Product guide

Getting started with heyEO

heyEO is built around one loop: measure, choose one change, ship it, measure again, and explain what you observed. This guide walks through each step and where to find it in the app.

On this page

Start here

Seven steps, one loop. Jump to the step you're on.

  1. Step 1Set up a projectSite, customer, competitors and prompts.
  2. Step 2Understand the first resultsSite checks, AI answers and audience evidence.
  3. Step 3Choose a recommendationTrack one fix you can finish.
  4. Step 4Hand off a fixA ticket or a coding-agent prompt.
  5. Step 5Record the changeThe page and the date it went live.
  6. Step 6Measure againA comparable scan against your baseline.
  7. Step 7Explain what was observedA before/after receipt and the weekly brief.
  8. See real results firstTour heyEO's own public report

What heyEO is

heyEO helps a small team see how its website and AI answer engines represent the product, decide on one change, and check what moved after it shipped. Three kinds of evidence are kept apart:

  • Site checks are deterministic: the same fetched page state gives the same result.
  • AI answers are sampled: one answer per prompt and search track per scan, and answers vary between runs.
  • Audience (ICP) evidence is a model's estimate of who your published pages seem to be for.

Want to see real results before signing up? Tour heyEO's own public report.

Your first week

  1. Day 1 — set up and scan. Add your site, describe your customer, save a few competitors and prompts, then run your first scan. The cost is shown before it starts.
  2. Day 1 — read the results. Look at site health, where AI answers name you, and any audience drift. Set this first completed scan as your baseline.
  3. Day 2 — pick and hand off one fix. Track one recommendation you can finish and hand it to whoever ships it.
  4. Day 3–4 — ship and record it. Record the change with the date it went live.
  5. Day 5–7 — measure again. Run a comparable scan, open the comparison and copy the before/after receipt. Treat any movement as observed, not proven.

1. Set up a project

Tell heyEO about your site, your customer and the questions buyers ask.

Sign in

Continue with Google, GitHub, LinkedIn or X. heyEO never sees your password. After signing in you continue to project setup. If your account didn't share an email address (X often doesn't), add one under Settings → Email address to get the weekly digest, alerts and workspace invitations.

Project setup

Enter your website and product name. Fill in your declared ICP, competitors and markets yourself, or switch on Suggest with AI for any section to get a draft from your website. You review and edit every suggestion before the project is created.

Project setup wizard step asking for the website URL, showing heyeo.io
Screenshot · Sept 2026Project setup — start with your website URL.The current screen may look slightly different.
Onboarding review step showing AI-drafted project details ready to edit
Screenshot · Sept 2026Review — edit suggestions before creating the project.The current screen may look slightly different.

Describe the audience as specifically as you can: company size, role, industry, maturity, pain points, buying triggers and use cases. heyEO compares this declared ICP with what your published pages appear to say.

Monitored prompts and prompt coverage

Where to find it:WorkspaceSettingsMonitored prompts

Monitored prompts are the buyer questions heyEO asks the answer engines. Under Buyer question coverage, Settings counts your active prompts by intent — awareness, comparison, alternatives and buying — and lists the markets they cover. When an intent has no prompts, it offers an example you can start from. Examples use your product name; markets are not filled in for you.

IllustrationSettings → monitored prompts: coverage, an example, and the Add step. Simplified drawing, not a screenshot.
When it helps
Before a scan, to check your prompts cover the questions buyers actually ask — not just questions about you by name.
What you do next
Press Use this on an example, edit the wording in New monitored prompt, then press Add. Nothing is saved until you press Add.

Run a scan

Press Run scan in the top bar. Before anything is spent, the dialog shows what will run — your monitored prompts, how many pages to crawl, and the exact credit cost. Every fetched page is audited.

A full scan asks each monitored prompt once on the search tracks selected in Settings. Choose any non-empty combination of OpenAI search, Google search, Perplexity search and the neutral web benchmark. If a track answers without proof that it searched the web, heyEO leaves that cell unmeasured rather than guessing. Results are directional evidence, not consumer-app rankings or traffic claims.

Run a scan dialog with page-allowance options and the cost shown before starting
Screenshot · Sept 2026Run a scan — choose how many pages and confirm the cost before you start.The current screen may look slightly different.

From your agent: ask it to price a scan first. It shows you the quote and starts only after you approve. Agent recipes

2. Understand the first results

Each page answers one question. Start with the dashboard, then follow the evidence.

