SEO tooling — decision record

Context

Reviewed on 2026-07-21 while evaluating CrowdReply’s backlinks marketplace (https://crowdreply.io/features/backlinks-marketplace). Broadened into a full comparison of the SEO tooling stack we already pay for (Ahrefs) plus two candidates (OpenSEO self-hosted, WebCEO) across two very different use cases:

HB Plastics internal — six brand domains (Perspex Online, Plastic Online, PM Plastics, Holland, EXP, HB Plastics parent).
LocalSearch agency — hypothetical ~1,000 client accounts, actively managed SEO.

What we measured (live, this session)

Pulled current data via the OpenSEO MCP (which wraps DataForSEO) for perspexonline.com.au and plasticonline.com.au:

Metric	Perspex Online	Plastic Online
Organic keywords (AU)	1,418	2,271
Est. monthly organic sessions	20,323	32,292
Top 50 SERPs showing an AI Overview	33	31
*Cited inside* the AI Overview**	19 (57% of eligible)	19 (61% of eligible)
Referring domains	62	195
Refdomain trend (12 mo)	59 → 62 (flat)	241 → 195 (−19%)

Both sites are already winning ~60% of AI-Overview citation slots on their top 50 keywords with almost no external link authority. That signal is what drives the decisions below.

Decisions

1. Don’t buy CrowdReply

Its tracker half is redundant — we already dominate the signal it monitors.
Its paid-link marketplace half is Webspam risk on our #1 revenue domain, and we’re winning AI citations without paid links, so it solves a problem we don’t have.

2. Don’t add Ahrefs Brand Radar addon

Same reason. Brand Radar is Ahrefs’ AI-citation tracker; we already sit at 57–61% citation rate. Nothing to optimise via a tracker.

3. Keep Ahrefs, keep OpenSEO — different jobs

	Use for
Ahrefs	Weekly/monthly strategic work. Backlink monitoring, competitor gap analysis, keyword planning for the ~50 application pages, rank trend tracking across the WooCommerce → Astro cutover. Ahrefs’ backlink index is deeper than DataForSEO’s — matters for outreach.
OpenSEO (hosted, MCP)	On-demand agent work in Claude sessions and in workers. Pay-per-credit, MCP-first.

Both cost real money — but they cover different parts of the workflow, and the alternative to OpenSEO (Ahrefs’ MCP) is gated behind addons we don’t have.

4. HB Plastics internal — self-host OpenSEO (small side project)

License: MIT.
Upstream: 100% DataForSEO (no Semrush/Ahrefs/SerpAPI/Moz — confirmed by reading the source at github.com/every-app/open-seo).
Deploy paths: Docker OR one-click Cloudflare Workers on the free plan, with CF Access auth and R2 caching of DataForSEO responses (7-day lifecycle).
Cost: DataForSEO pay-per-use direct (~$50/mo at our internal usage) vs OpenSEO hosted’s $10/mo + 28% DFS markup.
Why do it: the win isn’t the $10/mo — it’s that our MCP endpoint stays private (no third party sees which competitors we watch), and marketing-worker can consume it natively for the weekly executive report cron.
Effort: ~1 hour to deploy, plus a runbook.

5. LocalSearch agency (1,000 accounts) — DO NOT self-host OpenSEO

Buy WebCEO Agency Unlimited instead.

Cost math at 1,000 accounts:

	WebCEO	OpenSEO self-hosted
Base + white-label	$99 + $80	negligible (CF free)
Per project	1,000 × $2 = $2,000	0
Rank queries (50 kw × weekly)	$800	~$3,500 (DFS credits)
Backlinks	~$100	~$500 (DFS credits)
Claude API for automation	0	$500–2,000
Total ≈	$3,100/mo	$5,500–7,000/mo
Multi-tenant / white-label	✅ out of the box	❌ build it (6–10 weeks)
MCP / agent-scriptable	❌	✅

WebCEO is ~half the cost, mature, white-labelled, and works tomorrow. Self-hosting OpenSEO for the agency use case = building an inferior WebCEO clone. Wrong tool.

6. LocalSearch — where OpenSEO + Claude could fit

Only as a premium AI-analyst upsell layer on top of WebCEO, not a replacement. What Claude can do that no agency tool can:

Weekly narrative summaries per client, in our voice
Anomaly detection with context (“rank drop is Google shifting the local pack, not the site”)
Auto-drafted content briefs from SERP shifts
Proactive workflow triggers into Pipedrive / Slack / email

Only worth building if it’s a paid premium tier. Not v1.

Recommended sequencing (if LocalSearch scales up)

Month 1: buy WebCEO Agency Unlimited, onboard first 100 clients.
Months 2–4: self-host OpenSEO on our stack, wire Claude worker on top, sell as premium tier at 2–3× standard price.
Month 6+: review usage data. If Claude layer drives retention + margin, consider moving rank tracking off WebCEO. Otherwise keep both.

What NOT to do

Do not buy CrowdReply.
Do not buy Ahrefs Brand Radar addon.
Do not build a WebCEO clone in-house.
Do not self-host OpenSEO for the LocalSearch agency use case first — it’s the differentiator layer, not the base.

marketing-worker-hbplastics skill — where the OpenSEO MCP would plug in
aest-timezone-hbplastics skill — for any weekly-report cron scheduling
OpenSEO repo: https://github.com/every-app/open-seo (MIT)
WebCEO Agency Unlimited: https://www.webceo.com/seo-agency-software-all-tools.htm

Decision record — hand-written, no source_of_truth file dependencies. kb-sync will not auto-regenerate this. If the reasoning becomes outdated (new tool, price change, our AI-citation rate collapses), edit this file directly and bump last_reviewed.

SEO tooling — OpenSEO vs Ahrefs vs WebCEO (decision record)