FAQ

Common questions, plainly answered.

What CMO.ie does, how it works, what it costs, and how your data is handled. If your question isn't here, email hello@howl.ie and we'll add it.

Jump to a section

What CMO.ie does

The 30-second version. If you only read one section, read this one.

What problem does CMO.ie solve?
Customers increasingly ask AI tools (ChatGPT, Claude, Perplexity, Gemini, Google AI Overviews) for recommendations instead of typing a Google query. CMO.ie tracks whether your brand is mentioned in those answers, where competitors are mentioned and you aren't, and what to publish or change to start showing up.
Is this just SEO?
No. Traditional SEO optimises for Google's blue-link results. AI search engines synthesise answers from a different mix of sources, weight Irish publishers differently, and don't index a page just because Googlebot can read it. We track the AI side specifically, with a separate methodology — it complements, not replaces, your existing SEO work.
Who's it for?
Irish brands with a digital footprint who care about being recommended by AI. Most useful if you have at least a basic website + some product/service pages, and you compete in a category customers research before buying. Agencies use it to manage multiple client brands from one workspace.
What's the shortest path to value?
Sign up, create a project for your brand, fill in (or auto-extract) the brand profile, click Generate suggestions to get ~10 customer-style prompts, click Generate full set if you want 30-50. Once prompts are saved, the daily run picks them up and we check each prompt against the AI engines on your plan. Within 24-48 hours you have a real visibility baseline.

Getting started

How do I sign up?
Visit /signup, give us your name, company, email and password. We provision a trial organisation for you with one project, one prompt, and three AI models — enough to see the product working before you pick a paid plan.
What happens during onboarding?
After signup we walk you to /onboarding which guides you through: (1) confirming your brand name + website, (2) auto-extracting a brand profile from your homepage, (3) seeing 10 suggested prompts, (4) picking competitors, (5) running your first daily check. Total time: ~5 minutes.
Do I need an API key from Anthropic / OpenAI / Google?
Trial accounts: yes (BYOK). Paid plans: no — we use our managed keys by default. Power users on paid plans can switch to their own keys via Settings → API Keys if they want full control over spend or want to run more checks than their plan covers.
Can I add more than one brand?
Starter is one project. Pro is three. Advanced is unlimited. Agency tier is built around multi-client management with a credit pool that can be allocated across client projects.

Prompts

Prompts are the questions your customers might ask AI when researching your category. They're the unit of measurement — every check runs each prompt against each model.

What's a good prompt?
A natural question a real customer would type — not a keyword string. "Best digital agencies in Dublin for a B2B SaaS launch?" is good. "digital agency dublin" is not. We auto-suggest 10 prompts based on your brand profile so you don't have to write them from scratch.
How is the suggested-prompts list built?
Claude reads your structured brand profile (segment, audience, products) — never the raw HTML — and generates 10 conversational prompts split across awareness (~3), consideration (~4), and decision (~3) intents. We hard-lock to your stated industry so you don't get prompts about adjacent businesses.
What's the difference between Generate suggestions and Generate full set?
Generate suggestions returns ~10 prompts, fast. Generate full set returns 30-50 covering the broader funnel + scores each one's importance (1-5) + maps each to its closest plain-English Google query so you can see the keyword volume behind it. Use Generate suggestions for a quick start; use Generate full set when you want comprehensive coverage.
Can I write my own prompts?
Yes — the Add your own input lives directly under the suggestion list. We'd recommend starting with the suggester even if you intend to override it; the AI tends to phrase prompts in customer language better than we do.
How many prompts can I track?
Trial: 1. Starter: 25. Pro: 50. Advanced: unlimited. Agency: unlimited. The cap is on active prompts; archived prompts don't count.

