Projects

AI visibility doesn't improve on its own. It improves because someone publishes better content, earns a placement in a respected outlet, or restructures a site to be more machine-readable. The Projects dashboard lets you connect those specific efforts to measurable outcomes, so you can prove what worked, justify the investment, and plan what to do next.

Product overview: Daily Prompt Tracking — the three metric tiers, explained for newcomers.

Why Projects Matter

Without a project, your AI visibility data is a stream of numbers with no narrative. Metrics go up or down, but you can't answer the question your stakeholders care about: did this work?

A project creates that link. It defines what you're doing, when you started, what success looks like, and which prompts to monitor. Spyglasses then tracks everything within that scope and alerts you when a goal is achieved.

For SEO and AEO Professionals

You're optimizing pages to appear in AI-generated responses; rewriting content for machine readability, adding structured data, improving topical authority. But AI visibility is noisy. Metrics fluctuate report to report because AI outputs are non-deterministic.

Projects help you cut through that noise:

• Before/after framing: Set a start date when you publish optimized content, then filter your Historical Metrics and Citation Intelligence dashboards to that project's date range. Instead of looking at all-time data with dozens of confounding factors, you see the impact of this specific effort.
• Page-level goal tracking: Link a page group to an SEO project. The moment any page in that group is cited by an AI platform, the goal is marked as achieved and everyone subscribed to the project's goal alerts gets an email. No more manually checking reports to see if your work landed.
• Prompt scoping: Assign only the discovery queries relevant to this effort. If you're optimizing for "best CRM for small teams," you don't need results from your broader query library cluttering the picture.

For PR and Communications Teams

You're running earned media campaigns; pitching stories, securing coverage, building relationships with journalists and publishers. Traditional PR measurement tracks placements and impressions. But in an AI-first search landscape, the real question is: are AI platforms citing the publications that covered us?

Projects let you answer that:

• Coverage tracking at every granularity: Track a whole publisher domain (techcrunch.com), one specific article you placed, everything written by a particular journalist, or an entire coverage group — a bulk-imported list of your placements tracked as a single goal. When an AI platform cites a matching source in response to a relevant query, the goal is hit.
• Influence rates, not just first hits: Every match is recorded, so each goal shows what share of the project's AI citations your tracked coverage accounts for — "this article is 2% of all sources AI used for these prompts."
• Campaign measurement: A PR campaign has a defined timeline. Create a project with start and end dates that match your campaign window, and filter dashboards to see exactly how AI visibility changed during and after the push.
• Client reporting: The project detail page shows goals, prompts, annotations, and links to filtered dashboard views — and the exported PDF/PPTX reports include per-goal influence rates. Everything you need to build a campaign report showing the connection between outreach work and AI visibility outcomes.

Quick-Start: Creating Projects from Other Dashboards

You don't always need to start from the Projects page. Two other dashboards have one-click shortcuts that skip the manual setup and pre-populate the form for you.

From the AI Readiness Audit (SEO Projects)

After a site audit completes, look for the folder icon in the audit card header. Clicking it:

1. Creates a page group containing all pages from the audit, named Audit – [date]
2. Opens the New Project form with type pre-set to SEO and the new page group already selected

This is the natural next step after an audit: you've identified the pages you want to optimize — now create a project to track whether that work results in AI citations.

From Citation Intelligence (PR Projects)

In the Citation Intelligence dashboard, the Top Third-Party Citation Sources table has checkboxes on each row. Select the domains you want to monitor and click the Create PR Project button that appears:

1. Opens the New Project form with type pre-set to PR
2. All selected domains are pre-filled as publisher targets

This is useful when you've identified which third-party sites are generating AI citations in your space and want to start an outreach or monitoring campaign around them.

---

Creating a Project

Navigate to Projects in the property sidebar under Configuration, then click New Project.

Step 1: Choose a Project Type

The project type determines what goals are available. You can always track core metrics regardless of type.

Step 2: Add Details

• Name (optional): A recognizable label like "Q1 Content Refresh" or "Product Launch Media Push"
• Description (optional): Context for your team about what this project covers
• Start Date: Defaults to today. Set it to when the effort began for accurate before/after comparison
• End Date (optional): Set this for time-boxed campaigns. Leave blank for ongoing efforts — you can complete the project later

Step 3: Set Goals

For SEO projects, select a page group. Page groups are managed in the Property Pages dashboard. When any page in this group appears as a citation in an AI response to one of your tracked prompts, the goal is marked as achieved.

If you haven't created page groups yet, you'll be prompted to do so first.

For PR projects, you can track coverage at four levels of granularity, and mix them freely within one project:

Publisher goals answer "is this outlet influencing AI?"; article and author goals answer the more precise question "is the coverage we placed influencing AI?" — a publisher goal fires even when the citation is an article your client had nothing to do with.

