For consultants, in-house teams, and agencies

Connect a site. Get vetted SEO work items every week.

SEO Ops reads a site's search data, the live pages and your brand brief together, then files five to nine work items a week on a board — each one deduped against everything filed before, vetted by a second pass, and carrying the evidence it was built from. Scout™ does the thinking. You do the work — or your agent does.

Point it at your own site $10/mo per workspace →

Start on a site you know well — you can judge every recommendation yourself, and you do not need a client's permission to try it. Free trial, no card, and you will know within one run whether the work is any good.

You can watch it work — a run, stage by stage

Reading search data

1,284 queries · 90d

Organizing signals

4 segments

Scanning your site

rendered · captured

Checking Brand Profile

7 / 11 signals

Scout is thinking

the analysis

Building work items

3 new · 2 merged

Verifying

median 82 · 1 held

Delivering

board · webhook · digest

00:00

What you get

This is one card. You get five to nine a week.

Shaped exactly like a card from a real run. Type before words, numbers before prose: what kind of work, which page, how big, how sure — in a glance. Open it for the argument.

⌗Schema/guides/trail-blazes/

primary

FAQPage schema + answer the question in the first 40 words

Ranks, no clicks

5.1position

1.7kimpressions

0clicks

0%ctr

seen ×2

Why it is on the list: Top-6 for two high-volume trail-marking queries with zero clicks — the clearest SERP-feature-suppression shape in the data. A featured snippet is taking the click above position 5. Add FAQPage schema and open the page by answering “what do trail blazes mean”.
Done looks like: CTR ≥ 3%within 30 days of shippingday 7 · 30check-inspos < 8by day 30 → revert, check cannibalization
Evidence: query:trail blaze meaningspos 5.1 · 1,707 impr · 0 clicksquery:trail marker colorspos 6.4 · 1,136 impr · 0 clicksbrand:focus“trail navigation guides” is a stated focus phrase
History: Filed twice — a later run reached the same conclusion and bumped this card instead of duplicating it.

Pricing

Priced per workspace, not per seat.

SEO Ops is $49 a month per workspace — one price whether you are one consultant or a twelve-person agency. While we are onboarding pilot partners it is $10, and we will tell you before that changes; you will not find out from an invoice. Core plans elsewhere in this category start at $103–$129 a month.

Trial

Free

no card · for anyone checking the work on their own site

Full product while your trial is active
Start on your own site — no client access needed
Self-serve signup, no call required
Fine for one person and one site

Early bird

Workspace

Regular price $49 Early-bird price $10/mo

per workspace · for consultants, in-house teams and small agencies

One property included
$8/mo per additional property
2 runs per property per week included
Need more? Buy extra runs any time
API keys, signed webhooks, export
Client digests from your own brand

Scale

Talk to us

for agencies at volume · many properties

Volume property pricing
Higher run allowance
White-label digests and reports
Direct line for onboarding

Runs reset Monday, and unused ones do not roll over. Nothing is metered per seat or per user, and there is no charge for a run that fails. Included allowances are deliberately conservative while we measure real costs — buy extra runs whenever a property needs them.

Why the output is worth acting on

Scout reads three things at once. You only author one of them.

Search data alone produces generic advice. So does a crawl. Read together, with your brand's own positioning folded in, they produce work a senior SEO would recognise — in the time it takes to make coffee.

Given

What the market is actually asking for

Ninety days of query data, carved into four buckets — page-two positions, low click-through, protected rankings, and impressions earning nothing — before the model reads a single row.

Observed

What you actually shipped

The rendered page, not what the CMS claims. Scout escalates until something renders and reports which method worked, then keeps a capture of the homepage as it saw it.

Authored

What you are trying to become

Goals, focus phrases, markets, positioning, competitors. The only input a human writes — which makes it the one dial you can turn to sharpen every future run.

Which is how one person covers what used to need a team.

How you run it — shown, not described

Runs in the console. Everything on it is also the API.

Every tab is a real result next to the exact thing that produced it. Copy the block on the right, get the thing on the left.

Work items brindlewhaite.example

Move a card — or tab to one and press Space to pick it up, ← → to change column, Space or Enter to drop.

Backlog

In progress

On the wire

Move a card to see the request it makes.

