From findings to work
This is the part the SEO industry keeps skipping. Identifying a problem is easy and nearly free — converting it into something a specific person can pick up, complete, and have objectively verified is the actual product.
⚠ 47 issues found
• Redirect chains detected (18)
• Pages with noindex (84)
• Orphan pages (12)
• LCP needs improvement
• Broken backlinks (9)
…42 more
No owner. No sequence. No effort estimate. No definition of done. Nobody can pick this up on a Monday morning, so nobody does — and three months later it is still in the shared drive.
noindex injected at the CDN edge on /products/*. Not in application source.
Acceptance criteria
Published, like everything else. Priority deliberately rewards small high-impact jobs over large ones, because the single biggest predictor of whether SEO work gets done is whether the first item is achievable this week.
priority = (severity_weight × impact_weight × result_factor) ÷ effort_weight
severity_weight critical 4 · major 3 · minor 2 · advisory 1
impact_weight high 3 · medium 2 · low 1
result_factor fail 1.0 · advisory 0.6
effort_weight S 1 · M 2 · L 3 · XL 40.5–2 hrs
Config or single-template change.
2–8 hrs
Template work, needs a deploy and QA.
1–3 days
Multi-template or content programme.
3+ days
Architectural. Needs a scoped project.
Worked example. IX-01 is critical (4), high impact (3), failing (1.0), and effort S (1): 4 × 3 × 1.0 ÷ 1 = 12. PF-01 is major (3), high impact (3), failing (1.0), effort L (3):3 × 3 × 1.0 ÷ 3 = 3. Both matter — but the two-hour job that clears a critical fault goes first, and the three-day performance project waits.
The check code from the standard, so the job traces back to a published criterion. Never a vague description.
The specific URLs, headers, values or measurements observed. A developer should not have to reproduce the diagnosis.
What is in and what is explicitly out. This is what prevents the scope creep that poisons most SEO engagements.
Developer, SEO, Content, Design or Owner. Most stalled SEO work is sitting with the wrong person.
S, M, L or XL with an hour range. Bands rather than false precision — anyone quoting exact hours before seeing the codebase is guessing.
The conditions that must be objectively true for the job to be done. These are the recheck conditions, written before the work starts.
A range based on observed market rates for that work type and effort band. Not a quote, and labelled as an estimate.
Exactly which check re-runs and what it must return. The recheck is not a matter of opinion.
Generated from the sample inspection. The board view is how the work is tracked from open through to verified.
| # | Check | Job | Owner | Effort | Est. | Pri |
|---|---|---|---|---|---|---|
| 1 | IX-01 | No accidental noindex on indexable pages failDeveloper | Developer | S | NZ$80–240 | 12 |
| 2 | AR-01 | No orphaned pages failSEO | SEO | M | NZ$240–800 | 4.5 |
| 3 | AU-02 | No link equity lost to broken or redirected inbound links failSEO | SEO | S | NZ$120–480 | 4 |
| 4 | AI-01 | Answer-engine crawlers are not blocked unintentionally failOwner | Owner | S | NZ$120–400 | 4 |
| 5 | PF-01 | Largest Contentful Paint within threshold failDeveloper | Developer | L | NZ$800–3,200 | 3 |
| 6 | CR-02 | Server responds reliably under crawl load advisoryDeveloper | Developer | M | NZ$240–800 | 2.7 |
| 7 | IX-02 | Canonical tags are self-referencing and consistent advisoryDeveloper | Developer | M | NZ$240–640 | 2.7 |
| 8 | IX-03 | Important pages are actually indexed advisorySEO | SEO | M | NZ$320–960 | 2.7 |
| 9 | OP-01 | Title tags are unique, present and descriptive advisoryContent | Content | M | NZ$240–960 | 2.7 |
| 10 | CR-03 | No redirect chains, loops or soft-404 redirects failDeveloper | Developer | M | NZ$240–600 | 2 |
| 11 | AR-03 | Internal linking supports priority pages advisorySEO | SEO | M | NZ$240–960 | 1.8 |
| 12 | AI-02 | Primary content is available without JavaScript execution advisoryDeveloper | Developer | L | NZ$800–3,200 | 1.8 |
Diagnosed and scoped. Waiting on a decision: do it yourself or assign it.
Accepted by your team or a matched specialist. Acceptance criteria agreed up front.
Work underway. Scope is locked — a change needs your explicit approval.
Recheck re-ran the failing check and it passed. This is the only definition of done.
Verified is the only status that counts. A job is not done because someone said it was done. It is done because the check that failed now passes — and rechecks are free within 28 days, so there is no cost to proving it.
The estimate range is derived from observed market rates for that work type and effort band, published per check in the standard. A specialist bidding through SEOWof can quote inside, below or above the band — but if they quote outside it, the reason is shown to you.
They can request a rescope with a reason before starting. You approve or decline it. Nobody can escalate cost mid-job — that is the failure mode this whole structure exists to prevent.
The recheck does. Acceptance criteria are written into the work order before anyone starts, and they are derived from the check's pass criterion. It is not a judgement call and it is not an argument.
No. Every work order includes the full remedy steps so your own team can action it. Assigning it out is a convenience, not a requirement, and the inspection result is identical either way.