Most retailers wait too long to replace a Magento development partner. The pattern is usually that frustration accumulates for six to twelve months, the team rationalizes the issues as one-off problems, and the decision to switch only happens after a major incident or a quarter of missed targets. By the time the decision gets made, the cost of staying with the wrong partner has compounded into delayed launches, performance regressions, and a backlog that no longer reflects the business priorities.
The right time to evaluate whether your partner needs replacing is when the pattern is clear but the damage is still contained. This article describes ten signals that show up in real Magento engagements that have gone wrong, drawn from rescue projects we have taken over from other agencies. Not every signal indicates an immediate switch, but the presence of three or more in combination usually does.
Sign 1: Velocity Is Trending Down Without an Obvious Cause
Healthy Magento engagements show consistent or improving velocity once the team has ramped on the codebase. Stories that took three days in month two should not be taking five days in month six unless the work has gotten substantively harder. Declining velocity without a corresponding increase in complexity usually indicates that the team composition has changed, that knowledge is being lost to turnover, or that the codebase has accumulated debt the team is now navigating around.
The diagnostic is to look at the actual sprint burndown over the last six months and see whether the trend is consistent. If velocity has dropped twenty percent or more without a clear reason, that is a structural problem with the partner relationship, not a temporary one.
Sign 2: The Senior People Named in the Proposal Are No Longer on Your Project
The architect who pitched the engagement is now on three other projects. The lead developer rotated to a new client and a less experienced developer took their place. The project manager you started with left the agency and the replacement has not built context. This pattern is one of the most common causes of engagement decline and one of the easiest to detect once you start looking.
A healthy partner protects continuity for their best clients. An unhealthy partner uses the best resources to win work and rotates them off once the contract is signed. The signal is unambiguous when you compare the names in the original proposal against the names in the current standup. If the overlap is less than fifty percent and you did not approve the changes, the agency is treating your account as second-tier.
Sign 3: Core Web Vitals Have Regressed Since Launch
Magento storefronts launched with strong CWV scores should maintain those scores through subsequent releases. CWV regressions show up when the team is shipping features without performance discipline – new third-party scripts get added without monitoring, image budgets get violated, JavaScript bundles grow without code splitting, and the cumulative effect produces slower pages over time.
The diagnostic is to pull a PageSpeed Insights report for your top five landing pages today and compare against a report from six months ago. If LCP, INP, or CLS have gotten meaningfully worse, the agency is not maintaining the performance discipline that the original launch reflected. This is particularly visible on Hyvä-themed storefronts, where the whole point of the theme is to make CWV easier – regressions there indicate that the team is not using the tools available.
Sign 4: Hyvä, B2B, or Headless Work Is Being Outsourced Without Telling You
A growing pattern in mid-market Magento engagements is that agencies subcontract specialized work to offshore partners without disclosing the arrangement. The Hyvä migration that was supposed to be done by your agency’s frontend team is actually being done by a subcontractor in another country. The Adobe Commerce B2B configuration is being handed off to a different specialist firm. The headless implementation is being executed by a partner agency.
The pattern is not inherently wrong if it is disclosed, scoped, and quality-controlled. The problem is when it happens without disclosure, which usually correlates with quality issues, communication gaps, and accountability dilution when something goes wrong. The diagnostic is to ask explicitly who is writing code in your repository this sprint. If the answer is vague, the answer is probably “someone you did not hire.”
Sign 5: Estimates Consistently Exceed Actuals by More Than Twenty Percent
The estimate-to-actual variance is the cleanest single metric for whether an engagement is healthy. Some variance is normal – discovery surfaces complexity, scope evolves, and individual stories run long. Sustained variance of more than twenty percent across multiple sprints, however, indicates that either the estimation process is broken or the execution capability is not what was sold.
The diagnostic is to look at the last ten completed work items and compare the original estimate against the actual hours. If the average is meaningfully over the estimate, the agency is either underestimating to win approval or is unable to execute at the velocity their estimates assume. Both are structural problems that compound over time.
Sign 6: Releases Require Hotfixes With High Frequency
Healthy release processes produce hotfixes occasionally – every team encounters edge cases that did not surface in QA. Unhealthy release processes produce hotfixes on most releases, which indicates that QA depth is insufficient, that the testing environment does not match production, or that the team is shipping without adequate validation.
