Magento Agency Discovery Call: 18 Questions That Reveal Whether They Can Actually Build

The first 45 minutes you spend with a prospective Magento agency will tell you almost everything. Not because they will reveal their hourly rate or staffing model, but because a real builder answers questions differently than a sales engineer does. A real builder reaches for a story. They mention a specific extension. They tell you what they had to undo last quarter. A sales engineer reaches for slide 4.

This is the gap most in-house engineering leaders are trying to close when they shop for help on Adobe Commerce. The market has matured into roughly three camps: practitioners who write Magento code on a Monday, hybrid shops where senior architects sell while junior teams deliver, and offshore body shops who staff against any keyword you bring. The discovery call is your single best chance to figure out which one is sitting across from you before you sign anything.

Bemeir runs a lot of these calls from the other side of the table, and the pattern is consistent. Founders who walk in with 18 sharp questions cut their evaluation cycle in half. Those who let the agency drive the conversation usually end up rebuilding their first attempt 18 months later. The questions below are the ones we wish every CTO asked us, because the agencies that cannot answer them clearly should not be touching your codebase.

The four buckets that matter

Strong discovery questions cluster into four categories. Architecture and platform depth tells you whether they understand Magento as a system, not just as a product. Delivery operations tells you how work actually gets shipped. Team composition tells you who you will hire if you hire them. Commercial structure tells you what failure looks like and who carries the cost.

Walk into the call with a printed list. Take notes. Score answers on a 1 to 5 scale during the call, not after, because memory blurs by Friday. The point is not to interrogate. It is to make sure both sides leave with a shared, accurate picture of what a partnership would actually look like.

Architecture and platform depth

The first six questions test whether the agency thinks in Adobe Commerce primitives or in generic marketing-agency abstractions. A team that builds Magento for a living will answer in module names, indexer behavior, and topology decisions. A team that resells Magento will answer in benefits and case studies.

1. Walk me through the last full-stack performance regression you fixed on Adobe Commerce and what the root cause was.
2. How do you decide between a custom module, a third-party extension, and a Hyvä-native rebuild for new functionality?
3. What does your default development environment look like for Magento Cloud versus self-hosted infrastructure?
4. Which indexers do you most often see misconfigured on inherited projects, and how do you stabilize them?
5. How do you handle GraphQL schema extension when the storefront is decoupled?
6. What is your perspective on Adobe Commerce 2.4.7-p3 specifically, and what changed that we should care about?

If the answers feel rehearsed, ask a follow-up that goes one layer deeper. Real practitioners get more specific when pushed. Sales-led teams pivot to a different topic. Magento expertise is not something you can fake in real time, which is why this section is the highest signal portion of the call. Our team at Bemeir’s Magento development practice builds Adobe Commerce stores as our primary craft, and we expect every agency you talk to should be able to hold this kind of conversation without flinching.

Delivery operations

The next four questions reveal whether the agency has a system or improvises. Adobe Commerce projects fail more often from broken delivery than from broken code. You are essentially auditing their ability to coordinate.

1. What is your standard sprint cadence, and how do you handle scope changes mid-sprint?
2. Walk me through your code review and quality assurance process from commit to deploy to production.
3. What is your incident response posture if production goes down at 2am Eastern on a Saturday?
4. How do you document architectural decisions so we are not held hostage by tribal knowledge later?

The best answers here are boring. You want to hear about written runbooks, branching strategies, peer review thresholds, and on-call rotations. Excitement and improvisation are red flags. Adobe Commerce stores making 30 million dollars in annual revenue need predictable delivery, not heroics. The official Adobe Commerce DevDocs deployment guide gives you a useful reference point for what mature shops should be doing as table stakes.

Team composition

These four questions force the agency to be specific about who you will work with, not who they wish you would work with.

1. Who specifically would be the technical lead on our account, and how many other clients does that person carry?
2. What is the ratio of senior to mid-level to junior developers on a typical Adobe Commerce engagement?
3. What is your annual attrition rate on the engineering team, and what do you do about it?
4. Can we meet the proposed lead developer on a second call before we sign?

The fourth one is the killer. Agencies that cannot produce the actual lead developer in front of a prospect are usually planning to bait-and-switch you. We have rebuilt enough projects to know how this story ends. Our internal staffing rule at Bemeir is that the person who pitches you is one of the people who builds with you, because anything else creates a knowledge transfer cliff that costs you weeks.

Commercial structure and exit

The final four questions are about money, risk, and what happens if it does not work. Founders skip these because they feel awkward in early conversations. Skipping them is how you end up locked into a three-year retainer with an underperforming partner.

1. What does your standard contract say about IP ownership of custom code we pay you to write?
2. What is your default termination clause, and how does it handle in-flight work?
3. If we wanted to bring some engineering in-house in 12 months, would your contract help or hinder that?
4. What is the smallest engagement you take on, and what is the largest? Why those limits?

The last question is sneaky. It tells you their actual sweet spot, which is rarely the one on the homepage. An agency that says they take on 25,000 dollar projects and 2.5 million dollar projects is lying to one of you. Find out where they really win. The Federal Trade Commission’s guidance on B2B contracts is also worth a glance before you sign anything substantial.

How to score the call

Use a simple matrix. The table below is the version we recommend in-house teams use after a discovery call so that the evaluation is comparable across three or four agencies.

A composite score under 3.5 is a soft no. Between 3.5 and 4.2 is worth a second technical conversation. Above 4.2 is a real candidate. We have used variations of this rubric at Bemeir’s eCommerce success consultancy when running competitive RFPs for clients, and the pattern is consistent: agencies that score under 3.5 in the first 45 minutes rarely improve in proposal stage.

What to do with the results

Run this exact list with three agencies, not one. The comparison is what generates insight. A single discovery call always feels good in the moment because the agency is on their best behavior. Three calls back to back surface the differences in depth, candor, and operational maturity that matter once the contract is signed and the work starts.

If you are evaluating across platforms as well, run the same architecture-depth questions against agencies on Shopify Plus, Shopware, and BigCommerce so you can compare apples to apples. The same person who cannot articulate Adobe Commerce 2.4.7-p3 changes will often struggle to articulate the equivalent on any other platform too. Depth of practice travels across stacks.

The discovery call is not a formality. It is the first deliverable of the partnership, and it is the only one you can evaluate before you wire money. Treat it that way and you will avoid most of the war stories that founders end up telling at the end of bad engagements. Ask the 18 questions, score the answers, take real notes, and let the third call tell you what the first one was hiding.