Start With the Problem, Not the Product
Most DEX selections go wrong in the first week, when the team starts comparing feature lists before agreeing what problem it is solving. Every serious platform has telemetry, dashboards, and automation. The differences that matter show up only against your specific constraints, your existing stack, and your operating model. This guide gives you a sequence, a weighted scoring worksheet, a question bank for the RFP, and a design for a proof of value.
Disclosure. Dexterity Digital's delivery practice is Nexthink-first, and we say so on every page where it matters. This guide is written to be vendor-neutral: the criteria and questions work for any platform, including ones we do not implement.
The Selection Sequence
- Name the top three constraints. Which problems cost the most productive time for the most employees? Write them down with whatever data you have. These become the test cases for every vendor.
- Inventory the stack. Endpoint management (Intune, Workspace ONE, Ivanti), ITSM (ServiceNow, Jira), VDI platforms, security agents already on the device, and existing analytics. This determines which "already in your stack" options are realistic and what integrations matter.
- Agree the weights before seeing a demo. Adjust the worksheet below to your priorities and sign it off with stakeholders first. Weights set after demos tend to drift toward whichever vendor presented best.
- Shortlist three to four platforms. Use the DEX Tools Landscape and your stack inventory. Include at least one option you already license, such as Intune Advanced Analytics or ServiceNow DEX, as the baseline.
- Issue the RFP. Use the question bank below and require written answers, not links to marketing pages.
- Run a proof of value on real devices. Two finalists, 30 to 60 days, a representative device group, and success criteria agreed in advance.
- Score, decide, and plan the operating model. The platform is half the decision. Who will run it, and how, is the other half.
Weighted Scoring Worksheet
Suggested starting weights total 100. Score each vendor from 1 (weak) to 5 (strong) per criterion, multiply by the weight, and sum. Re-weight for your environment: a VDI-heavy organization should raise VDI support; an organization with a strict CISO should raise security and agent footprint.
| Criterion | Weight | What to test | Score |
|---|---|---|---|
| Telemetry depth and coverage | 12 | Which data is collected on Windows, macOS, Linux, VDI, and mobile; collection frequency; agent footprint | 1 to 5 |
| Real-time visibility and alerting | 8 | Time from an issue starting to IT seeing it; alert quality and noise | 1 to 5 |
| Remediation and automation | 10 | Library size, custom workflow building, ITSM-triggered actions, staged rollout | 1 to 5 |
| Automation governance and AI controls | 8 | Approval thresholds, impact assessment, kill switch, action audit trail | 1 to 5 |
| Employee sentiment and engagement | 8 | In-context surveys, notifications, correlation of sentiment with telemetry | 1 to 5 |
| Application experience | 7 | Web and SaaS page loads, transactions, application reliability | 1 to 5 |
| VDI and Cloud PC support | 5 | Session-level visibility for Citrix, AVD, Horizon, Windows 365 (weight higher if VDI is large) | 1 to 5 |
| ITSM and service desk integration | 8 | Live device context in tickets, ticket data back into the platform, one-click fixes | 1 to 5 |
| AI tool usage and shadow AI visibility | 6 | Which AI tools are in use, by whom, with what experience; governance policies | 1 to 5 |
| Analytics, querying, and reporting | 7 | Query language, dashboards, history retention, executive reporting | 1 to 5 |
| Security, privacy, and data residency | 8 | Certifications, data minimization and anonymization options, regional hosting | 1 to 5 |
| Vendor viability, roadmap, and ecosystem | 5 | Analyst positioning, customer references in your industry, partner capacity | 1 to 5 |
| Total cost of ownership | 8 | License, modules, implementation, ongoing operations headcount | 1 to 5 |
| Total | 100 | Weighted score = sum of (weight x score) / 5. Maximum 100. |
RFP Question Bank
Data and coverage
- List every data category your agent collects on Windows, macOS, and Linux, and the collection frequency for each.
- What is the agent's typical CPU, memory, disk, and network footprint, and how is it measured?
- How do you cover VDI sessions, Cloud PCs, and mobile devices, and are these separate licenses?
- How long is raw and aggregated data retained, and can retention be extended?
Remediation and automation governance
- How many prebuilt remediations ship with the platform, and how are custom ones built and tested?
- Can we restrict automated actions by device group, set device-count approval thresholds, and stage rollouts?
- For AI-initiated actions, what impact assessment is shown, who approves, and what is recorded?
- Is there a global kill switch for automated and agent-initiated actions?
Employee experience
- How are employee surveys and notifications delivered, and how is survey fatigue managed?
- How is sentiment correlated with technical data in reporting?
- What self-service or employee-facing assistant is available, and on which channels?
Integration
- Describe your ServiceNow (or other ITSM) integration: what appears in the ticket, what flows back, and what actions agents can trigger.
- Which endpoint management, identity, and collaboration platforms do you integrate with natively?
- What APIs and data export options are available for our data warehouse or SIEM?
AI capabilities
- Which AI features are generally available today, and which are beta or roadmap?
- Can the platform show which AI tools employees use, including unsanctioned ones, and the experience quality of those tools?
- Which models power your AI features, where is data processed, and is customer data used for training?
Security, privacy, and commercial
- Provide current security certifications and your most recent third-party audit reports.
- What anonymization and role-based data access controls exist for user-level data?
- Where will our data be hosted, and can we choose the region?
- Provide a three-year cost breakdown: core license, each module, implementation, training, and support.
- Provide three references of similar size and industry that we may contact directly.
Designing the Proof of Value
- Use real devices, not a demo tenant. A representative sample of 500 to 2,000 devices across models, locations, and personas reveals far more than a scripted demo.
- Test your constraints. Each finalist should show what it finds, explains, and fixes for your top three problems, on your data.
- Include the service desk. Have agents work real tickets with the platform and record time to resolution and escalation rates.
- Agree success criteria in writing before day one. For example: identifies the root cause of the top constraint within 14 days; measurable reduction in a target ticket category; agent footprint within a stated limit.
- Plan for the end. Decide in advance how agents will be removed if a vendor is not selected.
Red Flags
- Answers that point to marketing pages instead of specifics.
- AI features demonstrated only on prepared data, or described without saying whether they are generally available.
- No clear answer on agent footprint, data retention, or data residency.
- Pricing that only becomes clear after modules are added late in the process.
- No plan for who will operate the platform after go-live. Tooling without a DEXOps operating model produces dashboards, not outcomes.