You’re a procurement officer with a deputy minister waiting on a recommendation and an RFP timeline that didn’t leave room for forty interview questions per shortlisted vendor. You get twenty minutes on a discovery call with each of three consultants. Which eight questions do you ask? The answer matters more than the volume of the conversation. The eighty-question template comforts the file. Eight load-bearing questions tell you which vendor can actually deliver.
Worth naming: the procurement officers who get the most from this format are not the ones who have already decided. They’re the ones who arrive with two vendors they genuinely can’t separate on paper. A discovery call is not a demo. It’s a structured diagnostic, and the question that usually settles the comparison is the one about what went wrong on the last similar project, not what will go right on this one.
This is for procurement officers, contracting authorities, and senior buyers running a public-sector WordPress engagement with real signoff exposure. The regulatory examples below are Canadian because that’s where most of my engagements run, and those are the regimes I can speak about with evidence. The questions themselves travel, every public-sector procurement regime I’ve worked alongside has a version of the same conversation, just with different statutes and different officer titles.
Why eight, and why not eighty
A long evaluation checklist is comfort. It feels procurement-grade because it covers every dimension a deputy minister might later ask about. It’s also the document that gets ignored on the actual discovery call, because nobody can answer eighty questions in twenty minutes and most of them aren’t load-bearing anyway. The conversation that produces a defensible vendor choice is not the same conversation that produces a thick file.
If you are evaluating options, I offer WordPress newspaper development for Canadian publishers services across Canada.
Eight questions, chosen carefully, cover the capability dimensions that actually predict whether a WordPress vendor can deliver in a Canadian public-sector context, architecture, accessibility, governance, training, hosting, data residency, ongoing support, and procurement readiness. Each one is designed so the answer to it can’t be substituted for the answer to another. If a vendor gives the same answer to questions two and four, that’s the signal, not the content of either answer.
I’ve sat in the room on both sides of these conversations. I’ve been the vendor across the table from an eighty-question template, the kind that asks for CFTA Chapter 504 conformance and Schedule III handling in line one but never gets around to what happens when the comms team publishes the wrong post. I’ve also sat beside the procurement officer afterward, watching the same shortlist produce very different recommendations depending on which eight questions made the cut and which forty got skipped. The eight below are the ones that have separated vendors who actually deliver for federal departments, provincial agencies, school boards, and municipal teams from vendors who have a sales deck claiming they could.
Eight questions to ask a WordPress vendor
- What is your reference engagement at our buyer profile? Not “federal departments” or “school boards.” A specific department, a specific contract band, a named scope, and a contact procurement can call. Vendors who can’t name one haven’t delivered at your profile yet, that may still be acceptable, but only if the engagement is structured for it.
- How do you certify accessibility on delivery? The useful answers reference a named standard (WCAG 2.2 AA, AODA, the Accessible Canada Act regulations), name the tools (axe, WAVE, manual keyboard audit, screen-reader walkthrough on NVDA), and describe the signoff document the vendor produces. “We follow WCAG” is a starting point, not an answer.
- What is your editorial-workflow default? Roles, capabilities, who can publish without review, who reviews, how scheduled posts behave, how the comms team recovers from a bad publish. The Publishing Teams piece covers what a real editorial workflow looks like in WordPress.
- How do you handle data residency? Hosting provider, region, where backups live, where staging data lives, where logs are retained, what happens to personal data in form submissions. The Treasury Board cloud-usage profile and Schedule III of the Privacy Act don’t get answered with “we use Amazon Web Services.”
- What is your training model for staff turnover? Public-sector teams turn over more than vendors expect. The useful answer includes a recorded library, a written operator reference, and a re-training trigger (new comms hire, new accessibility officer, ministry restructure). Training that doesn’t survive turnover is a year-three problem.
- What does ongoing support cost? A monthly retainer band, a response service-level agreement, what’s in scope and what isn’t, and the upgrade path when the work outgrows the retainer. A vendor who can’t price ongoing support on the call hasn’t run support before, or is hoping you won’t notice.
- How do you respond to procurement-evidence requests? Insurance certificates, security questionnaires, accessibility statements, performance test reports, the paperwork procurement needs in the file. Mature vendors have these ready and can email them within a business day. Less mature vendors treat each request as a fresh project.
- What will you refuse to do? The question most agencies don’t get asked, and the one that tells you the most. A vendor that will do anything is selling a deck, not a practice. The honest refusals reveal what the vendor knows about its own limits.
| Question | Ready What a ready vendor sounds like | Not‑yet‑ready What a not‑yet‑ready vendor sounds like |
|---|---|---|
| Your reference engagement at our buyer profile? | Names the department, the contract band, the scope, and a procurement contact you can phone. | “Many federal departments,” “several school boards”, a category standing in for a name you can call. |
| How do you certify accessibility on delivery? | Cites a standard (WCAG 2.2 AA, AODA), names the tools, and offers a redacted signoff sample. | “We follow WCAG”, a category, with no tool and no signoff document behind it. |
| What is your editorial-workflow default? | Walks through the roles, who publishes, who reviews, and how the team recovers from a bad publish. | “The standard roles handle it”, no review step, no recovery plan named. |
| How do you handle data residency? | States the host, the region, and where backups, staging data, and form submissions actually live. | Names the provider (“we use Amazon Web Services”) as if that settled where the data lives. |
| What is your training model for staff turnover? | A recorded library, a written operator reference, and a trigger that re-trains the next hire. | One handoff session. The training leaves with the person who sat through it. |
| What does ongoing support cost? | A monthly band, a named response time, what’s in scope, and the path when the work outgrows the retainer. | Pricing arrives as a hedge (“it depends on scope”) rather than a band you can defend. |
| How do you respond to procurement-evidence requests? | Insurance, security questionnaire, and accessibility statement on file, emailed within a business day. | “We can put that together”, each request handled as a fresh project. |
| What will you refuse to do? | A specific refusal: “I won’t ship without an editorial-workflow signoff from your comms lead.” | Flexibility where a limit belongs, “we work with whatever the client needs.” |
The two columns are encoded in words and a glyph, not by colour alone: Ready names something specific you can check; Not‑yet‑ready offers a category in place of a name. Listen for which one you are hearing.
