Blog / Web management /

How I estimate SEO work for a client

A client asks how long an SEO task will take. The tempting answer is a neat number, and I’ve learned that a neat number is usually a trap. The useful answer starts earlier: what does “done” mean, what access do I have, and which dependencies could change the work?

Here’s how I think about estimating, as someone who handles client sites at Nivonix.

Key takeaways

  • Estimate tasks, never rankings.
  • State assumptions, exclusions and what could expand the scope.
  • Count review and hand-off time, not only the hands-on work.
  • Revisit the estimate when new evidence arrives.

Break the request into stages

I split most requests into discovery, implementation, review and reporting. They behave differently. Discovery depends on access and how tidy the site is. Implementation depends on who can touch the code. Review depends on other people’s calendars.

Then I write down assumptions, exclusions and anything that could expand the scope. “Assumes admin access to the CMS and DNS” is one line, and it saves a lot of awkward conversations.

An estimate should never promise rankings. Search doesn’t change on a date I choose. I estimate work, and I say plainly what the work is.

Work from an example

Say a technical review turns up a template issue that affects hundreds of URLs. The first estimate should cover the audit itself. Implementation is a separate decision once the evidence is on the table, because the fix might be an hour in the theme or a week of developer time.

Pricing that second part before you’ve seen the problem is guessing. I’d rather say “the audit takes this long, and I’ll estimate the fix once I know what it is” and be honest than give one big number and hope.

Separate what I control from what I don’t

A useful estimate leaves room for uncertainty without becoming a blank cheque. I separate work I control from external dependencies such as access, approvals or a developer release.

I also describe the deliverable in plain language: a list of findings, an implemented fix, or a check after release. Those are three different tasks and they take different amounts of time. Clients tend to hear “SEO fix” and assume all three are included.

Count the review time

One detail often missing from estimates is review time. A draft recommendation may need an engineer, a copywriter or the client’s owner to assess it, and those hand-offs affect delivery. Say who is responsible for each part and what information is needed before it can begin. That makes the estimate easier to revise when the task turns out larger than expected.

Write a scope note

A scope note is part of a good estimate. Write down whether keyword research, writing, implementation and QA are included, and whether you or the client supplies access and approvals. If the work depends on another person, call that out before giving a delivery date.

When new facts arrive, update the scope and explain why. Agree the next step before the work expands, not after.

Check the evidence

Use the tools that answer the question in front of you. Search Console shows queries, indexing and landing pages. Analytics shows what visitors do after arrival. Neither explains every cause by itself, so compare like with like and note the date, the page group and the change you’re evaluating.

Common traps

  • Treating a tool recommendation as an instruction without understanding the page.
  • Assuming one observation applies to every URL.
  • Expecting an SEO edit to solve a product, policy or usability problem.

Each of these wastes time. Go back to the real question and verify what actually changed.

Keep it repeatable

After a redesign, content update or platform change, revisit the important pages and keep a short record of decisions. That record makes the next estimate far more accurate than my memory ever is.

Think about the last SEO job you quoted. What did “done” mean in that quote, and did the client agree? If you can’t answer both, start there.