Buyer guide
A buyer scorecard for Microsoft Dynamics 365 specialists
Short answer
Score every proposed Microsoft Dynamics 365 specialist on the same evidence: one short task from your app and role, a four-level rubric, and checks of how the person communicates and documents work. MS Dynamics Guys publishes this scorecard for comparing candidates from any supplier before anyone gets access to your systems. Your interview still decides fit.
How to run the scorecard
Use the same five steps for every candidate, whichever supplier proposes them. The scorecard compares people for a role; choosing a company to run a whole implementation needs other criteria. Everything you need is on this page: print it, or copy the tables into your own hiring document.
Set the must-haves before you read profiles
Mark the rubric criteria that the role cannot do without, and write them down. Do this first, so one impressive profile cannot move the bar for everyone else.
Give everyone the same task and time limit
Choose a task from the task bank below, or write one from a real backlog item. Every candidate gets the same instructions, the same data and the same time limit.
Use invented data, in a sandbox only
Here, a sandbox means a test environment that holds only invented data. The candidate works in their own sandbox or development environment, or on paper. Nobody gets an account in your systems at this stage.
Two reviewers score alone, then compare
Each reviewer scores every criterion from 1 to 4 and writes one line of evidence. Then the two compare. Where they differ, they reread the evidence and agree on one score, with a written reason.
Stop at a 1 on any must-have
A must-have scored 1 ends the process for that candidate, whatever the other scores are. Tell the supplier which criterion ended it, and why.
Two reviewers are possible even in a small team. If only one person knows the product, that person scores the technical criteria. A second colleague can score communication and documentation, which need no product knowledge.
The four-level scale
Every criterion uses the same four scores. A score describes what the reviewers saw in the work, not how much they liked the candidate.
| Score | What the reviewers see | What it means for your decision |
|---|---|---|
| 1: wrong | The answer is wrong, missing or unsafe, and a hint does not fix it | On a must-have, the process for that candidate ends here |
| 2: prompted | The answer becomes right only after a reviewer points at the problem | A gap to probe in your interview, or to support in the first weeks |
| 3: tested and explained | The answer is right without help, a test or check proves it, and the candidate explains why it works | Meets the role |
| 4: adds limits and a simpler option | Everything in a 3, plus where the solution stops working and a simpler option, such as a standard feature | More than the role needs on this criterion |
The rubric: six criteria for every role
The same six criteria fit developers, consultants and engineers. The rubric describes a 1 and a 3 for each criterion. A 2 is a 3 reached only with help. A 4 is a 3 plus the limits of the solution and a simpler option.
| Criterion | Score 1: stop | Score 3: meets the role | Evidence to keep |
|---|---|---|---|
| Product and version fit | Works in the wrong product or version, or relies on something your version does not support | Uses supported extension points or standard settings of your product and version | The version the answer assumes, noted by the reviewers |
| Task result | The result fails, or nothing shows that it works | The result runs in the sandbox, and tests or a test script prove it | Code, a configuration export or test results |
| Failure handling | Cannot say what happens when a step fails | Explains what rolls back, what is retried and who receives an alert | The candidate's answers to your failure questions |
| Configuration first | Writes code for something a setting or standard feature already does | Tries the standard feature first, and gives a reason for each custom part | The written reason for each custom part |
| Communication | Asks no questions and states no assumptions | Asks about gaps, writes assumptions down and explains the result plainly | The candidate's questions and written assumptions |
| Documentation | Leaves nothing that another person could reuse | Leaves a design note, release steps and the work in source control | The documents themselves |
Next to every score, write one line about the evidence behind it, and keep that evidence: notes, code, test results or documents. When a supplier or a candidate asks about a decision, you can then show what each score rests on.
Task bank: one task for each role
Each task fits into one walkthrough and works with invented data. Pick the row for your role, then adapt it to your app and version.