Dashboard

Where to find it:MonitoringDashboard

Pick a project in the switcher. The dashboard shows the latest scan, site findings, answer measurements, audience alignment and the highest-priority recommendations. A brand-new project shows empty states until its first scan completes — never sample data.

When it helps
Right after a scan finishes, to see what was measured and where to look first.
What you do next
Open the area that needs attention — Site Audit, AI Visibility or ICP Drift — or go straight to Recommendations.

Site Audit

Where to find it:MonitoringSite Audit

Findings are deterministic and grouped by category, so the same page gives the same result on every scan.

When it helps
To see which technical basics — canonicals, titles, headings, structured data, indexing — fail or warn, and on which pages.
What you do next
Expand a finding to see the affected pages. Use the page icon beside a page to see everything else heyEO knows about it.

AI Visibility

Where to find it:MonitoringAI Visibility

The matrix shows each monitored prompt in its own words against each search track. “Brand absent” means that sampled answer did not name you; an unmeasured cell means no usable answer was recorded. Measured answers start collapsed — expand one to read its mentions, competitors and citations. Older scans from before the retrieval panel are labelled as model answers, not search results.

If a scan sampled no prompts, add prompts in Settings and run again.

When it helps
To see, prompt by prompt and track by track, whether answers name your brand, who they name instead, and which sources they lean on.
What you do next
Open a measured answer for its mentions and cited links, then use the source explorer and competitor evidence below.

Source explorer

Where to find it:MonitoringAI VisibilityWhere the answers get their information

Domains are listed by how often they were cited in this scan's sample, with the number of distinct answers that cited each one. Your own domain is tagged Your site. The first ten are shown; Show all lists the rest. Counts describe this sample only.

IllustrationAI Visibility → Where the answers get their information. Simplified drawing, not a screenshot.
When it helps
When you want to know which sites the answers cite, and whether your own site is one of them.
What you do next
Expand a domain to see its URLs and the answers that cited it. Press an answer to jump to it in Measured answers.

Competitor evidence

Where to find it:MonitoringAI VisibilityWho fills the gap

Press a competitor's name to open What buyers are told about…. For each answer that named them, you see real supporting sentences as quotes, name-only mentions as plain names, the sentiment of each, whether your brand was also present, and the sources that answer cited.

IllustrationAI Visibility → Who fills the gap → a competitor name. Simplified drawing, not a screenshot.
When it helps
When a competitor keeps appearing and you want to read what buyers are actually told about it.
What you do next
Read the quotes and sources, then decide whether a recommendation — or a missing page of your own — addresses the gap.

ICP Drift and reveal mode

Where to find it:MonitoringICP Drift

ICP Drift compares your declared audience with the audience inferred from your pages, dimension by dimension, with the quotes each inference is based on. If a later scan skipped audience analysis, heyEO keeps the latest evidence and shows which scan it came from rather than presenting it as new.

Switch on Reveal mode to see who your site sounds like it's for before the comparison. Press Reveal your declared audience to bring the comparison back. Reveal mode lasts for that visit only.

IllustrationICP Drift → Reveal mode. Simplified drawing, not a screenshot.
When it helps
To check whether your site still speaks to the customer you declared — or, with reveal mode, to see how it reads with fresh eyes.
What you do next
Review the dimensions that drift and the page quotes behind them, then update the page or your declared ICP.

Page evidence drawer

Where to find it:Site Audit, Recommendations, ICP Drift or AI Visibilitypage icon beside a link

Next to links to pages on your own site there's a small page icon, What we know about this page. The link itself still opens the page. The drawer gathers that page's technical findings, recommendations, audience quotes and the measured answers that cited it, with the date each was measured. If part of the scan couldn't be loaded, the drawer says which part.

IllustrationAny page link on your own site → the page icon beside it. Simplified drawing, not a screenshot.
When it helps
When one page keeps coming up and you want everything heyEO recorded about it in one place.
What you do next
Pick the recommendation that covers the page, or press Open answer to read an answer that cited it.

3. Choose a recommendation

Pick one thing you can finish this week.

Recommendations

Where to find it:Act on itRecommendations

Recommendations turn scan evidence into an ordered list. Each shows where it came from (site audit, AI visibility or audience), its effort and the evidence behind it. Priority balances impact, confidence and effort; it is a planning aid, not a prediction of search or revenue results.

From your agent: ask “what should I fix first?” and have it track the one you choose. Agent recipes

When it helps
When you've read the results and need to decide what to do first.
What you do next
Open a recommendation and press Track this recommendation. Tracking is what lets you hand it off, record the change and see it reassessed later.

