Web Performance Consulting for Adobe Experience Manager and Magnolia Fixed Core Web Vitals target values, checked with every release

The agency implements what it has been commissioned to do and measures its work against those requirements. If they contain no target value for performance, performance cannot be signed off either. An external role sets this target value and signs off releases against it. This is not competition for the agency but a role that is missing on the client side. In most projects, this leads to close cooperation with the agency, because clear acceptance criteria make its work easier too.

The score comes from a simulation without a tag manager, cookie consent, personalization or real devices. Google, by contrast, evaluates real user experiences over the last 28 days. Edge Delivery is a good starting point. Whether the values hold in production depends on what marketing and third parties add afterwards.

Lighthouse in the pipeline prevents major outliers in the code. It does not see tags from the tag manager, consent logic, A/B tests or differences between markets. Ongoing control adds real-user data, performance budgets per page type and an approval step for everything that reaches the page from outside the repository.

Implementation is usually done by your development team or agency. The deliverables are tickets with acceptance criteria and answers to questions, followed by sign-off against the agreed target values. Separating specification from implementation is the prerequisite for keeping the sign-off against target values independent.

No, and nobody who is not responsible for implementation should. The values result from your code, your content, your campaigns and your service providers' scripts. What is committed is the service that comes from outside: measurement based on real-user data, target values per page type, prioritized tasks with acceptance criteria, sign-off of every release against these target values, and monthly evidence of where your platform stands. A commitment to a measurement result would be a commitment about your team's work.

The collaboration follows your cadence: participation in refinements as needed, the meetings in which the team clarifies tasks for the next sprint, tickets in your system, for example Jira or Azure DevOps, and checks in the test environment. On your side, this requires a fixed point of contact and access to the test environment, real user monitoring and Search Console.

The report names affected URL groups, not the cause. The first step is to split the main content load time into server response time, resource load time and render delay, separated by page type and device. Only then is it clear whether the cause lies in delivery, in the frontend or in the scripts from the tag manager and consent banner. This breakdown is the subject of the Performance Assessment.

Google calculates the values from a rolling 28-day window. A measure that goes live today is fully visible after four weeks at the earliest. How much the values improve depends on how many of the prioritized tasks the team implements and how much the platform changes during operation. In projects with a working sprint cadence, the first effects are usually measurable after two to three releases. After that, the task is to maintain the level reached.

Reliable figures come from A/B tests: 8 percent more sales after a 31 percent reduction in main content load time, and 8.4 percent higher retail conversion per 0.1 seconds faster mobile load time. Both figures are listed with their sources in the box at the top of the page. The economic effect arises before ranking, in user abandonment behavior. As a ranking signal, the Core Web Vitals are one among many.

Page speed optimization usually means one-off technical measures on a website, often measured by the Google PageSpeed Insights score. Web performance consulting starts earlier and ends later: target values for real users, priorities for the development team, sign-off of every release and a process that keeps the values from getting worse again. The score is a diagnostic tool, not a goal, because Google does not use it for ranking.

In three places, almost always in this order: at the cache layer in front of delivery, when parameters, sessions or personalized areas bypass it; at the third-party scripts from the tag manager and consent banner that weigh on Interaction to Next Paint; and on pages where teasers, assets and personalized components load later without reserved dimensions. What this means for each system is described on the pages on AEM performance and Magnolia CMS performance.

Each service has a scope defined in writing. It depends on the number of sites and markets, the release cadence and the number of teams. After the free initial analysis, you receive a written proposal with scope and process. Ongoing control has a minimum term of three months.

The service description and the draft contract are available before the first meeting with your procurement team, as is the distinction from temporary agency work. The engagement is with a self-employed consultant on invoice, usually under a service agreement or framework agreement. Your side provides budget approval and a fixed point of contact who organizes access to the test environment, real user monitoring and Search Console.

The service is described in terms of results: defined deliverables, no attendance obligation, no duty to follow instructions, no integration into your organization. Your team is responsible for implementation; specification and sign-off against the agreed target values come from outside. This separation is also the feature that distinguishes a service contract from temporary agency work and false self-employment.

The focus is on Adobe Experience Manager and Magnolia, because that is where the longest projects and the most complex release processes are. For other systems of enterprise size, the initial analysis clarifies the individual case. For smaller websites, a provider that also handles implementation is the better choice.