In this table, CRM means customer relationship management: apps such as Microsoft Dynamics 365 Sales and Customer Service, which run on Microsoft Dataverse.
| Role | The task | A strong answer |
|---|---|---|
| CRM developer | Review a short demonstration plug-in, and explain what happens to a record save when the plug-in throws an error | Names the stage and mode of each step. Explains that an error in a synchronous step cancels the save and rolls back the whole operation. Asks first whether a declarative option, such as a business rule or a flow, would do the job. |
| Business Central developer | Add one rule to a standard process with an event subscriber in AL, the programming language of Business Central, and cover the rule with a test codeunit | Uses an event instead of copying standard code. Tests the rule with a positive and a negative case. Hands in code analyzer output with no warnings, or with a reason for each warning that remains. |
| Finance and operations developer | Design an extension in X++, the programming language of finance and operations apps such as Microsoft Dynamics 365 Finance and Supply Chain Management. A design on paper is enough. | Wraps the standard method with Chain of Command and calls next, so Microsoft's logic still runs. Spots methods that cannot be wrapped, such as those marked [Hookable(false)] and final methods without [Wrappable(true)]. Plans a unit test in SysTest, the X++ test framework. |
| CRM consultant, including Sales consultants | Qualify a demonstration lead into an opportunity, then write a user acceptance testing (UAT) script for that step. For a Customer Service consultant, use a case that must reach the right queue instead. | Predicts which account, contact and opportunity records the qualification settings create, and whether users see a side pane first. Checks how duplicates are caught: the default rules cover accounts and contacts, so leads need a rule of their own. Adds a negative test to the script. |
| Business Central consultant | Trace a posted demonstration sales invoice to its general ledger (G/L) entries, then reconcile inventory with the G/L | Names the setup that chose each account, and checks the trace with Find entries. Gives every inventory difference a cause, such as cost that is not yet posted to the G/L. |
| Power Platform developer, Power Apps developer or Power Automate developer | Diagnose a demonstration cloud flow whose run failed halfway, and the canvas app that lists the records the flow creates | Finds the failed action and its cause. Adds error handling with Run after settings, Try and Catch scopes and a retry policy. Plans a rerun that creates no duplicates. Notices a filter that cannot be delegated, so the app reads only the first 500 records by default. |
| Integration engineer | Map one interface between Microsoft Dynamics 365 and another system: the records, the direction, the timing and the failure path | Names the system of record for each record type and the keys that match records on both sides, so a retry updates a record instead of creating a duplicate. Waits for the time in the Retry-After header when Dataverse returns 429 Too Many Requests. Names who owns a failed message and how it is sent again. |
| Data migration engineer | Read the error report of a trial load, then reconcile the loaded data with the source extract | Reconciles record counts and totals per table, and gives each error a cause and an owner. In Dataverse, keeps source created dates by mapping them to overriddencreatedon, and knows which other audit columns an import cannot set. |
| Solution architect | Critique a demonstration design, and write each finding as an entry in an architecture decision record (ADR), the log of design decisions | Each entry states the context, the options considered, the decision with its tradeoffs, a confidence level and a status. A changed decision gets a new entry that supersedes the old one, and accepted entries are not edited. |
This task bank is public, so candidates can read it too. Before you use a task, change the scenario, the data and one condition.
A role in this table is not an offer to supply it. The hiring hub lists the roles that MS Dynamics Guys offers.
Communication checks
Communication decides how much of your lead's time a specialist will need. Build these checks into the task, and score them under the Communication criterion.
| Check | How to run it | What good looks like |
|---|---|---|
| Questions first | Leave one input out of the task on purpose, such as the product version or one business rule | The candidate asks for the missing input before building, or writes down an assumption and builds on it |
| Written assumptions | Ask for the assumptions in writing, together with the result | Each assumption is specific enough for your lead to confirm or reject |
| A failure explained plainly | Ask the candidate to explain one failure case to a business user | Short sentences, every abbreviation explained, and a clear next step for the user |
| Your hours, confirmed in writing | Ask the supplier to confirm which of your working hours the person will cover | A written answer before any start date, not a promise made in a call |
Documentation checks: what the candidate hands in
Ask every candidate for the same five documents with the task. Short documents are fine. Score them under the Documentation criterion with one question: could another person continue the work from them?