General projects don't have specific citation goals. They're for tracking overall metric movement.

Step 4: Assign Prompts

Select which discovery queries should run nightly for this project. Only prompts assigned to active projects are executed, so this controls both what gets tracked and your prompt usage.

Choose prompts that are relevant to what you're optimizing for. If you're optimizing content about CRM solutions, assign CRM-related queries, not your full library.

As you toggle prompts, two feedback cards update live to help you make an informed selection.

Coverage Impact

The Coverage Impact card shows how your selection compares to the full library's buyer-journey × brand-dimension matrix. It mirrors the scoring used in the Coverage tab of the Discovery Queries dashboard, so the same framework applies here.

You'll see:

• A composite Coverage score (0–100) with a color-coded number.
• Three progress bars for the sub-scores:
  • Matrix breadth (60% weight) — share of (stage × dimension) cells your selection covers.
  • Stage coverage (25% weight) — prompts per buyer-journey stage vs. minimum targets. Decision-stage is weighted highest because it's closest to purchase.
  • Framework spread (15% weight) — how many distinct generation frameworks are present in your selection.

The bars move as you add or remove prompts. It's a fast way to confirm your project isn't unintentionally concentrated in one corner of the matrix.

Auto-pick for max coverage

Inside the Coverage Impact card, the Auto-select button picks prompts for you using a greedy coverage algorithm. Set a maximum prompt count (defaults to 10, max 50) and click the button. Spyglasses evaluates the full library, picking at each step the prompt that most improves the composite coverage score, until it hits the cap or no prompt can improve the score further.

This is useful when:

• You're starting a project and want a sensible baseline selection.
• You have a budget constraint ("no more than 8 prompts running nightly") and need the best coverage within that budget.
• You want to avoid hand-picking from a large library.

Auto-select fully replaces your current selection. You can still tweak the result with the checkboxes afterward.

Data Accuracy

Once you've selected at least one prompt, the Data Accuracy card appears below Coverage Impact. It shows the margin of error on your nightly observations at 90% confidence, given the number of prompts × 30 nightly runs per month:

• Reliable (< 8% margin of error) — green. Trend analysis is sound.
• Directional (8–15%) — amber. Movements are suggestive but noisy.
• Exploratory (> 15%) — red. Treat as a rough signal only.

Coverage and accuracy are complementary: coverage tells you what you're tracking, accuracy tells you how confidently. Aim for green on both.

Understanding the Project Detail View

After creating a project, the detail page shows:

Overview

The project header displays type, status, date range, and quick links to View Metrics (Historical Metrics dashboard filtered to this project) and Citations (Citation Intelligence dashboard filtered to this project).

Goals

Each goal shows its current status:

• Pending: No citation match detected yet
• Hit: A citation matching the goal criteria was found. Shows the date of the first hit and a link to the cited URL
• Missed: The project was completed without this goal being achieved

When a goal is achieved by a nightly run, Spyglasses automatically creates an annotation marking the date and emails everyone subscribed to the project's goal alerts. Who receives that email is opt-in and controlled from the Notifications card — see below.

Influence rates

Spyglasses records every citation that matches a goal, not just the first. Each goal with hits shows its citation count and influence rate — the share of all citations the project's prompts produced that came from that tracked target:

TechCrunch: 6 citations — 4% of this project's 150 citations

Influence rates give you the sentence every PR report needs: not just "we got cited" but "our placements account for X% of the sources AI relies on for these prompts." Adding or removing goals never changes the denominator (it's all citations from the project's prompts), so rates are comparable across goals.

Top-line impact metrics

Above the Goals card, every project shows the two numbers that connect your work to the metrics clients track — share of voice and brand mentions:

• Share of goal influence — distinct citations matched by any goal, divided by all citations the project's prompts produced. "How much of what AI cites comes from our tracked targets."
• Attributed brand mentions — of those matched citations, how many actually generated a brand mention in the answer. This is the direct-impact number: "our placements didn't just get cited, they got the brand named."

A citation counts as attributed in two cases: the answer text credited the brand for that source, or — on Google AI Overviews — the overview mentions the brand while one of your tracked sources is in its source list. AI Overviews don't attribute their sources individually, so we credit the matched source and label how many attributions came in through this rule. Treat the AIO portion as strong-but-circumstantial: the brand and your source appear together, but AIO doesn't say which source produced the mention.

A smaller line beneath tracks the near misses among matched citations: pages that mention your brand in content without an inline citation (hidden earned media), pages the AI read but didn't cite, and pages that appeared in the AI's search but weren't picked up.

