This report was produced on 2022-04-20 by Lobo Data. We use machine learning and other statistical techniques to profile UK public procurement contracts currently being tendered.
The report first forecasts which companies are likely to bid for the requested contract. It details each of these company’s relevant public procurement experience, as well as their links to government (for example via meetings with MPs and donations). The report then forecasts the number of bids the contract is likely to receive, before profiling both the buyer and the market for the goods/services they are procuring. The analysis is based on a proprietary database of over 300,000 historic UK procurement contracts.
Much of the report’s content is interactive, enabling our clients to dig deeper into specific data-points (e.g. to see the specific contracts a competitor has won historically, or view their political connections). So we recommend the report is viewed in a browser, such as Chrome or Internet Explorer, rather than printed. All the content is available offline.
Tender details
This procurement intelligence report relates to the following contract:
| Title: | MIS/DISINFORMATION MONITORING AND ANALYSIS SERVICES |
| Buyer:* | DEPARTMENT OF DIGITAL CULTURE MEDIA AND SPORT |
| Amount: | £500,000 |
| Find a Tender reference: | ocds-h6vhtk-0328c8 |
| Contract Finders reference: | 0261b240-62fd-4d0c-8411-88816720c269 |
| Tender published: | 31 March 2022 |
| Tender deadline: | 30 April 2022 |
| Contract description | The Authority requires a Supplier to support cross-Government efforts to build up a comprehensive picture of potentially harmful misinformation and disinformation online through provision of Social Media monitoring and analysis of English language |
*This name may not be the same name as on the tender. Some buyers procure goods and services under multiple different names: we use a single, common name to refer to such buyers so that we can build a complete dataset of contracts they have previously issued. This ensures our metrics and analysis are based on the most comprehensive data possible.
Potential competition
Our algorithm has identified the companies listed in the first column of the table below as the most likely to bid for this particular tender.
The table also shows the number of relevant contracts each company has won. Pressing the black-arrow beside any number will reveal more detail about each of those contracts, so that our clients can assess how relevant their experience is to this particular tender.
When testing our algorithm on a dataset for this type of contract it hadn’t previously seen, it identified the ultimate winner of the contract 80% of the time. And so the list ought give our client a strong indication of which of the roughly 60000 companies that have won UK public contracts in the past 4 years are their closest competitors for this contract.
Table 1: Likely competition for this tender, including summary of tenders they have previously won.
*The drop-down arrow in the first column will only show the details of the largest 100 contracts a supplier has won. This is to prevent the overall report becoming too large. If the supplier has won more than 100 contracts, and you’d like to see all of them, please email daniel@datalobo.com
We estimate that in 2020 almost 10% of contracts (by value) went to companies which had some demonstrable link to politicians (e.g. they had met with them, donated money to them or employed them). The table below shows the known links between government and the likely competitors for this particular tender. Pressing the black arrow will reveal more detail on each identified connection, so that our clients can assess how salient they are to this tender (and thus inform their decision on whether to bid).
Table 2: Likely competition for this tender, including summary of political connections.
Winner profile
The ‘potential competition’ tables above only include companies that have won at least 3 public sector contracts in the past 5 years: to train our algorithm, we need at least some historic data to profile companies and thus predict which current tenders they’re likely to bid for.
And so they won’t include companies with less public sector experience, often SMEs. For this reason, we also use our algorithm to estimate what the chances are of a company with low/no experience (or an SME) winning the contract.
Chart 1: Likelihood that the contract will be won by an SME or a firm with low/no experience.
This contract is unlikely to be won by an SME, but SMEs have a better chance of winning this tender than previous contracts for similar goods and services.
There is a good chance this contract will be won by a firm with some experience of winning public procurement contracts, but not necessarily an extensive track-record.
Expected bids
Based on the number of bids historic contracts have received, our algorithm can estimate how many bids this particular contract is likely to receive. Our clients may wish to tailor the amount of resource and time they spend preparing for a bid, based on the number of competing bids they will face.
It expects this contract to recieve three bids or fewer, though contracts for this type of goods/service often recieve very few bids.
Chart 2: Expected number of bids for this tender.
Winning price
The buyer expects to pay around £500,000 for the goods/services being tendered.
Historically, this buyer has tended to award contracts for around the advertised price. On the occasions it has deviated from this, it has awarded contracts for both above and below the advertised price.
Chart 3: Winning price relative to price quoted on tender.
“Well below” - Award price was 85% or below of value listed on original tender. “Below” - Award price was between 85% and 98% of the value listed on original tender. “Around advertised” - Award price was within 2% of the value listed on original tender. “Above” - Award price was between 102% and 115% of the value listed on original tender. “Well above” - Award price was 115% or above of value listed on original tender.
Buyer profile
To help our clients better tailor their bids, the following section uses historic contract data to profile the issuer of this particular tender. The chart below plots the historic value of contracts issued by Department Of Digital Culture Media And Sport.
