, this could help give a good idea of what practices are actually improving key soil metrics (eg SOM levels) on different soil types, over time.

However, even this is potentially limited, and would require a large number of sample farmers to build a representative picture. Additionally, the concept of ‘forcing’ farmers to share data with DEFRA / RPA might not go down well.

A better idea may be for the government to utilise datasets already collected by trusted third party organisations (for instance DEFRA already works with ADAS’ YEN scheme, learning from the relatively large dataset they have built up over 10 years of operations). There are other, larger datasets of existing farm data (eg those collected by the large agronomy firms, or those that companies like Soil Benchmark are building up). While farmers would never want their information to be shared directly the DEFRA / RPA, it would likely be possible for the firms who are trusted by farmers to handle this information to be able to report back to DEFRA on the progress being achieved.

Perhaps DEFRA should do more to work with these managers of large datasets as an elegant and cost-efficient mechanism for monitoring progress of their key policies at the national level, without having to get into the expensive and potentially infeasibly business of trying to monitor representative groups of farmers directly.



  1. What changes do we need to see in the wider food and agriculture sector to encourage better soil management and how can the Government support this transition?

One change that we believe would help to encourage better soil management is to support more farmer-to-farmer learning. It is clear that this is how farmers prefer to learn, rather than from ever more top-down guidance, either from NGOs, researchers, or government.

Part of this is around supporting networks which encourage this sharing of knowledge, such as AHDB’s Monitor Farm network, or the emergence of farmer clusters across the country, both of which should be supported by government where possible.

However, we also believe that there should also be a change in how the actual research is done. There should be much more of a focus on farmer led research.

Currently best practice for agricultural research is to run plot trials in carefully controlled conditions, with replicates, and then peer reviewed results are trickled down to farmers. Surely Liz Truss proved to destruction that trickle down is not the best way of running things? The results of studies and trials are too often irrelevant to farmers who are in different contexts to those the trial was conducted in – on different soils say, or with a different climate. Or they simply couldn’t work at the scale of a real farming system.

There are admirable organisations such as Innovative Farmers who help to run trials on real farms – moving away from simply plot trials run at research centres – and these initiatives should be further supported by redirecting central government R&D funding which too often goes to researchers who sit too farm from the farm level to answer real problems farmers are facing in managing their soils.

However, we think there is also a potentially more transformative way to increase our knowledge of what practices actually work to improve soil health on real farms across each of the more than 1,800 different soil types we have in the UK. This is to recognise that all farmers are naturally ‘researchers’ who have already been experimenting on their fields for decades in most cases. Every year in every field on every farm is an ‘experiment’. When you start looking at many farms, going back many years, you quickly see that farmers have effectively run many millions of these on-farm ‘experiments'. With the range of farms trying new things in terms of soil management – some for many years – between them will be many of the answers to how we should best care for our soils.

And while they haven’t been monitoring each one as precisely as a scientific trial, there is still a wealth of information that is recorded in the normal course of business on each of those ‘experiments, and has been for many years in most cases. Either this is recorded in standard farm record keeping (yield records, livestock movements), automatically through modern farm machinery (yield maps, operations logs), or is collected due to cross-compliance, FRfW, or in future the new SFI regulations (soil samples, detailed of inputs applied). This information is currently just sitting siloed and under-utilised on most farms.

Projects like the AHDB-funded Soil Management Information System (SMIS) (co-ordinated by Cranfield University, but driven solely by the information shared by hundreds of growers) have shown that by looking at this siloed historical information together at scale it’s possible to uncover new insights that are useful for farmers in improving the health of their soils.

The SMIS project has shown that this approach can start to push forward our knowledge about what constitutes soil friendly farming, answering questions like ‘what rotation most improves organic matter levels on a particular soil type?’

There are so many potential questions that can be answered in this way. And this one resource of on-farm information can answer many more specific questions than the ‘one problem at a time’ approach of traditional replicated trials can. Added to that, traditional research approaches struggle to look at the long cycle of a full rotation because of time-limited funding cycles. In contrast, looking at the existing information going back over one or two rotations from a range of farms can answer specific questions that reflect on-farm practice.

The key is to provide mechanisms which start to pull together this farm information so that machine learning driven platforms like those that drove the SMIS project can start to learn from this aggregated information. The large national agronomy firms, such as Hutchinsons (through their Omnia platform) and Agrii (through Rhiza) have collected large resources of historical farm information which could help to drive further research to improve soil health nationally. Soil Benchmark, through our work with many of the independent agronomy firms are building up a similar resource. The knowledge being derived from these platforms should be shared throughout the industry and with government – rather than simply becoming any one firms intellectual property. Reversing national (and international) declines in soil health is too important for this potentially transformative new knowledge to become trapped behind a ‘paywall’.

We need to encourage an industry wide approach to pooling (anonymously if necessary) this information to allow government and farmers to learn as much as possible from it. If this can be achieved, we can learn from the combined historical experience of the many millions of ‘experiments’ that each farmer has effectively been running.

Not only would such a resource help farmers to learn what is likely to work best for each of their fields, with its unique location and soils, but it will help government to make better policy.

February 2023