Use the All citations / Brand-attributed toggle on the Goals card to flip every goal's count and influence rate between the two views — "everything our targets matched" vs. "only what generated brand mentions."

Publisher and article goals also display the publisher's AIPVS score so you can see target quality alongside performance.

Drilling into a goal's citations

Click a goal's citation count to open its citations page — every citation the goal matched, with the prompt that triggered it, the platform and model, the date, and (for inline citations) the context around the reference. Click any row to read the full AI answer, opened to the platform that cited you. Coverage-group goals link to the group's own AI Citations page instead.

The page is scoped to the project window and uses the same attribution rule as the project's top-line metrics, so every count here matches the project page exactly. The brand-relevance breakdown:

• All citations — everything the goal matched within the project window
• Attributed to your brand — the answer credited the brand for this source (this card reproduces the goal's "attributed citations" count from the project page; AIO-rule attributions are labeled Attributed via AI Overview)
• Mention your brand, not cited inline — pages whose fetched content carries your brand, but the AI never credited them in the answer (hidden earned media — AI reads your placement while composing, without citing it). Each of these shows the mention context from the page content
• Brand in answer, not this citation — the AI answer mentions your brand somewhere, but the mention isn't attached to this citation. Without this class, you'd open the answer, see your brand, and wonder why the citation says "not brand-related"
• Page cited inline — the page itself was referenced in the answer text. This is a citation of the page, not necessarily a brand mention — an answer can cite your placement while talking about something else. For the brand-focused number, use the Attributed card

Coverage-group goals drill into the same page (citation-level, project-scoped); a link from there opens the group's own coverage page, which counts URLs rather than citations — a different unit, deliberately labeled as such.

The cards filter the list below, and a relevance filter separates attributed to brand / mentions brand in content / brand in answer / not brand-related. This is the honest version of a publisher goal's count: "the domain earned 31 citations; 12 carry our client; 4 got credited by name." The influence rate itself is unchanged — it still counts all matched citations.

Each hit also carries a citation type telling you how far the page got through the AI's retrieval pipeline: cited inline (the page was referenced in the answer — not necessarily with a brand mention), evaluated source (the AI read the page while composing but didn't reference it), or search result (the page appeared in the search the AI ran, but the AI didn't pick it up — it may never have read it). "Search result" is the one people misread: it means your page was on the menu, not that the AI used it. Citation types are about the page; the brand story is the relevance breakdown above. See how citation types work for the full breakdown and per-platform caveats.

Adding goals to a running project

Click Add goal on the Goals card to add a publisher, article, author, or coverage-group goal at any time. New goals are matched against the project's past citations automatically — if the article you just added was already being cited, its hits and influence rate appear within moments, and the goal is quietly marked as Hit (dated to the earliest matching citation, without triggering notification emails).

For long goal lists, the card shows the first 10 goals with a Show all toggle — though if you're tracking dozens of placements, a coverage group keeps them as a single goal instead.

Tracked Prompts

The project page shows the tracked-prompt count; View & manage opens the full list on its own page — searchable, with per-prompt removal — so projects tracking hundreds of prompts stay readable.

Annotations

Shows all annotations tied to this project, both manual entries and auto-generated ones (like goal achievements). These same annotations appear as diamond markers on your time-series charts.

Notifications

A project sends two kinds of email, and both are opt-in per recipient from a single Notifications card on the detail page:

• Goal alerts — sent the first time each goal is cited (see Goals above).
• Weekly summary — a Monday-morning recap of metrics and goal progress (see Weekly Summary Emails below).

The card lists every organization member with two toggles — one for goal alerts, one for the weekly summary — so you decide exactly who hears about each project. Toggling saves immediately; there's no separate save step.

• New projects start quiet. Only the person who creates the project is subscribed (to both), so creating a project no longer emails the whole organization. Turn on the people who are actually working the account.
• Organization members are toggled individually. Someone who doesn't work on a given client property simply stays off its notifications.
• External addresses — clients, executives, or teammates without Spyglasses accounts — can be added by email and given either or both notification types.

Projects created before opt-in notifications were introduced have all current members subscribed to both streams, so nothing changed for them. Trim each project's list down to just the people who need it.

Using Projects with Other Dashboards

One of the most powerful features of projects is the ability to filter other dashboards by project.

From the project detail page, click View Metrics or Citations to open the corresponding dashboard with the date range automatically set to the project's start and end dates and the project filter applied.

You can also apply a project filter directly on the Historical Metrics or Citation Intelligence dashboards using the project dropdown. This narrows all charts to the project's date window, giving you a focused view of metrics during that specific effort.

Weekly Summary Emails

Every Monday morning, Spyglasses sends a summary email for each active project. The email is designed to keep stakeholders in the loop without requiring them to log in, and to give you something you can forward to clients or executives as a self-contained update.

