Most agencies that outsource SEO worry about the wrong thing. They ask about price and turnaround. What actually goes wrong is the site: a staging robots.txt pushed to production, a redirect map with gaps, a plugin installed on a Friday that doubles page weight.
We run an SEO outsourcing team in India that works inside other agencies' client sites every week. This is the checklist we ask agencies to hold us to. It works just as well if you hire someone else, or a freelancer.
1. Give access by role, not by password
The fastest way to lose control of a client site is to email an admin login.
- Search Console: add the outside team as a Restricted user first. They can read every report. They can't remove URLs or change settings.
- Analytics: Viewer role is enough for reporting.
- CMS: a separate account per person, with Editor rights, not Administrator. On WordPress, plugin installs should need an admin.
-
Code: a fork or a branch, never direct push to
main.
When the work ends, you remove four named accounts instead of rotating a shared password across the client's whole stack.
2. All technical changes go through a pull request
Title tags, schema, redirects, internal links: if it lives in code, it goes through review like any other change.
git checkout -b seo/fix-duplicate-titles
# make the change
git commit -m "SEO: unique <title> on /services/* (21 pages)"
git push origin seo/fix-duplicate-titles
# open a PR, tag the client's developer
The PR description should say what changed, why, and how to check it. A developer who reads "added FAQ schema to 8 pages, validated in the Rich Results Test, screenshots attached" can approve it in two minutes.
3. Keep a change log the client's developers can read
Every change, one line each. A shared sheet is fine.
| Date | Page | Change | Why | Reverted? |
|---|---|---|---|---|
| 2026-11-03 | /pricing | Title 72 → 58 chars | Truncated in search | No |
| 2026-11-05 | site-wide | Removed noindex from /blog/page/* |
Pages dropping out of index | No |
When rankings move, this is the first place anyone looks. Without it, every drop turns into a guessing game between the agency, the SEO team and the developers.
4. Never let staging rules reach production
One of the most common outsourced-SEO mistakes we still see is this:
User-agent: *
Disallow: /
It's correct on staging and catastrophic on the live site. Two cheap guards:
- Serve robots.txt from code that checks the environment, not from a static file that gets copied.
- Add a check to CI that fails the build if production's robots.txt contains
Disallow: /on its own line.
curl -s https://example.com/robots.txt | grep -qx "Disallow: /" && \
echo "Production robots.txt blocks everything" && exit 1
5. Migrations need a redirect map before launch day
If a migration is in scope, the outsourced team should hand over a full URL map at least a week before launch, not on the day.
old_url,new_url,status
/services/seo-audit,/services/seo-audit-services,301
/blog/1,/blog,301
/old-landing-page,,410
Then test it against the live list of URLs that get traffic, which you can export from Search Console's Pages report. Any URL with impressions and no row in the map is a ranking you are about to lose.
6. Set a performance budget so SEO work doesn't slow the site
Schema scripts, chat widgets, tracking pixels and "SEO plugins" all add weight. Agree a budget up front and let the build enforce it. With Lighthouse CI:
{
"ci": {
"assert": {
"assertions": {
"largest-contentful-paint": ["error", { "maxNumericValue": 2500 }],
"cumulative-layout-shift": ["error", { "maxNumericValue": 0.1 }],
"total-byte-weight": ["warn", { "maxNumericValue": 1600000 }]
}
}
}
}
If an SEO change breaks the budget, it gets fixed before it ships, not after the Core Web Vitals report turns red a month later.
7. Validate structured data automatically
Schema is easy to break silently: a CMS update renames a field and every Product loses its price.
Run a structured data check on a handful of key templates after each deploy, and keep the validated JSON-LD in the repo so a diff shows exactly what changed. Google's Rich Results Test is fine for spot checks. For anything you rely on, script it.
8. Agree what "done" means for links and content
This one isn't technical, but it's where most trust breaks down.
- Links: every link is reported with the page it's on and the page it points to. No private blog networks and no link farms. If a team won't show you the list, assume you won't like it.
- Content: written by a person, edited by an SEO lead, with the target keyword and brief in the doc. Not one prompt pasted into a chatbot.
- Reports: a sample report before you sign, so you know what your client will see.
9. Paperwork that protects the agency
If you resell the work, two documents matter more than the price:
- An NDA, because the outside team will see client data.
- A non-solicit, so they never contact your clients directly.
Starting with a single client or a small project, an audit or a content batch, is the cheapest way to find out whether the process above actually happens.
The short version
- Role-based access, removed when the work ends
- Every code change through a pull request
- A change log developers can read
- Environment-aware robots.txt, checked in CI
- Redirect map a week before any migration
- A performance budget the build enforces
- Automated structured data checks
- Every link and every brief visible
- NDA and non-solicit before kickoff
None of this is specific to India or to outsourcing. It's how you'd want any team, in-house or outside, to touch a site that pays the bills.
What would you add? I'm especially curious how other teams handle robots.txt across environments.
Top comments (0)