November 11, 2026 · 7 min read
Ask AI About Your Social Numbers: A Monthly Report Prompt for Brokers
A broker does not need to be a data person to read a monthly social report. Here is a fill-in prompt for Metricool and Claude, a worked example and a five-step check.

A broker can read a monthly social report with AI by connecting Metricool to Claude, pasting one fixed prompt that names the month, the accounts and five plain questions, and requiring every number to carry its source. Then check three figures against Metricool before the summary reaches an agent or a partner.
Our page asking does a real estate team need a marketing dashboard covers requests, approvals and the schedule: where the work lives. It does not read your results for you. This prompt does, and that is the part most brokers skip because the analytics screen looks like homework.
The usual failure is plain: the numbers sit in a tab nobody opens, so next month's plan rests on a feeling. Below is the prompt, block by block, with a worked example and a check that keeps the answer honest.
Why reading results is now the AI task to learn
Metricool's 2026 State of AI in Social Media Study surveyed more than 700 social media professionals. Its press release says AI use for metrics analysis climbed from 32% to 59% since last year. Reading results is no longer rare.
The same release adds that 54% measure AI's impact through time saved, compared with 6% who measure content quality and 4% who track post performance. Most people ask how many hours AI saved. Few ask if the posts did better.
These respondents are social media professionals, not real estate agents, so treat the figures as a signal, not a brokerage benchmark. The lesson carries over: judge AI by what it helps you decide next, not by how fast it wrote.
What you need before you paste anything
Metricool's press release on its Claude and ChatGPT connectors, dated 10 September 2026, says the connectors are available to all Metricool users, including those on a free plan. Once connected, you can access data from your social accounts, analyze content performance and manage content planning inside the conversation.
The setup is short. You need four things:
A Metricool account with your brokerage accounts already connected, so there is data to read.
Metricool connected to Claude, using the steps on Metricool's announcement.
One goal for the month, in a single sentence: booked listing appointments, seller inquiries, recruiting conversations or event turnout.
One result from outside Metricool, such as the inquiries your team logged from social. Your CRM is a separate system, so you type it in.
Use the last full calendar month and the month before it. A partial month makes every comparison look like a drop.
The prompt, block by block
Paste the five blocks into one message and replace everything in square brackets.
Block 1: Role and reader
You are the marketing analyst for [brokerage name]. Your reader is the broker-owner, who is not a data person. Write short sentences. The first time you use a metric, define it in a few words.
This sets the reading level, so reach and impressions come with a translation.
Block 2: Scope and permissions
Use my connected Metricool account. Look at [account names] on [networks] for [last full month] and compare with the month before. Read the data only. Do not create, edit or schedule any post.
The connector can also schedule content, so say plainly that this session only reads. Reading and publishing belong in separate conversations.
Block 3: Five questions
Answer these in order. 1. Which three posts did best, and what do they have in common: format, topic, listing or not? 2. Which format did we post most, and did it earn that share of attention? 3. How did listing posts compare with community and recruiting posts? 4. What changed since last month that I should not ignore? 5. What should we do more of, less of and keep the same next month? Give one action for each.
Five questions is the limit. Question 5 forces the answer into a decision, which is the point of a report.
Block 4: Output shape
Start with one decision line. Then list the three best posts with a link or title. Then give a short paragraph for each question. Keep the whole answer under 500 words.
A fixed order lets you compare this month with last month at a glance and hand the result to an agent as one page.
Block 5: Honesty rules
Our goal this month is [one sentence]. Outside Metricool, our team logged [your number] [inquiries or appointments] from social. For every number you give, name the account, network and date range. If the data does not show something, write "not in the data" instead of guessing. Never say a post caused an inquiry. Label anything you inferred as "inference."
This block carries your goal and outside result, and it blocks the two habits that sink AI reports: confident guesses and invented cause.