What's in the email

• Headline metrics — current values for brand consistency, share of voice, mentions, and citations, with week-over-week change indicators
• Goal progress — how many of the project's goals have been hit vs. how many are still pending
• PDF attachment link — a download link for the latest generated project PDF (if one has been exported), so recipients have a portable report they can share offline
• Open in Spyglasses — a link to the project detail page, included only for recipients who have a Spyglasses account

If your organization is on an agency plan with white-label branding configured, the email uses your agency logo instead of the Spyglasses logo, so it's ready to forward to clients as-is.

Who receives it

Weekly summary recipients are managed from the Notifications card on the project detail page (see Notifications) — the same place you control goal alerts. Turn the weekly summary toggle on for any organization member, or add an external address for a client, executive, or teammate without a Spyglasses account.

Recipients are opt-in: a new project only emails its creator until you add others, so members are no longer subscribed automatically. (Projects that predate opt-in notifications keep all current members subscribed, preserving their prior behavior.)

External recipients see the headline metrics and PDF link but not the "Open in Spyglasses" button, since they don't have an account to log into.

How recipients unsubscribe

Every project email — both weekly summaries and goal alerts — includes an unsubscribe link in the footer.

• Organization members on a weekly summary land on their personal email preferences page, where they can opt out of onboarding emails or the newsletter in addition to project summaries.
• External recipients land on a one-click unsubscribe page that removes their address from that project's recipient list — the weekly list or the goal-alert list, depending on which email they clicked. They don't need a Spyglasses account to do this.

Unsubscribing removes the recipient from that specific list for that specific project; the toggle in the Notifications card flips off to match. If they need to be added back later, re-enable them from the project detail page.

Managing Project Lifecycle

Active Projects

Active projects have their assigned prompts run nightly. Keep a project active while the effort is ongoing and you want continued monitoring.

Completing a Project

When the work is done, click Complete on the project detail page. This sets the end date to today and moves the project to a completed state. Completed project prompts stop running nightly, but all historical data remains accessible.

Reactivating

If you need to extend a completed project, click Reactivate. This clears the end date and resumes nightly prompt execution.

Archiving

Archive projects you no longer need in your active view. Archived projects retain all data but won't appear in the default project list.

Best Practices

Start the Project Before the Work

Create the project and assign prompts a few days before you publish optimized content or launch a campaign. This establishes a baseline in the metrics that you can compare against.

Keep Prompt Assignments Focused (But Reliable)

Assign prompts that directly relate to what you're optimizing — don't load up your entire library just because you have it. At the same time, keep an eye on the Data Accuracy card:

• 8 or more prompts clears the "Reliable" bar (under 8% margin of error), which is the safest foundation for trend analysis.
• 3–7 prompts is acceptable when your project is narrowly scoped and the Coverage Impact score stays reasonable — you'll be in the "Directional" band, meaning movements are suggestive rather than conclusive.
• Fewer than 3 is only appropriate for highly exploratory projects. Treat the data as rough.

The Auto-select button is tuned to balance these two forces for you: give it a count that meets your accuracy target and it will pick a set with the best coverage at that size.

Use Descriptive Names

"Q1 Blog Refresh: CRM Category Pages" is far more useful than "Project 1" when you're reviewing results three months later or generating reports for stakeholders.

One Effort Per Project

Resist the temptation to bundle multiple unrelated efforts into one project. If you're doing an SEO content refresh and a PR campaign at the same time, create two projects. This lets you attribute metric changes to the right effort.

Review After Two Weeks

AI visibility changes take time to materialize. Don't check daily and get discouraged. Set a reminder to review the project's filtered metrics two weeks after making changes, then again at four weeks. Look for sustained trends, not single-report jumps.

Explore projects with an AI assistant (MCP)

If you've connected the Spyglasses MCP server, you can analyze a project in plain language instead of clicking through the dashboard. Ask your assistant to "analyze my Q1 content project" and it calls list_projects and get_project_insights to pull the same metric deltas, weekly trend, goals — including per-goal hit counts and influence rates — and annotation timeline you see here, then summarizes progress and recommends next moves. The Analyze a project prompt runs this end to end.

See MCP → Account data for the full tool reference.

Related

• MCP Server — Analyze projects and account data from an AI assistant
• Historical Metrics — Track trends over time, filterable by project
• Coverage Groups — Bulk-import placement lists and track them as single project goals
• Citation Intelligence — Analyze citation patterns, filterable by project; create PR projects from top sources
• AI Readiness Audit — Audit pages for AI readiness; create SEO projects from audit results
• Annotations — Mark important dates on your charts
• Discovery Queries — Manage the prompts assigned to projects
• Property Pages — Create page groups for SEO project goals