- A design note: what was built, why, and which options were rejected.
- A change description that a reviewer can follow without the candidate in the room.
- Test evidence: the results themselves, not a statement that the tests passed.
- Rollback steps: how to undo the change, or turn it off, if a release goes wrong.
- A handover note: what the next person needs to know to continue.
Supplier checks: the same questions for every supplier
The candidate is only half of the decision. Ask each supplier the same questions, ours included, and keep the answers with the scores.
| Ask the supplier | Why it matters | An answer to look for |
|---|---|---|
| Who assessed this person, and with which task? | A score means little without the task and the reviewer behind it | The reviewer's role and product track, the task, and the gaps found, in writing |
| Will the person we interview be the person who starts? | A substitute has not passed your task or your interview | No change of person without your written approval, and the same task for any substitute |
| Can we run our own task and decide ourselves? | Only your team can judge the fit with your people and your system | Yes, with no conditions attached |
| What happens if the person leaves? | Continuity is part of what you buy | A replacement process and a knowledge transfer plan, written into the agreement |
| Who removes access at the end? | Accounts that stay open after an exit are a security risk | Your administrator removes every account on the last day, and the supplier confirms that no copies of your data remain |
| Where will the work be stored? | Work kept outside your systems leaves with the person | Your repository and your workspace, from the first commit |
Microsoft certifications: check them, then keep the task
Candidates and suppliers often list Microsoft certifications. A certification is one piece of evidence, so check it before you give it any weight. Two checks are enough:
- Ask for a transcript link. In their Microsoft Learn profile, the candidate selects Share link to create a link to their transcript. The transcript lists the certifications and training modules they completed. Prefer the link to a PDF copy, because the link shows Microsoft's own record.
- Check that it is current. Associate, expert and specialty certifications expire after one year. The holder renews one for free by passing an online assessment on Microsoft Learn, which opens six months before the expiry date. Fundamentals certifications do not expire. An expired certification is regained only by passing the required exams again.
Calibration example: a 2 and a 4 on the same posting trace
Can we use this scorecard on specialists from MS Dynamics Guys?
Yes, and we suggest it. Give our candidates the same task and rubric as candidates from any other supplier. Our assessment and your scorecard then work together:
- Our summary comes first. Every specialist we propose has passed a role-specific assessment. You receive a written summary of what we tested, where the candidate was strong and the gaps we found. How we assess specialists describes the method.
- Your task covers the gaps. Aim your task at the criteria our summary marks as weak or untested, and at what is special about your system.
- Your systems stay closed. Our candidates do your task outside your systems. They get access only after you choose them and the agreement is signed.
- You decide. Your scores and your interview decide who joins. A pass in our assessment does not oblige you to accept anyone.
- The person you approve starts. We propose a different person only with your written approval. That person goes through the same assessment, your task and your interview before starting.
- The work stays with you. Code and solution files live in your repository, with their full history. When a specialist leaves, their knowledge transfer is written down and handed over.
Replacement terms are agreed in writing for each engagement. After you choose someone, the work follows our delivery model.
When MS Dynamics Guys fits
MS Dynamics Guys fits buyers who want to compare our specialists with other candidates on the same evidence, and who keep the final decision themselves.
MS Dynamics Guys fits when
- You will score every candidate with the same task and rubric, ours included.
- You want a written summary of our assessment before you spend interview time.
- You want the code in your repository and the hiring decision in your team.
- The role works in Microsoft Dynamics 365 Sales or Customer Service, on Dataverse.
- The role is AL development for Business Central.
- The role is functional consulting for Business Central.
- The role works in Power Apps or Power Automate.
Look elsewhere when
- You want a company to vet candidates that other suppliers propose. We assess only the specialists we propose. Score other candidates with this scorecard and your own reviewers.
- You want to recruit an employee for your payroll, or you need interview coaching, salary data or employment advice. MS Dynamics Guys supplies specialists under an agreement between two companies.
Questions about the hiring scorecard
How does the scorecard differ from our interview?