Team focus

Optional, in Settings → Team focus. Split your current delivery capacity across four kinds of work — on-site technical; owned content and evidence; earned citations and off-site; measurement and monitoring — in whole percentages that total 100. Each recommendation then shows a focused priority next to its base priority. Clearing Team focus restores the base order.

Team focus settings with capacity split 40/30/20/10 across four work types totalling 100 percent
Screenshot · Sept 2026Team focus — an optional capacity split across four kinds of work.The current screen may look slightly different.

4. Hand off a fix

Give whoever ships it the evidence, not a vague ask.

Hand off this fix

Where to find it:Act on itRecommendationsa tracked recommendationHand off this fix

Opens a read-only preview with two tabs. Team ticket gives the title, the affected pages, the recorded evidence, any suggested fix and how to verify it. Coding agent prompt quotes the evidence and asks the agent to find the relevant code or content, check the finding still applies, propose the smallest change and explain how to verify it. It never invents file paths. Recommendations written by a model carry a review-before-acting note. Copying doesn't change the recommendation.

IllustrationRecommendations → a tracked recommendation → Hand off this fix. Simplified drawing, not a screenshot.
When it helps
When someone else — a teammate, contractor or coding agent — will make the change.
What you do next
Paste the text into your ticket tracker, chat or coding agent. Then come back to record the change once it's live.

5. Record the change

A dated record is what makes the next comparison meaningful.

Record change

Where to find it:Act on itRecommendationsa tracked recommendationRecord change

In the tracked recommendation, fill in Changed page URL, Date the change went live and Earliest date a follow-up should count, then press Record change. Move the recommendation through New, Accepted, Implementing and Implemented as work progresses — Implemented is only available once a change is recorded. You can also attach a specific completed scan as the follow-up.

IllustrationRecommendations → a tracked recommendation → Record change. Simplified drawing, not a screenshot.

From your agent: tell it what you shipped and it can record the change for you, after you confirm. Agent recipes

When it helps
As soon as the change is live on your site.
What you do next
Wait for, or schedule, a comparable scan. The recommendation shows Waiting for reassessment until one arrives.

Updates

Where to find it:EvidenceUpdates

After a completed scan you may see a notice that a tracked recommendation looks actioned, with a Review updates link. It doesn't start a scan. Differences seen after the change are recorded alongside it, not attributed to it.

Updates inbox empty state: No recommendation updates yet
Screenshot · Sept 2026Updates — empty until a tracked recommendation looks actioned after a later scan.The current screen may look slightly different.
When it helps
After a later scan, when heyEO thinks a tracked site-audit recommendation looks actioned.
What you do next
Choose Confirm actioned, Not this change or Review later. Only your confirmation records the change — detection never does it alone.

6. Measure again

Same prompts, same pages, a fair comparison.

Scan history and baseline

Where to find it:EvidenceScan history

Every scan is listed with a readable name. Setting a baseline changes future comparisons only; past scans aren't modified, and earlier baselines stay listed.

Scan history with readable run names and baseline selection
Screenshot · Sept 2026Scan history — readable run names and baseline selection.The current screen may look slightly different.
When it helps
To choose which earlier scan later scans are compared with.
What you do next
Press Set as baseline on a completed scan, then Compare with baseline on a later one.

Scheduled scans

Where to find it:WorkspaceSettingsScan schedule

Available on paid workspaces; owners create and manage schedules, and members are asked to contact the owner. Before saving, heyEO shows the expected runs and credits per month from your current pricing. Nothing is charged until a scan runs. You can pause and resume a schedule, and you're told if credits run out.

When it helps
When you want comparable scans to keep arriving without starting each one by hand.
What you do next
Choose the cadence, time zone, run time, scan type and pages, check the expected monthly credits, then press Create schedule.

Baseline comparison

Where to find it:EvidenceScan historyCompare with baseline

Site findings, AI answer measurements and ICP evidence are compared separately. If prompts or tracks differ, the affected section is marked partial and explains why instead of hiding differences. Scans from the older model panel can't be compared with retrieval-panel scans for AI visibility — set a new retrieval scan as your baseline.

Baseline comparison page showing comparable runs and site findings
Screenshot · Sept 2026Baseline comparison — differences against the selected baseline.The current screen may look slightly different.

From your agent: ask “did it improve?” — it compares the latest scan with your baseline. Agent recipes

When it helps
After a follow-up scan, to see what's new, resolved or changed since the baseline.
What you do next
Copy the before/after receipt to share what was observed.

7. Explain what was observed

Share what moved, with the dates and the caveats intact.