Daily runs + visibility

How often do you check my prompts?
Trial: 2 runs per month. Starter: 4. Pro: 30. Advanced + Agency: unlimited. A run fires every active prompt against every selected model, captures the response, parses citations, and computes share-of-voice across your brand + competitors.
Which AI models are checked?
ChatGPT (OpenAI), Claude (Anthropic), Perplexity, Gemini (Google), and Google AI Overviews. Trial includes 3 of those; Starter 2; Pro 4; Advanced + Agency 5.
What's share-of-voice?
Across all the prompt × model checks in a run, what share of mentions belonged to your brand vs each tracked competitor. It's the headline number that tells you whether you're improving against the field over time.
Why does the result for the same prompt change?
AI models are stochastic — they don't return identical answers every time. We run each prompt multiple times where the plan allows it and aggregate. Over a month, the noise smooths out and the real trend emerges.

Brand profile

Why does the brand profile matter so much?
Every downstream feature — prompts, action plans, briefs, monthly playbooks — uses the brand profile as ground truth for what your business actually is. If the profile says you're a digital agency, suggestions will be agency-customer questions. If it's wrong, everything downstream is wrong. The profile is the most important thing you tell us.
How is the brand profile created?
On project creation, we fetch your homepage, parse the visible text, and ask Claude to fill out five fields: short description, market segment, brand identity, target audience, products/services. You see it before it's saved and can edit any field.
What if the auto-extract gets it wrong?
Two paths. (1) Click Re-extract — most failures are transient (rate limit, JS-only homepage, redirect). (2) Edit the fields manually — you know your business better than the model, and the manual version is what every downstream feature uses.
Why isn't extraction working?
Most often: your site is behind Cloudflare bot protection, or the homepage is JS-rendered with no server HTML, or it's a redirect to a country-specific domain. The error message usually tells you which. Editing the profile manually unblocks you regardless.

Gaps + actions

What's a gap?
A specific domain or URL the AI models keep referencing when answering your prompts where competitors are mentioned and you aren't. The Gap Analysis page ranks them by opportunity — high-traffic sources where you're absent are where you have the most to gain by appearing.
What's an action plan?
Click Generate plan on the Actions tab and we run a three-Claude pipeline (Gap Analyst → Strategist → Brief Writer) that turns your top gaps into a prioritised list of moves. Each action has effort + impact ratings, root-cause analysis, and concrete steps.
What's a brief?
A click-to-generate content brief targeting a specific gap. Tells you who the audience is, what to write, what to include, and what voice to use — so you (or your writer) can publish a piece of content that has a real chance of earning the citation you're missing.
What's the action tier on each plan?
Trial + Starter: 'gaps' (gap identification only). Pro: 'strategy' (gaps + prioritised plan). Advanced + Agency: 'full' (gaps + plan + briefs + draft + polish requests).

SEO audits

What's the SEO audit?
A 9-phase deep audit of your site — keyword landscape, on-page review, content gaps, technical SEO, AI search resilience, competitor benchmarking, backlinks, local SEO, prioritised action plan. Calibrated for the Irish market. Delivered as a markdown report you can share with your team.
How long does it take?
60-120 seconds end-to-end. The page shows a stage-by-stage progress indicator (crawling → reading meta tags → checking Core Web Vitals → mapping keywords → writing the report). Safe to close the tab — the audit keeps running and shows up in your list when done.
How much does it cost?
Pro: 1 free audit per month. Advanced: 3 free per month. Trial / Starter / Agency overflow: €49 per audit. Agency tier includes 1 per active client project per month.
Why did my audit fail?
Most common: your site blocked our crawler (Cloudflare, Webflow, or a 403). The audit row shows status 'unavailable' with the specific reason. Some sites also fail mid-audit when PageSpeed Insights returns transient errors — a re-run usually works.

Monthly playbooks

What's a monthly playbook?
On the 1st of each month we generate a one-pager for every active project — the three highest-leverage moves to make this month based on the previous 30 days of data. Delivered to every team member's email + viewable inline at /admin/playbooks.
How are the three moves chosen?
Top gaps + visibility deltas + share-of-voice changes get fed to Claude with the brand profile. Claude returns the moves in CMO.ie house voice (practical, Dublin-inflected, no corporate filler). Each move has a specific 'do this by Friday' task — not generic 'consider improving content' advice.
Can I regenerate a playbook?
Yes — admins can force-regenerate from /admin/playbooks. Useful if Claude misread the data or the prompt template changed. Doesn't trigger a fresh email send by default.