The scorecard measures one task against fixed criteria, so candidates from different suppliers are compared on the same evidence. Your interview covers what a task cannot show: how the person works with your people, your system and your rules. Run the task first. Then spend the interview on your must-haves and on the criteria that scored 2, instead of repeating the task.
Can the task use our own system or backlog?
Use your backlog, but keep your system closed. Rewrite one real backlog item with invented names, amounts and records. The candidate then works in their own sandbox or development environment, or on paper. Your accounts and real data wait until you have chosen someone and signed an agreement. Keep the task small: it should show judgment, not deliver free work for your backlog.
Do Microsoft certifications replace the task?
No. A certification shows that the person passed Microsoft's exam for a product area. It does not show how they work with your version, your extensions and your data. If your contract or policy requires a certification, make it a must-have and check it before the task, as described above. Then let the task and your interview decide.
How do we score a consultant who writes no code?
Use the same six criteria, and read them for configuration work:
- Task result: the configuration works in the sandbox, and a test script that a business user can run proves it.
- Failure handling: the consultant plans what the process does with wrong input, such as a missing required value, and adds a test that proves the rejection.
- Configuration first: standard features come first, and every step that needs code is marked with a reason.
- Documentation: a setup record lists each setting, the reason for it and who approved it.
Product and version fit, and communication, read the same way for every role.
Should we add the scores into one total?
We publish no standard weights, because the right weights depend on your role and your backlog. A total can also hide a weak must-have behind strong scores elsewhere. Read the scores criterion by criterion: the must-haves first, then the criteria your next work items depend on. If you do use weights, set them before the first task and keep them the same for every candidate.
What if a candidate scores 2 on a must-have?
Only a 1 ends the process automatically. Before the first task, decide for each must-have whether a 2 is acceptable. If it is, plan support: pair the person with your reviewer for the first items, and score that criterion again on the first accepted work. If it is not, treat the 2 like a 1, and apply that rule to every candidate.
Discuss a technical assessment for your role
Tell us the Microsoft Dynamics 365 app and version, the role, and the task you plan to use. We reply with what our assessment covers for that role and what your own task should add.
Product documentation referenced on this page
- Handle exceptions in plug-ins Microsoft Learn, checked October 5, 2026
- Use plug-ins to extend business processes Microsoft Learn, checked October 5, 2026
- Extensibility overview (Business Central) Microsoft Learn, checked October 5, 2026
- Testing the application overview (Business Central) Microsoft Learn, checked October 5, 2026
- Using the code analysis tool (Business Central) Microsoft Learn, checked October 5, 2026
- X++ language reference Microsoft Learn, checked October 4, 2026
- Class extension: method wrapping and Chain of Command Microsoft Learn, checked October 5, 2026
- Testing and validations (finance and operations apps) Microsoft Learn, checked October 5, 2026
- Qualify and convert a lead to opportunity Microsoft Learn, checked October 5, 2026
- Set up duplicate detection rules to keep your data clean Microsoft Learn, checked October 5, 2026
- Set up queues to manage activities and cases Microsoft Learn, checked October 4, 2026
- Posting group setup Microsoft Learn, checked October 5, 2026
- Finding related entries for documents Microsoft Learn, checked October 5, 2026
- Reconcile inventory costs with the general ledger Microsoft Learn, checked October 5, 2026
- Work with dimensions to track and analyze data Microsoft Learn, checked October 5, 2026
- Employ robust error handling (Power Automate) Microsoft Learn, checked October 5, 2026
- Understand delegation in a canvas app Microsoft Learn, checked October 5, 2026
- Service protection API limits Microsoft Learn, checked October 5, 2026
- Use Upsert to create or update a record Microsoft Learn, checked October 4, 2026
- Run data import (Dataverse) Microsoft Learn, checked October 5, 2026
- Maintain an architecture decision record (ADR) Microsoft Learn, checked October 5, 2026
- View and share your transcript Microsoft Learn, checked October 5, 2026
- Microsoft Certification renewal Microsoft Learn, checked October 5, 2026
- Renew your Microsoft Certification: FAQ Microsoft Learn, checked October 5, 2026