Before/after receipt

Where to find it:EvidenceScan historyCompare with baselineCopy before/after receipt

The receipt lists the before and after scan dates, the linked change when one is found, the comparison status and coverage, and what changed for site findings, AI answers (mention rate, position, cited domains) and audience. If the scans aren't comparable it says why instead of listing changes. A result with no change is still a useful receipt. It always ends by saying differences were observed after the change, not caused by it.

IllustrationScan history → Compare with baseline → Copy before/after receipt. Simplified drawing, not a screenshot.
When it helps
When you need to tell someone what happened after a change, in a few lines they can trust.
What you do next
Paste it into your update, ticket or client report.

Weekly brief

Where to find it:MonitoringManagement report

The page is headed with your project name and weekly brief, the measurement date and what it's compared with. Sections: This week at a glance, What changed, Where to focus and Next measurement. Rows for site health, AI visibility, ICP and connected Google data each carry a status: Improved, Declined, Unchanged, Changed, Unavailable or No baseline. ICP is reported as changed, never as better or worse. The brief appears after your first completed scan.

IllustrationMonitoring → Management report. Simplified drawing, not a screenshot.

The same data is available to automation at GET /api/v1/projects/{projectId}/management-report. See the API Docs.

When it helps
Once a week, or before a team or investor update.
What you do next
Press Copy weekly update to paste a summary into Slack or email, or Print it. Use the arrows on each row to open Site Audit, AI Visibility or ICP Drift.

Reading your report

  • The headline figure is the latest measurement. The change beside it is the difference from the comparison scan.
  • No baseline means there was no comparable earlier scan, so no change is estimated.
  • Unavailable means that source had nothing to report — for example, no measured answers or no imported Google data.
  • The movement note is a reminder that changes are observed associations. A rise after an update doesn't prove the update caused it.
  • Dated events mark scans and recorded changes, so you can see what happened during the period.

Search Console and Analytics

Where to find it:WorkspaceSettingsGoogle

Open What heyEO can see to check exactly which permissions Google granted. The brief compares the last 28 days with the 28 days before; Google's data usually arrives a few days late, so recent days can be marked partial. Pages that stand out — for search visibility, low click-through or low engagement — are listed as opportunities to investigate, not problems. You can disconnect at any time; imports stop until you reconnect.

When it helps
When you want search and site-traffic context next to your scan evidence.
What you do next
Connect, choose one Search Console property and one GA4 property, then look for Opportunity to investigate in the weekly brief.

Email digest

Where to find it:WorkspaceSettingsEmail digest

When fresh scan or Google evidence arrives—or tracked work still needs attention—the brief chooses one next action, links to the relevant screen and shows up to three supporting measurements. The selection is rules-based and uses no AI credits. Other workspace members receive the same brief only if they are opted in. The digest needs a confirmed email address; if you signed in without one, add it under Settings → Email address.

When it helps
When you want one evidence-backed decision in your inbox instead of another dashboard reminder.
What you do next
Switch Send me the weekly digest on or off. It's personal to you, and heyEO skips quiet weeks with no useful next move.

Workspace and automation

Inviting your team

Where to find it:WorkspaceTeam

  • Owners manage billing, invites and who has access.
  • Members see the workspace's projects and results.
  • Owners can revoke a pending invite before it's accepted.
When it helps
When someone else should see results or help ship fixes.
What you do next
Invite by email. The teammate signs in from the invite link to join.

Automation tokens

Where to find it:WorkspaceSettingsAPI tokens

Tokens are bound to one project. You can revoke a token at any time. For conversational use, connect an agent instead — see the agent guide.

When it helps
When a script or CI job needs to read results or start scans without you.
What you do next
Copy the token when it's shown — it appears once — and store it as a secret. Then follow the API Docs.

Common questions

Why does my report say “No baseline”?
There's no comparable earlier completed scan yet. Your first scan becomes the reference point; heyEO doesn't estimate a change without one.
What does “Brand absent” mean?
The sampled answer for that prompt and track didn't name your brand. It describes that answer only, not every possible answer.
Why is something marked partial or not compared?
The two scans differ for that item — for example, different prompts or pages — so heyEO shows what it can compare and says what it can't.
Where did my “Hand off this fix” button go?
It appears once a recommendation is tracked. Open the recommendation and press Track this recommendation first.
Why are my Google figures behind?
Search Console and Analytics data arrive with a delay, so the most recent days may be missing or partial. The brief shows freshness rather than filling gaps.

New to a term? See the glossary. Still stuck? Jump back to the journey.