Crawlability tool

What's the crawlability tool?
Free, login-free check at /crawlability. Tests whether 40+ AI bots (GPTBot, ClaudeBot, PerplexityBot, GoogleBot, etc.) are allowed by your robots.txt + can fetch a sample URL. Tells you which AI engines can actually read your site — a precondition for ever showing up in their answers.
Why is crawlability free?
We use it as a marketing top-of-funnel — give value before asking for an email. If your site is bot-blocked, no amount of CMO.ie tracking will help; better you find out for free.

Plans + billing

Pricing is on /pricing — this section answers the questions the pricing table can't.

What's a check?
One prompt run against one AI model = one check. If you have 50 prompts × 3 models × 30 runs/month, that's 4,500 checks/month on Starter. Each plan has a soft total-checks cap (most plans are unlimited; Trial is capped at 10).
Do failed checks count against my limits?
No. Only successful checks count. If a model API errors out, we retry with backoff and eventually skip — that doesn't burn your runsPerMonth allowance.
How does cancellation work?
Cancel anytime from Settings → Billing. Access stays live until the end of your current billing period. We don't pro-rate refunds, but data is preserved for 30 days after cancellation in case you change your mind.
Do I get a discount for paying yearly?
Email hello@howl.ie — annual deals are case-by-case at the moment, especially for agency tier. No public discount code.
What if I exceed my plan's prompt limit?
You can't add the next prompt — the form blocks it. Either upgrade your plan or archive prompts you no longer need. Archived prompts preserve their history but don't count towards the cap.

Agency tier

What's different about Agency?
Multi-client management. One workspace, many client projects, with a credit pool that can be allocated across clients (instead of a per-project plan). BYOK by default — your Anthropic / OpenAI / Gemini keys, your spend, your control. Includes the MCP server + REST API on Scale tier and above.
What's the credit pool?
1 prompt × 1 model × 1 day = 1 credit. Pre-allocate credits across client projects monthly so a high-priority client gets more frequent checks than a maintenance one. See /agency for the full model.
Can my clients log in to see their own data?
Not yet on the customer side — the agency tier is built around the agency operator running it. White-labelled client logins are on the roadmap but not v1.

Data + privacy

Where's my data stored?
Supabase (Postgres) in the EU region (Frankfurt) for all customer data. Vercel hosts the app — also EU regions. We don't store your AI provider's responses any longer than needed for trend analysis (90 days by default; longer on Advanced + Agency).
What's sent to the AI providers?
Just the prompt text + the basic context the run engine needs (brand name + market). We never send your customer data, your team's emails, or anything outside what's needed to run the visibility check.
Is CMO.ie GDPR-compliant?
Yes. Howl.ie (Ireland) is the data controller; we publish our processing basis at /privacy. Subject access requests and right-to-erasure requests are answered within 30 days; email privacy@howl.ie.
Do you sell or share my data?
No. We don't sell to third parties, don't share data across customer accounts, and don't run ads inside the product. Subprocessors (Anthropic, OpenAI, Google, Vercel, Supabase, Stripe) are listed in /privacy.

Account

How do I add a teammate?
Settings → Team → Invite. They get an email with a sign-up link tied to your organisation. New teammates default to 'member' role; only owners can grant admin.
How do I change my password?
Settings → Account → Reset password. Or use the 'Forgot password' link on /login if you can't sign in.
How do I delete my account?
Settings → Account → Delete account. This permanently removes your organisation, projects, prompts, history, and team. We don't keep a soft-delete copy.
Where do I report a bug?
Email hello@howl.ie or use the chat icon in the bottom-right of the dashboard if it's enabled. We aim to respond within one business day; bugs go straight to the engineering backlog.

Still stuck? Email hello@howl.ie — we'll usually reply same day.