The right procurement intelligence platform helps your team find relevant work, understand the evidence and act in time. Write down your target buyers, jurisdictions and contract types. Then ask shortlisted providers to complete the same test using public notices you select.
This guide is published by Duke, a provider in this category. It offers a purchasing method and a source-linked shortlist, not an independent product ranking. Vendor descriptions were checked on 13 September 2026. They describe advertised positioning, not verified coverage, contractual entitlements or comparative performance. We have not conducted a controlled hands-on comparison of these products.
Define the job before comparing features
| Your bottleneck | What the platform must demonstrate | Evidence to retain |
|---|---|---|
| Relevant tenders are missed | Discovery across your required sources and stages | Reconciliation against known notices |
| Alerts contain too much noise | A profile that separates relevant work from lookalikes | Reviewed examples with reasons for inclusions and exclusions |
| Qualification takes too long | A traceable route from summaries to original documents | A completed decision checklist with references |
| Research starts too late | Buyer history and qualified forward-looking signals | Source dates, supporting records and uncertainty labels |
| Decisions and deadlines get lost | Assignment and review using your team's roles | A demonstrated sample pursuit |
Write a buying brief: “We sell [offer] to [buyers] in [jurisdictions]. We need [notice stages and lots]. Today [problem] costs [measured time]. Success means [observable improvement].” Agree on this with the people who will use the tool.
Test coverage against known notices
Select 20–30 public notices across your sources. Include a current opportunity, amendment, award, framework and a difficult below-threshold example where publicly available. Record each native identifier, source URL, publication date and check time.
Ask the provider to find every record, show the source and explain absences. Separate notices found, documents accessible and records available for monitoring. Do not compare a count of notices with another provider's count of procedures. One procurement can generate several notices and lots.
For recent publications, measure delay over several observations. A single fast result does not establish an always-current service. Ask which sources are monitored, which records are historical imports, and which documents require separate access. Include required coverage and its limitations in the agreement you buy.
Test relevance with your own examples
Give each provider the same offer description and a labelled sample: relevant, irrelevant and uncertain. Include ambiguous terms, neighbouring classifications and geographic edge cases. Review the same period and record known misses as well as noisy results.
Illustration: 24 relevant results among 40 reviewed alerts gives 60% sample precision. It does not tell you how many relevant opportunities are absent. Estimating recall requires a separately assembled set of relevant records. Alert volume alone measures neither quality nor completeness.
Use the monitoring guide to define a profile and the bid/no-bid checklist to define useful qualification.
Demonstrate the full decision workflow
Take one difficult tender from discovery to a recorded decision. Identify the lot, current deadline, mandatory requirements, evaluation method and submission channel. Check whether summaries link to the original document passage and how missing, replaced or contradictory documents appear.
For buyer intelligence, ask how organizations are identified, joint awards and lots are represented, and estimates are distinguished from award values. A possible renewal inferred from duration must remain distinct from a published future procurement. The buyer-research guide explains these checks.
Involve an actual user in assigning work, recording a reasoned decision and retrieving it. Test the integrations, permissions, export formats and limits included in the proposed subscription. A presentation of an integration logo is not an integration test.
Providers to investigate
These examples are not ranked. Confirm the current source coverage, features and commercial terms with each provider. Absence from this list does not imply a product is unsuitable.
| Provider and source | Advertised emphasis | Question to test |
|---|---|---|
| Stotles | Public-sector sales and bid workflows, buyer intelligence and qualification | Can your sales and bid teams complete a sample pursuit in the proposed subscription? |
| Tendium | Public-sector sales and tender workflows | Which of your portals, languages and document types are supported? |
| Tussell | Government contracts and spending intelligence | Can the records answer your specific buyer and account-planning questions? |
| Hermix | Public-contract intelligence and analysis | Can you trace the relevant market and supplier findings to source records? |
| Duke | Tender monitoring and procurement intelligence | Can we demonstrate your required sources, document evidence and operating workflow? |
Apply the same test to additional candidates. We do not infer a competitor's limitations from its homepage or assign untested ratings. Request current prices and verify the exact plan; public positioning alone does not establish an entitlement.
Compare cost using measured work
Ask for the full annual cost: subscription, seats, markets, onboarding, integrations, exports, support and usage allowances. Confirm renewal and cancellation terms.
Illustration: saving six research hours a week at an internal cost of €50 an hour over 46 working weeks represents €13,800 of annual capacity. It is not guaranteed cash savings or revenue. Subtract the time spent reviewing alerts and maintaining profiles. Keep expected additional contract wins separate unless you can support an attribution model.
Record the purchase decision
Set mandatory gates first: required sources, usable original documents, acceptable data handling and viable price. A high aggregate score must not compensate for failing a mandatory gate.
For providers that pass, weight dimensions according to your bottleneck. Illustrative weights are 35% relevant discovery, 25% qualification, 20% buyer research and 20% team operation. Set your own before the demonstrations. Mark each result as demonstrated, provider-stated, not tested or failed; retain the evidence and subscription tested.
If you want to evaluate Duke, book a demonstration with your buying brief and sample notices. Apply the same standard to us.