Not every bad quarter with an agency means you should fire them. Magento engagements are complex, dependencies are real, and even strong teams have rough patches. The harder skill is distinguishing the temporary slump from the structural problem. The nine signals below are pattern markers, not isolated complaints; when three or more appear together over two quarters, the relationship is almost always unrecoverable through better account management.
Bemeir’s Magento team has inherited enough mid-market stores from struggling prior agencies to recognize how these patterns show up in practice. The version of each signal you see in a real engagement is usually subtler than the textbook description, which is part of why merchants miss them for so long.
1. Estimates that keep getting longer for similar work
The clearest signal of capacity decline is when the same caliber of work that took two sprints a year ago now takes five sprints. The agency will have a specific reason for each extension, often a real one, but the pattern across tickets is what matters. If your six-month moving average of estimate-to-actual ratio is creeping up, the team has either lost senior bandwidth or no longer has the codebase knowledge to move efficiently.
2. Patches and dependencies falling behind
Adobe Commerce security patches ship on a predictable cadence. Magento Open Source has its own patch flow. If your store is more than two patch cycles behind, your agency is either understaffed for the workload or not treating security as a priority. Either is grounds for replacement. The same logic applies to PHP versions, MySQL/MariaDB versions, and third-party module versions; a store that is three years stale across the dependency surface is one exploit away from a real incident.
3. Persistent quality regression
New features break existing features. Bug fixes introduce new bugs. The number of “we found this regression in production” tickets is going up quarter over quarter. This is not the same as having a bad sprint; it is the codebase trending in the wrong direction. The structural cause is usually that the team has lost the engineer who knew the architecture, or that testing discipline has eroded.
4. Operational opacity
You cannot get clear answers to basic questions: what is the patch level, what is our Core Web Vitals status, what is the open security exposure, what would it cost to add this feature. The agency answers in generalities, deflects, or provides numbers that change between conversations. Opacity is a competence signal but also a trust signal; agencies that know their numbers share them.
5. Communication breakdown across multiple cycles
You have raised concerns formally. You have asked for changes in process, team, or approach. The response has been verbal acknowledgment without behavioral change. This is the most common pattern in failing engagements and the easiest to ignore because the agency is still pleasant in meetings. The signal is the gap between what was agreed and what changed.
6. The senior engineers you signed for are gone
Most Magento agency engagements are sold by senior engineers and delivered by junior or offshore teams. This is not inherently a problem if the senior bandwidth is there for architecture review, code review, and escalation. It becomes a problem when the senior staff has left or been reassigned and nobody mentioned it. Ask who is reviewing the code that ships to your store, by name, and what their actual involvement looks like in any given week.
7. Roadmap velocity declining without explanation
Velocity is allowed to slow when scope expands. Velocity slowing while scope holds steady is the signal. Many merchants do not measure velocity rigorously and only notice the decline when a feature that should have shipped in Q1 is still in flight in Q3. A simple sprint-burndown over the last 18 months, plotted against story points or completed tickets, will surface the trend.
8. The agency cannot explain its own architecture decisions
When you ask why a custom module was structured a particular way, the answer should be defensible: “we chose plugin-based extension because the upstream module changes frequently and we wanted upgrade safety.” If the answer is “that is how it was done” or “the original developer no longer works here,” you have inherited an architectural orphan. One orphan is normal in a complex codebase; multiple orphans signal a team without architectural ownership.
9. Pricing pressure replaces strategic conversations
The early phase of an agency relationship is dominated by strategic conversations: where is the business going, what does the platform need to do for it, what investments make sense. As an agency loses senior staff or focus, those conversations are replaced by pricing conversations: scope changes, change orders, hourly tracking disputes. The shift in conversation type usually predates the visible problems by six months.
How the signals map to action
| Signals present | Likely state | Recommended action |
|---|---|---|
| 1 signal | Bad quarter or specific situation | Raise it formally, give a defined improvement window |
| 2 signals | Stress on the relationship | Document concerns, set explicit milestones, monitor for 60 days |
| 3 signals | Structural problem | Begin quiet evaluation of alternatives; do not commit to new scope |
| 4+ signals | Unrecoverable | Plan transition; the cost of staying exceeds the cost of leaving |
| Signal 2 alone (patches behind) | Security exposure | Treat as urgent regardless of other signals; this is a board-reportable risk |
What a healthy alternative looks like
When merchants ask Bemeir what to look for in a replacement, the answer is the inverse of these signals. A healthy Magento agency relationship shows: estimates that hold their line over time, patches applied on a documented cadence, regressions trending down, named senior engineers reviewing every code path that ships, communication that closes loops, and architecture decisions the team can defend in writing.
You do not need every healthy signal to be present on day one. You need enough of them to be visible after the first quarter that the relationship is moving in the right direction, not the wrong one. The Adobe Commerce technical resources and community resources make it easy for any reasonable agency to maintain that posture; the agencies that fail to do so are not failing because the information is unavailable.
When the signals appear, what comes next
A merchant who has recognized three or more of these signals has a 60-day window to act before the situation gets meaningfully worse. The right sequence is to (1) commission an honest pre-replacement audit of the codebase, (2) get two or three credible agencies to bid against that audit, (3) plan a transition that minimizes operational risk, and (4) execute the change before the technical debt grows another six months of compounding cost.
The hardest part of this work is emotional rather than analytical. Firing an agency feels harsh even when the data justifies it. But the merchants who delay are the ones who pay the highest cost; the merchants who act on the signals at the right time are the ones who can hand the next team a recoverable codebase. The signals exist for a reason, and the cost of ignoring them is the part of the engagement nobody bills you for.
