Reporting

A report is not a data dump. It has three jobs:

1. Show progress so the client sees the value of the retainer.
2. Explain results so movements do not cause alarm or confusion.
3. Agree the next steps so the meeting after it is productive.

If the client reads only the first page, they should still know where they stand and what happens next.

The white-label report is the live source of truth. It gives the client access to the data while your monthly review adds interpretation and direction. Cover:

1. Current scorecard. AI Visibility, Brand Knowledge and AI Sentiment, by assistant and category.

2. Competitor comparison. Share of voice and who assistants recommend instead.

3. Notable answers and sources. Real examples that explain the movement.

4. Recommendations and tasks. What CiteMind suggests, what you approved and what is completed.

5. Monthly narrative. A short summary of what changed, why it changed and what happens next.

6. Scope and context. The tracked questions, assistants, markets and baseline date.

The client can open the report between meetings and always see current data, while your monthly summary keeps the conversation focused on decisions.

Adapt the text below. Replace every placeholder with real findings.

Headline: {{client}}'s AI visibility {{improved / held / declined}} this month, mainly on {{question_type}} questions.

What changed:

• AI Visibility: {{direction}} across {{assistants}}, driven by {{reason}}.
• Brand Knowledge: {{direction}} in {{category}}.
• AI Sentiment: {{direction}} on {{category}}, with {{issue_or_win}}.
• Competitors: {{competitor}} {{gained / lost}} share on {{question_type}} questions.

What we did: {{actions_completed}}.

What we will do next month: {{next_actions}}.

Things to know: {{model_update_or_site_change}}. AI assistants generate answers that vary, so we report trends over time.

CiteMind's white-label report uses real-time data and is fully branded for your agency. You can configure the colors, upload your agency logo and present the report as part of your own service. The client does not need to leave your ecosystem or learn a new CiteMind-branded interface.

Share one report link per client, for example:

geo.{{agency_domain}}.com/share/{{report_id}}

The final identifier is unique to that shared report. Use the link in your monthly email, client portal or meeting agenda. The report stays current as CiteMind collects new data, so you do not need to export and resend a static file every time a result changes.

Set up the client-facing URL from CiteMind's Settings. Enter the agency subdomain you want to use for reports, then follow the DNS instructions shown there in the domain provider used by your agency. CiteMind tells you which DNS record to create and what value to point it to.

A typical setup is geo.{{agency_domain}}.com. Once the DNS change has propagated and the domain is verified, shared reports use your agency URL. If you change providers or the subdomain later, update the setting and repeat the verification steps shown in Settings.

Before sending the first report:

• Add the agency logo.
• Set the agency colors so the report matches your brand.
• Configure and verify the reporting subdomain in Settings.
• Create the client report and check the shared URL.
• Confirm that the visible data, recommendations and tracked scope are correct.
• Decide who receives the link and how it will be reviewed each month.

Do one complete check yourself before the client opens it. The report is live, so the client experience should be right from the first share.

• Send it on the same day each month. Consistency builds trust.
• Lead with the story, then the numbers.
• Use the same question set so comparisons are valid.
• Include a bad month honestly. Explain the cause and response.
• Link actions to results only where the evidence supports it.
• Keep jargon low. Write for the person who will forward it to their boss.

• Sending screenshots with no interpretation.
• Reporting every metric. Choose the ones that matter to the client's goals.
• Hiding negative movements.
• Changing the format every month.
• Claiming that an action caused a change without evidence.
• Sending the report with no follow-up meeting.

Fluctuation is normal. Common causes:

• Generated answers. The same question can produce different wording and brand lists.
• Model updates. A new model version or retrieval change can shift results.
• Source changes. A page the assistant relied on is updated, removed or replaced.
• Competitor activity. A competitor publishes content or gains presence on a source.
• Your own changes. Site updates, new content and technical changes.
• Question phrasing. Small wording differences change answers.

None of these means the work is failing. They mean you need a method that tolerates variation.

Before the first report, tell the client:

1. Answers vary, so we report trends and not single results.
2. Small movements up and down are normal.
3. Model updates can shift results for reasons outside our control.
4. We will tell you when we see a real change and what we think caused it.
5. We will show progress through the trend, actions completed and improvements in specific questions and sources.

Put this in the proposal and repeat it in the first report.

Use a stable question set. Do not change questions each month. When you must, record the date and reason.

Track groups, not single questions. A consistent move across related questions is more meaningful. Group by question type, product line or category.

Compare periods, not days. Look at weeks and months, against the previous period and the baseline.

Check several assistants. A movement across several is stronger evidence than one assistant alone.

Look at direction over time. A steady direction matters more than a high or low day.

Pair the numbers with examples. Show actual answers and the sources behind them.

Maintain a simple log of events that could explain changes. Record the date, what happened and which questions or assistants it may affect. Log model updates, site releases, robots.txt changes, new content, presence work, competitor launches and company news. When a result moves, check the log first.

Example script: "Visibility on {{question_group}} dropped this month. Here is what we checked: {{checks}}. The most likely cause is {{cause}}. We have {{action}}, and we will watch the next {{period}} to see if it recovers. One month does not change the trend, but we want you to hear it from us."

Tell the client before they notice, separate what you know from what you suspect, show what you checked and say what you are doing about it.

When scores move slowly, report other evidence:

• Actions completed and their dates.
• Inaccuracies corrected in how assistants describe the brand.
• New sources where the brand now appears.
• Sentiment on a specific category improving.
• Referral visits from assistants, where analytics can identify them.
• Branded search or enquiry trends, as supporting evidence and not proof of cause.