The container, rebuilt so nothing breaks quietly.

Four things: an audit that tells you what is really firing, a rebuild on a documented data layer, consent implemented so measurement survives privacy controls, and governance so the container stays clean after we leave. The governance is the part clients underrate and later thank us for — workspaces, a second reviewer, a pre-publish checklist and version notes are what stop a container drifting back into sprawl within a year.

It is, once they are traced, and tracing is part of the work. For each tag we establish what it sends, where it sends it, whether the destination account is still active and whether anyone reports on it. Anything still in use is kept and documented. Everything else is removed in a versioned publish, so a single click restores the previous state if a stakeholder appears with a requirement nobody wrote down. In practice the risk of clearing dead tags is far smaller than the risk of leaving conversion data flowing to a platform nobody monitors.

It can, and the effect is worth measuring rather than debating. As web.dev puts it, tag managers affect Web Vitals indirectly by using up the bandwidth and main-thread CPU time a page needs to load quickly and stay responsive, and every additional vendor script compounds that. Two changes fix most of it: remove what is dead, and move what is critical server-side. In a published test, Semetis measured LCP down 23% and Total Blocking Time down 60% after switching to server-side tagging. We record your numbers before and after so the improvement is yours to verify.

Not always, and we will tell you when you do not. It earns its keep when conversion data drives bidding decisions, when tracking prevention is visibly eroding your numbers, when you need control over what leaves your infrastructure, or when page performance is a commercial problem. The mechanism is well documented: the browser sends one request per event and your container fans out the vendor calls. Against that sit a cloud bill and a maintenance responsibility, so we scope it as a decision with a price rather than a default upgrade.

By gating tags rather than blocking them, and by verifying the result instead of trusting the banner vendor. Configured properly, non-consented visits still send cookieless signals that keep aggregate reporting usable, while consented visits are measured in full. Getting this right matters more than the volume suggests, because consented users are typically 2-5x more likely to convert — so a misfiring consent setup skews the shape of your data, not just its size. We test every state across browsers and devices and document what is collected in each.

That is our preferred arrangement. We write the data layer specification, your developers implement the push events in the release cycle, and we build and QA the container side. It keeps the fragile part — reading values from the page — out of the equation, and it means your team understands the implementation rather than inheriting a black box. Where there is no in-house development capacity, we can implement in the site ourselves or work with your web agency, and we still write the specification first.

A short, named list, and everyone else works in workspaces. Most of the incidents we are called in to repair trace back to an unreviewed publish: a trigger widened, a variable renamed, a conversion duplicated. We set named users at the right permission level, remove shared logins, require a second reviewer for publishes, and insist on a version note describing the change and the reason for it. It costs a few minutes per change and saves the fortnight of forensics that follows a silent break.

You get a documented container, a change log and a handover pack, and the choice of running it yourself or keeping us on a light monthly arrangement for reviews, new tracking and monitoring. Either way the alerting stays in place, so a tag that stops firing after a deployment surfaces the same week rather than at the next quarterly review. Clients who take the documentation and run it internally are a success for us, not a loss.

Audit and clean-up is a fixed price. The rebuild — data layer specification, tag build, consent, QA and documentation — is quoted as a fixed-scope project once the audit shows what is actually needed, so you approve a number rather than an open-ended retainer. Server-side work is quoted separately with the cloud cost stated. Ongoing governance is a small monthly fee, and it can be paused when there is nothing to do.

We can, and the loop closes faster when the team reading the data is the team changing the campaigns. Our paid search and paid social teams work from the same container and definitions, and our GA4 practice sits alongside it. It is equally fine to buy the tagging work alone while other agencies buy your media: we document everything and share the specification with them, because a container only one agency can read is not much of an asset.