What the drop sends

PATCH/v1/items/wi_3f9c1a2b

Headers

Authorization	Bearer $SEO_OPS_PARTNER_KEY
Content-Type	application/json

Body

{"status": "in_progress"}

200 OK84 ms

{
  "id": "wi_3f9c1a2b",
  "status": "in_progress",
  "seen_count": 3,
  "last_seen_at": "2026-08-21T07:40:03Z"
}

curl -X PATCH \
  -H "Authorization: Bearer $SEO_OPS_PARTNER_KEY" \
  -H "Content-Type: application/json" \
  -d '{"status": "in_progress"}' \
  "/v1/items/wi_3f9c1a2b"

What happened

yourun SEO Ops on brindlewhaite.example and open the top three work items

GET/v1/properties/prop_61aa/items?status=backlog2007 items

PATCH/v1/items/wi_3f9c1a2b200in_progress

PATCH/v1/items/wi_9e02c7d1200in_progress

PATCH/v1/items/wi_4b77f0aa200in_progress

agentSeven cards in the backlog. I opened the top three by rank — a title rewrite on /packs/ultralight-frames/ (4.2k impressions, no clicks), the FAQ schema on /guides/trail-blazes/, and a verify-first on the knots post. Starting with the first:

✎Snippet/packs/ultralight-frames/

primary

Title + meta lead with ‘ultralight pack frame’ — 4.2k impressions at 0% CTR

Page 2, no clicks

10.5position

4.2kimpressions

0clicks

0%ctr

The whole setup

# 1 · mint a partner key: console → Settings → API keys
# 2 · drop it where your agent reads secrets
SEO_OPS_PARTNER_KEY=seoops_live_…

# 3 · leave a recipe in the project so the agent knows the rules
#     .seo-ops/recipe.md
#     - backlog:    GET  /v1/properties/{id}/items?status=backlog
#     - start work: PATCH /v1/items/{id} {"status":"in_progress"}
#     - never archive; ask first

# 4 · talk to it
"run SEO Ops on brindlewhaite.example
 and open the top three work items"

Any agent that can read a file and call HTTP — Claude Code, Cursor, your own runtime. Same hashed key the console uses; nothing it can do that you cannot.

Start a run

POSThttps://api.seo-ops.com/v1/properties/$PROPERTY_ID/runs

Headers

Authorization	Bearer $KEY
Content-Type	application/json

Body

{"days": 90}

202 Accepted310 ms

{
  "id": "run_8c2e4d7a",
  "status": "running",
  "stage": "reading_search_data",
  "started_at": "2026-08-21T07:31:44Z"
}

curl -X POST \
  -H "Authorization: Bearer $KEY" \
  -H "Content-Type: application/json" \
  -d '{"days": 90}' \
  "https://api.seo-ops.com/v1/properties/$PROPERTY_ID/runs"

Read the backlog when it lands

GEThttps://api.seo-ops.com/v1/properties/$PROPERTY_ID/items?status=backlog

Headers

Authorization

Bearer $KEY

200 OK96 ms

[
  {
    "id": "wi_3f9c1a2b",
    "title": "Title + meta lead with 'ultralight pack frame' — 4.2k impr at 0% CTR",
    "status": "backlog",
    "rank": 1,
    "impact": "High",
    "effort": "Medium",
    "seen_count": 5,
    "last_seen_at": "2026-08-21T07:40:03Z"
  },
  … 6 more
]

curl -X GET \
  -H "Authorization: Bearer $KEY" \
  "https://api.seo-ops.com/v1/properties/$PROPERTY_ID/items?status=backlog"

What arrives at your endpoint, every run

POSThttps://your.app/hooks/seo-ops

Headers

Content-Type	application/json
X-Agency-Event	items.upserted
X-Agency-Signature	sha256=3b1f…9e

Body

{
  "event": "items.upserted",
  "api": "v1",
  "occurred_at": "2026-08-21T07:40:05Z",
  "property_id": "prop_61aa…",
  "run_id": "run_8c2e4d7a",
  "data": { "item_count": 7, "items": [ … ] }
}

200 OKyour endpoint replies

— then push each item into ClickUp, Jira, Linear, or a sheet.
   report.ready fires first, with the run summary.

# sample payload your endpoint receives
curl -X POST \
  -H "Content-Type: application/json" \
  -H "X-Agency-Event: items.upserted" \
  -H "X-Agency-Signature: sha256=3b1f…9e" \
  -d '{
  "event": "items.upserted",
  "api": "v1",
  "occurred_at": "2026-08-21T07:40:05Z",
  "property_id": "prop_61aa…",
  "run_id": "run_8c2e4d7a",
  "data": { "item_count": 7, "items": [ … ] }
}' \
  "https://your.app/hooks/seo-ops"

Verify the signature, then file the items

import hmac, hashlib

def verify(raw: bytes, header: str, secret: str) -> bool:
    digest = hmac.new(secret.encode(), raw, hashlib.sha256).hexdigest()
    return hmac.compare_digest(header, f"sha256={digest}")

@app.post("/hooks/seo-ops")
def hook(req):
    if not verify(req.body, req.headers["X-Agency-Signature"], SECRET):
        return 401
    for item in req.json["data"]["items"]:
        clickup.create_task(list_id, name=item["title"],
                            description=item["detail"],
                            priority=item["impact"])
    return 200

Or skip the endpoint: GET …/export?format=json pulls the open backlog whenever you want it.

