Discussion starter: an agency question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.
- Agency
A developer implements your recommendation wrongly and traffic drops. Who owns the fix, and how do you contract for that?
Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.
1 reply
-
adminAdmin
Fix first, assign responsibility afterwards, and have the contract already say how both work. In the moment, the priority is restoring traffic: identify the faulty implementation, roll it back or correct it, and monitor recovery, whoever caused it. Responsibility is usually shared. The agency is responsible for recommendations that are clear, correct and testable, and for any quality checks it was contracted to do; the client and its developers are responsible for implementing them as specified and for releasing code. Contracts that define those roles, include a review step before release and limit liability to what each party controls prevent most disputes. Contracts that say nothing turn every incident into an argument.
Where things usually go wrong
- Ambiguous recommendations. “Add canonical tags to filter pages” without specifying which pages and which targets.
- No review of the implementation before it reaches production.
- Environment differences: correct on staging, broken in production because of configuration or caching.
- Unrelated changes shipped together, making the cause hard to isolate.
- No monitoring after release, so the drop is noticed weeks later.
Dividing responsibility clearly
Stage Agency Client and developers Recommendation Clear, specific, testable requirements with acceptance criteria Confirm feasibility and ask questions Implementation Answer questions during build Build as specified Pre-release review Check implementation on staging, if contracted Provide access and time for review Release Advise on timing and monitoring Deploy and own the release Post-release Monitor and flag problems, if contracted Fix or roll back promptly What to put in the contract
- Scope: the agency provides recommendations and, where agreed, reviews of implementation; it does not write or deploy code unless explicitly contracted.
- Review step: SEO-critical changes are reviewed on staging before release, with time allowed in project plans.
- Sign-off: the client signs off releases; implementation outside the specification is the client’s responsibility.
- Incident response: how quickly the agency will investigate a suspected SEO regression, and what that support costs if outside the retainer.
- Liability: limits appropriate to the fees, and no guarantee of rankings or traffic.
- Change control: changes to the specification during build are documented and agreed.
Handling an incident well
- Confirm the cause with evidence: a crawl comparison, logs and the timing of the release.
- Agree the fastest safe fix: often a rollback of the specific change.
- Monitor recovery with daily or hourly data.
- Hold a short review afterwards: what went wrong, which step would have caught it, what changes to the process.
- Discuss costs only after the site is recovering, referring to the contract.
Hypothetical example:
An agency recommends canonical tags pointing filtered category URLs to their unfiltered category. The developer implements canonicals pointing every category page, filtered or not, to the homepage, and category traffic falls within days. The agency spots it in monitoring, provides the crawl evidence, and the developer rolls back within hours. In the review, both sides agree the recommendation lacked concrete examples and the release skipped staging review. The contract is updated so SEO-critical changes require an example-based specification and an agency check on staging, billed as part of the retainer. The next canonical change ships without incident.
Watch for this:
Agencies that avoid reviewing implementations to limit their liability often end up blamed anyway, because clients assume the SEO expert checked. It is better to offer implementation review as a defined, paid service than to leave the gap open.
Monitoring shortens the damage
The difference between a minor incident and a serious one is usually how long the problem stays live. Automated checks after each release, covering robots directives, canonicals, status codes and key content on representative URLs, catch most implementation errors within hours. Daily monitoring of clicks and impressions by template, with alerts for unusual drops, catches the rest within days. Agree who receives the alerts and how quickly they respond. An agency that spots a problem the morning after release and a developer who can roll back the same day turn a potential traffic loss into a non-event.
Writing recommendations that are hard to misread
The best prevention is a recommendation a developer cannot misinterpret: the exact pages affected, the exact output expected, before-and-after examples of the HTML, edge cases, and how to test it. Acceptance criteria written as checks, such as “the canonical on /shoes/?colour=red is
https://example.com/shoes/“, can often be automated in the release pipeline. The extra effort in writing a precise specification is small compared with the cost of an incident.Fix the problem before discussing fault, share responsibility according to clearly defined roles, contract for implementation review on SEO-critical changes, write recommendations as testable specifications, and use every incident to tighten the process.
Need help with this on your own site?
Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.
Add a reply Cancel reply
Replies are open to approved members. Log in or apply to join.