The signal the table can’t show you
The table above scores each question on its own. The harder signal lives between the questions, and you only catch it live. Listen to how the eight answers relate to each other. A vendor with a practice draws on a different filing cabinet for each one: the accessibility answer reaches for a signoff sample, the support answer for a retainer band, the procurement-evidence answer for a security questionnaire that already carries a version number, the training answer for a recorded library someone actually watched. A vendor with a deck gives you eight answers that all trace back to the same paragraph, reworded. The accessibility answer and the support answer and the data-residency answer carry the same shape and the same easy “we handle that,” with the same nothing behind it. That sameness is the tell, and no single row in the table can surface it. You hear it only by holding all eight answers in your head at once.
Here’s the test I use on myself before a discovery call: for each of the eight, can I name the actual document I’d email afterward? The axe-core report, the AODA section 14 statement redacted from a real municipal delivery, the security questionnaire with last quarter’s date already on it. If I can name the file, I have a practice. If I’d have to build the file after the call, I have a sales deck, and so does any vendor your buyer is talking to.
What to do when the answers come back as categories
The right column of the table flags a gap; it rarely settles the decision on its own. A vendor who answers in categories rather than names has usually just not delivered at your buyer profile yet, and for a smaller engagement that can be acceptable, as long as the file says so out loud. The mistake I watch procurement officers make is hearing the hedges, deciding to proceed anyway, and then writing the contract as though they hadn’t heard them.
The work that protects the file happens after the call, in how the contract gets written. When the answers come back soft, the engagement gets reshaped to match: scope reduced to what the vendor has actually done before, a senior-oversight reviewer named in the proposal by person rather than by role, and a line in the file where the procurement officer records that they heard the gaps and accepted the risk on purpose. A full-scope engagement that proceeds as if “we can put that together” were the same answer as “here it is” is the one that lands back on the same officer’s desk a year later, with a re-tender attached.
Eight questions versus the eighty-question RFP
Hiring a vendor through a discovery call is closer to hiring a contractor for a renovation than it is to writing a thesis. You don’t ask the contractor eighty questions before letting them quote, you ask the few that tell you whether they’ll actually show up, whether they’ve done a job like yours, whether they’ll leave the site tidy, and whether they’ll be honest when something goes wrong. The eighty-question version produces a binder. The eight-question version produces a contractor.
The helpful-neighbour test applies here the same way it applies to tools and trades. A neighbour who answers your questions plainly, names their limits, and points you somewhere better when they aren’t the right hire is the neighbour you keep calling. A neighbour who has the same answer to every question and pivots every conversation back to a sales close isn’t a neighbour at all, they’re a salesperson with a yard.
What it costs when the wrong questions get asked
The cost of the call itself is twenty minutes per shortlisted vendor. The cost of skipping it, or of running it as a checklist exercise that misses the load-bearing questions, is whatever the wrong proposal turns into on contact with the actual project. On the public-sector engagements I’m brought into to clean up, that figure lands somewhere between eighty thousand and two hundred thousand dollars: re-tender, rebuild, or fixing what shouldn’t have been delivered in the first place. Those are the lines that show up in the briefing note when the file comes back for review, which is usually how I find out about them.
For the underlying cost math, and for the in-house developer alternative every procurement officer needs to weigh against the consultant route, the cost comparison piece walks the numbers.
How the eight fit Canadian public-sector procurement
The procurement framework these eight questions sit inside, in Canada, is the Canadian Free Trade Agreement Chapter 504 at the federal level, together with the Treasury Board Directive on Service and Digital, the Accessible Canada Act, and the relevant departmental procurement instruments. Provincially, the broader-public-sector directives in Ontario, British Columbia, Alberta, Quebec, and the Atlantic provinces set their own thresholds and evidence requirements. Each regime decides what shape of engagement is even available at a given dollar band; the eight questions decide whether the procured engagement actually delivers.
Three of the eight questions tie directly to standing Canadian-specific obligations, and each has a companion piece in this practice:
- Question two intersects with AODA in Ontario and the Accessible Canada Act federally, the AODA audit a federal vendor actually asks for goes deeper.
- Question four ties to the Treasury Board cloud-usage profile and Schedule III privacy obligations, Canadian data residency for WordPress hosts covers what each major host actually delivers.
- Question seven sits on top of departmental procurement-evidence retention, and the 2026 site audit checklist shows what evidence procurement should expect a vendor to produce on request.
The principles travel outside Canada. The thresholds change, the documents have different names, the offices have different titles. The questions are the same.
Product names referenced on this page — including WordPress — are trademarks or registered trademarks of their respective owners. Training offered here is independent and is not affiliated with, endorsed by, or sponsored by any of these companies.