What your client gets — from your brand, not ours

From Brindlewhaite Outfitters <reports@brindlewhaite.example>

To owner@summitgear.example

Subject Search work this week — 7 items, 2 already moving

Hi Dana — this week’s pass over summitgear.example found seven things worth doing. Two are already in progress on our side.

/packs/ultralight-frames/ — page 2 for ‘ultralight pack frame’, 4.2k impressions, no clicks. Title and meta rewrite. In progress.
/guides/trail-blazes/ — ranks 5, no clicks; a featured snippet is taking them. FAQ schema + opening paragraph. In progress.
/blog/19-knots-every-hiker-should-know/ — 89 queries land here; we’ll read it before touching it.

Four more on your board · Homepage as we saw it on Aug 21 attached

Turn it on

PATCH/v1/properties/$PROPERTY_ID

Headers

Authorization	Bearer $KEY
Content-Type	application/json

Body

{
  "digest": {
    "recipients": ["owner@summitgear.example"],
    "cadence": "weekly",
    "from_name": "Brindlewhaite Outfitters"
  }
}

200 OK71 ms

{ "digest": { "recipients": 1, "cadence": "weekly", "next_send": "2026-08-25T14:00:00Z" } }

curl -X PATCH \
  -H "Authorization: Bearer $KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "digest": {
    "recipients": ["owner@summitgear.example"],
    "cadence": "weekly",
    "from_name": "Brindlewhaite Outfitters"
  }
}' \
  "/v1/properties/$PROPERTY_ID"

Your name on it, your sending domain (SPF/DKIM, verified once). The homepage capture is the one Scout rendered during the run — proof of work without writing the email.

FOUNDATION · 10 candidates → 7 shipped · 3 held

A site still finishing its basics. The recipe is a sequence, not a mix: snippets first, then hygiene, batched into one campaign so four look-alike rewrites read as one job.

Snippet ×4 — one campaign, four pages
Hygiene ×3 — verify indexation before rewriting

3 held · ship_range_full(7/7) — the band for this shape is 5–7. Nothing padded, nothing dropped.

COMPOSE · 20 candidates → 7 shipped · 13 held

A dense site with plenty to choose from. The recipe composes: at most two of any shape, every strong lens represented, one slot always reserved for a risk.

Snippet ×2 · Schema ×2 · Content ×2
Check first ×1 — the guard slot, never crowded out

13 held · shape_slots_full(on_page_rewrite, 2/2) ×9 · shape_slots_full(structured_add, 2/2) ×2 · shape_slots_full(verify_first, 2/2) ×2 — no week is seven rewrites.

Every candidate is kept. The ones that did not ship are in the run’s record with the rule that held them — you can disagree with a rule; you cannot be padded.

recipe-selection.json · from the run archive

{
  "recipe": "FOUNDATION",
  "ship_range": [5, 7],
  "stats": {"n": 10, "basics_share": 0.70, "invisible_layer": false},
  "campaigns": {"c3": "ctr_fix_on_page_rewrite_pass", "c4": "…", "c6": "…", "c8": "…"},
  "shipped_n": 7,
  "held_n": 3,
  "held": [
    {"candidate": {"shape": "structured_add", "lens": "structure_schema", …},
     "reason": "recipe:ship_range_full(7/7)"},
    …
  ]
}

{
  "recipe": "COMPOSE",
  "ship_range": [7, 9],
  "stats": {"n": 20, "basics_share": 0.45, "invisible_layer": false},
  "guard_used": true,
  "shipped_n": 7,
  "held_n": 13,
  "held": [
    {"candidate": {"shape": "on_page_rewrite", "lens": "ctr_fix", …},
     "reason": "recipe:shape_slots_full(on_page_rewrite, 2/2)"},
    …
  ]
}

Where this is going: MCP — the same key, spoken natively by your agent — is next. Today, agents work against the API; the guide inside the console walks through it.

Start

Point it at your own site and see.

Sign up, run Scout on your agency's own site, and read the backlog it files. You will know within one run whether the recommendations are any good — which is the point of starting somewhere you can check the work.

Create your account Read the setup guide

Already have an account? Sign in.