A worked example: one team, one October
This is an illustration of the process, not real data, so no figures appear. A team under one brand runs Instagram, Facebook and LinkedIn. The broker pastes the prompt with her goal: more seller inquiries before the holidays, and the inquiries her team logged from social.
Claude's decision line says the listing carousels did the most for the team, and that the one open house Reel did well but was never repeated. The three best posts are two listing carousels and one agent introduction. On format, the team posted mostly single images, and its best posts were not single images.
On community versus listing posts, Claude says community posts drew more comments and listing posts drew more saves, and it marks the reason as "inference." It flags that LinkedIn posting dropped off late in the month. Its three actions: repeat the open house Reel, turn one weekly single image into a carousel and restart the LinkedIn schedule.
The broker did not need to know what a rate is. She read three actions and decided. The answer did not claim the Reel produced a seller inquiry, because Block 5 forbade it.
Check the answer before it travels
AI reads tables quickly. It can also misread a column or mix up two accounts. Treat the output like a draft from a sharp new hire and run this check every month:
Pick the three best posts Claude named and find them in Metricool. Confirm the title, date and network.
Pick one number from the answer and find it in Metricool. If you cannot find it, ask Claude where it came from.
Look at the date range Claude says it used. A wrong month is the most common slip.
Read every sentence with "because" in it. Any claim about cause should be tagged "inference" or removed.
Ask one follow-up that tests the answer: "What in this data would make your top recommendation wrong?"
Five minutes of this separates a useful report from a confident-sounding one.
The guardrail most brokerages are missing
The same Metricool press release says only 12% of respondents have a clearly defined AI policy, while 35% report having no restrictions or formal rules around how they use AI at work. A monthly report is a low-risk place to start one. Three lines are enough: reports use brokerage accounts only, no client names, phone numbers or lead details go into the prompt, and a person signs off before the answer goes to anyone else.
How Urban Marketing Edge handles this
We put the numbers in plain language every month. AI helps us move faster on drafts and variations, and strategy and taste stay with a person. Each plan comes with plain-language monthly reporting, monthly adjustments on what to repeat and what to drop, and a post approval system so nothing goes out unseen. The report feeds the next month's themes, so the prompt above is a small version of how we plan.
Our plans page shows what each level includes, and our work page shows results for Rocks Realty and Clareo Group. We say plainly when a smaller setup would do the job.
Questions brokers ask
Can I use AI to read my real estate social media report if I am not a data person?
Yes. Connect Metricool to Claude, paste the five-block prompt and ask for plain sentences with every number sourced. You judge the three recommended actions, not the math. Confirm the best posts and one number in Metricool before sharing.
Do I need a paid Metricool plan to use the Claude connector?
No. Metricool's press release on its Claude and ChatGPT connectors says the connectors are available to all Metricool users, including those on a free plan.
Will Claude post or change anything when I ask for a report?
It can, because Metricool says the connector can create and schedule content. That is why Block 2 says to read the data only and never create, edit or schedule a post. Run reports in their own conversation.
How do I know the AI did not make a number up?
You check. Find the three best posts in Metricool, find one number from the answer and confirm the date range. Tell Claude to write "not in the data" when it cannot see something.
How do I know if using AI for this is working?
Judge it by decisions and results, not hours. Metricool's 2026 study found 54% measure AI's impact through time saved, compared with 6% who measure content quality and 4% who track post performance. Track if the actions you took lifted your goal, such as logged seller inquiries.
Should agents run this prompt on their own accounts?
They can, with your three-line rule: no client details in the prompt and a person reviews the answer. Give agents the same blocks.
Book a 30-minute strategy call
If your numbers sit unread, bring last month's report and your goal. On the call we read it together, tell you which few figures matter for your brokerage, and show how this prompt fits your accounts. No pitch deck, no pressure.
Book a 30-minute strategy call and we will map the simplest reporting routine for your team. If you want to see the options before you talk, start with our 12 Week Program, which trains an in-house marketer to run this kind of review.