The diagnostic is to look at the last twelve releases and count how many required a follow-up hotfix within forty-eight hours. If the count is more than three, the QA discipline is not where it needs to be for production Magento work. The fix is not necessarily firing the agency, but it is a forcing function for a real conversation about QA scope and resourcing.
Sign 7: The Roadmap Is Reactive, Not Proactive
In a healthy engagement, the agency brings you ideas you would not have generated on your own. They propose architectural improvements before they become problems, surface security patches before they are exploited, recommend module replacements before deprecation, and identify performance opportunities before competitors do. The agency is part of your strategic thinking, not just an order-taker.
In an unhealthy engagement, every work item on the roadmap originates from your team. The agency executes what you ask but does not contribute ideas, anticipate issues, or push back on directions that look wrong. The relationship has become transactional, and you are paying agency rates for execution that an in-house team could provide more cheaply.
Sign 8: Documentation Quality Is Deteriorating
Documentation is the canary in the coal mine for engineering discipline. Teams that document their work also tend to test their work, review their work, and think through their work. Teams that stop documenting are usually starting to cut other corners as well.
| Documentation Quality Signal | Healthy | Warning |
|---|---|---|
| PR descriptions | Context, decisions, test coverage | One-line summary or empty |
| Architecture decisions | ADRs maintained in repo | Decisions exist only in Slack history |
| Runbooks | Updated with each release | Last updated months or years ago |
| Onboarding docs | New developers productive in days | New developers struggle for weeks |
| Integration docs | Auth flows, data contracts, error handling | Tribal knowledge in original developer’s head |
The diagnostic is to look at the documentation produced in the last quarter and compare it against the documentation produced in the first quarter of the engagement. A declining trend is a leading indicator that other quality measures are about to decline as well.
Sign 9: Critical Knowledge Lives in One Person’s Head
Bus factor risk is a quiet killer in Magento engagements. The pattern is that one senior developer at the agency holds most of the context about your codebase, your integrations, your customizations, and your operational quirks. When that person is available, the engagement runs smoothly. When they are on vacation, sick, or rotated off, the rest of the team takes substantially longer to do anything because the context is not documented or distributed.
The diagnostic is to look at what happens when the senior person is unavailable for two weeks. If velocity drops by more than thirty percent and quality issues surface that would not have happened on their watch, you have a bus factor problem. The agency should be deliberately distributing knowledge across the team and into documentation. If they are not, the engagement is fragile in a way that will cost you when it matters most.
Sign 10: The Conversation About Money Has Become Adversarial
Healthy partnerships have honest conversations about scope and cost. Unhealthy partnerships have negotiations. The shift from one to the other usually happens gradually – change orders start producing pushback, scope conversations become defensive, status updates start hiding bad news, and the energy that used to go into solving problems now goes into managing the commercial relationship.
The diagnostic is whether you find yourself preparing for difficult conversations with the agency the way you would prepare for a vendor negotiation. If the answer is yes, the partnership has degraded into a transactional relationship, and the costs of that degradation will continue to compound until something changes.
What to Do When the Signals Are Clear
The presence of three or more signals does not automatically mean you should fire your agency tomorrow. The first step is to have a candid conversation with their senior leadership about the patterns you are seeing. A healthy agency will treat that conversation as a wake-up call and will produce a credible recovery plan. An unhealthy agency will defend, deflect, or promise improvements without changing anything material.
If the response to that conversation does not produce visible change within sixty days, the right move is to start a parallel evaluation of replacement agencies. The cost of switching is real but bounded. The cost of staying with a partner who is not serving you compounds indefinitely. Bemeir specializes in taking over inherited Magento engagements and has structured the rescue project process specifically to minimize the discovery friction and risk that retailers fear most about switching.
The rescue project workflow starts with a comprehensive technical audit, a documented thirty/sixty/ninety day plan, and an explicit knowledge transfer protocol from the outgoing agency. The structure exists because retailers who have been burned once need to see the structure that prevents it from happening again. For a deeper view of how a structured Magento technical audit reduces switching risk, and how Bemeir’s Hyvä migration practice handles performance-critical takeovers, the conversation usually starts with a sanitized look at the current codebase and a candid discussion of where the existing relationship has gone wrong.
External reference points for evaluating agency health include the Adobe Commerce Solutions Partner directory, Hyvä’s official partner program, and broader research from Forrester on commerce services providers that frames the structural patterns behind agency performance.