Chart 4: Distribution of historic tenders, by value
The size of this contract is typical for Department Of Digital Culture Media And Sport. But it is larger than the average public sector procurement contract.
The table below shows the suppliers that have won the most tenders with Department Of Digital Culture Media And Sport over the past four years. Pressing the black arrow in each row will reveal the contract details.
Table 3: Companies which have won most contracts with Department Of Digital Culture Media And Sport, including summary of those contracts.
The chart below shows the goods and services that this buyer has procured in the past four years. Contract notices generally include a high-level description of the good/service being procured: this is shown by the inner ring of the chart. They often provide a more granular description too: these are shown by the outer rings of the chart. Where a contract is for multiple types of good/service, we weight each equally. Clicking a particular segment of any ring will allow the user to zoom in on that type of good/service.
Chart 5: Sector breakdown of contracts issued by Department Of Digital Culture Media And Sport
Some contracts are for multiple different types of good and service: each type is shown separately on the chart.
The table shows all active contracts for this buyer, ordered by the amount of time remaining on the contract. This will not include contracts without a specified start and end date - as we can’t be sure whether they are currently active.
Table 4: Current contracts issued by Department Of Digital Culture Media And Sport, including summary of those contracts.
Industry profile
To help our clients better tailor their bids, the following section uses historic contract data to profile the market for the good/service that is being procured.
Chart 6: Size of tender relative to historic contracts
The size of this contract is typical for this industry. But it is
larger than the average public sector procurement contract.
The tables below show the largest suppliers for tenders in this industry. We can typically define a contract’s industry at three levels, with division being the broadest and class being the narrowist. In this case:
| Division: | Administration, defence and social security services |
| Group: | Administration services |
| Class: | Supporting services for the government |
Division
Group
Class
Broader sector intelligence
This section is designed to help inform our clients’ longer-term planning. It highlights how the public procurement market is changing, focusing on how which types of goods/services are increasingly in demand, which are types are less-and-less in demand, where in the United Kingdom that contracts are located, and the total value of the public procurement market.
The chart below, for example, shows the total value of contracts procured in the UK. By default, it shows how this total value is split between different types of buyer. The reader can use the drop-down menu to change the split shown (for example, to show the proportion of contracts which were awarded without first issuing a competitive tender).
Chart 7: Size of the UK public procurement market
The table below shows the twenty fastest growing procurement sectors.
Table 4: Goods/services being procured, ordered by 12-month growth rate in value of contracts issued.
The following table shows the twenty fastest shrinking procurement sectors:
Table 4: Goods/services being procured, ordered by 12-month gall in the value of contracts issued.
The map below shows which regions are experiencing an increasing share of the public procurement market.
Chart 8: Regional growth rate in value of UK public procurement contracts issued.
Annex
Dataset details
This report is based on a proprietary database, constructed using the raw contract data published on the various websites used by government bodies to procure goods/services(OJEU, Contract Finders, Find a Tender). A series of algorithms are then used to:
Identify and correct (or remove) misreported information. It is not uncommon for Contract Award notices to, for example, misreport the actual value of a contract.
Identify contracts issued by the same government entity, but under different names. The Foreign, Commonwealth and Development Office, for example, has issued contracts using (i) old names (e.g. “Foreign & Commonwealth Office”), (ii) abbreviated names (e.g. FCDO), (iii) different syntax (e.g. “&” instead of “and”), (iv) with typos (e.g. “Foreign dand Commonwealth Office”), (v) including additional information (e.g. “FCDO, formally DFID”).
Identify and remove duplicate contracts. Some contracts are issued on multiple different websites, but with modestly different titles, contract-issuer names or publication dates. We have a series of algorithms to try identify and merge such duplicates.
Identify what type of the public-sector entity the buyer is (e.g. a ministerial department of central government, a housing association, or a form of local government such as a Council). These categories enable our machine-learning algorithms to better predict the competitors for recently-issued bids.
Identify the Companies House register number of winning bidders. This enables us to read in additional data on these companies
Match in other datasets (e.g. companies house data on incorporation date, electoral commission data on political donations).
Methodology
We use machine learning to estimate how many companies are likely to bid for this tender and profile those bidders. Our algorithm considers a large number of factors, such as previous winners of tenders awarded:
- In this sector and in closely-related sectors.
- By this buyer and by similar buyers (e.g. other Councils if the buyer is a Council)
- In this location and nearby locations.
- For this value of money and for this length of time.
This approach requires us to categorise buyers into groups. For the purposes of this report, our algorithm categorises DEPARTMENT OF DIGITAL CULTURE MEDIA AND SPORT as a Ministerial Department.
Attributations
In preparing this document we made extensive use of the tidyverse package (v1.3.0; Wickham et al., 2019), reactable (Lin G,2022), R markdown and R Shiny (both maintained by RStudio). Header image from “Vecteezy.com”.
Feedback
Any feedback on this report would be very gratefully received: please email daniel@datalobo.com. Ideas on additional datapoints we could include, how to present the information more clearly.