What Ad Intelligence Means for a Small Business

You have a little money set aside for ads this month. You pick a photo you like, write a caption that sounds right, and boost it. Two weeks later you can't say why it worked, or why it didn't, so next month you guess again.

Ad intelligence sounds like something for big companies with analysts. For a small business it means one habit: look at evidence before you spend, not after. You don't need special software for it. You need three sources of evidence and an hour a month.

The three inputs

Every good ad decision draws on some mix of these:

1. Your own past results.
2. What competitors and your category are already running.
3. What your customers actually ask.

Each one answers a different question. Your results tell you what has worked for you. The category tells you what everyone else is saying, and where nobody is saying anything. Customers tell you the words and worries that make someone buy.

Input 1: your own past ads and posts

Start with what you already own. Open your last two or three months of ads and organic posts and sort them from best to worst.

Look at the numbers that match what you wanted. If the goal was enquiries, look at cost per result, not likes. These four cover most of it:

Metric	What it tells you
Cost per result	What you paid for each enquiry, booking or order
Click-through rate (CTR)	Whether people cared enough to tap
Cost per click (CPC)	What each tap cost
Frequency	How often one person saw the ad, which spots a tired creative

Then ask what the winners had in common. Same photo style? A face in the first frame? A price in the headline? A question in the opening line? Write the pattern down in one sentence. Do the same for the losers.

Be honest about small samples. If an ad got a handful of clicks, treat the result as a hint, not a rule. Organic posts help here too: likes, comments and shares are a cheap way to see which message people respond to before you pay to push it.

Input 2: what competitors and your category are running

Public ad libraries let you see ads that businesses are running right now. The Meta Ad Library covers Facebook and Instagram. TikTok has the Creative Center. LinkedIn has its own Ad Library, and Google has the Ads Transparency Center. All are free to browse.

Search for your category and your nearest rivals, then look for repeats. If five businesses have run the same ad for a long time, it is probably earning its place. If every ad opens with a discount, the discount is no longer a way to stand out.

Collect, don't copy. For each ad, note:

• Hook: what the first line or first second says.
• Offer: what the viewer is asked to do.
• Format: photo, carousel, short video, or a text-heavy image.
• Language: English, Malay, Chinese, Arabic, or a mix.

The most useful finding is often a gap. Maybe every dental clinic near you shows smiling before-and-after faces and none explains what a first visit feels like. That is an angle you can own.

Input 3: what customers actually ask

This is the input most small businesses already have and never use. Customers hand you ad copy for free, in their own words.

Check four places:

• Comments under your posts and under competitors' posts.
• DMs and WhatsApp messages.
• Reviews on Google and your social pages.
• Sales calls or counter conversations, which you can jot down after each one.

Copy the exact phrases into a note. For example:

"Is it halal?" "Can I collect after 6pm?" "Will it work on sensitive skin?"

Count how often each question appears. The top three are your next three ads, because each one is a doubt that stops people buying.

Use the customer's words in the headline. A question you hear ten times beats a clever line you thought of in the shower.

A simple monthly routine

Block about two hours in the last week of the month. Don't do it during a busy day.

1. Review your results (30 minutes). Rank last month's ads and posts. Write one sentence on what the winners shared and one on the losers.
2. Scan the libraries (30 minutes). Browse the public ad libraries for your category. Save five to ten ads and note hook, offer, format and language.
3. Read your customers (30 minutes). Go through comments, DMs, reviews and call notes. List the top questions and objections, with a tally.
4. Pick three angles (15 minutes). Choose one that repeats a past winner, one that fills a gap you saw in the libraries, and one that answers a common question.
5. Draft two versions of each (15 minutes). Change one thing between versions, such as the photo or the opening line, so you can tell what made the difference.
6. Set a test budget and a stop rule. Decide the amount and the point at which you will pause a version before you launch anything.

Keep a running note with the date and what you learned. After three months it becomes your own playbook.

Worked example: a florist in Dubai

Say you run a small florist in Jumeirah, Dubai, that sells hand-tied bouquets and table arrangements, with a few hundred dirhams a month to spend on ads. This is an invented example, so the findings are made up to show the method.

Own results. Last month you ran three posts. The close-up of a bouquet being tied got the most saves and shares. The photo of your shopfront got few. The pattern you write down: hands and flowers up close beat wide shots of the shop.

Category. You browse the Meta Ad Library for florists in the UAE. Most open with a festive discount. Almost none talk about how fresh the flowers are. Gap noted: "cut this week, arranged today".

Customers. You go through DMs and reviews. The same questions keep coming up: can you deliver the same day, can I add a handwritten card, and will the flowers last the drive to Abu Dhabi. You tally them. The delivery question appears most.

Three angles:

• Repeat the winner: a close-up of a bouquet being tied, with the hands in the first frame.
• Fill the gap: a short video of the morning's flowers being unpacked, with the caption "Arranged today, delivered today".
• Answer the question: a post titled "Will it survive the drive to Abu Dhabi?" showing how you wrap and water it.

You draft two openings for each angle, set an example test budget of AED 300 across all six versions, and decide that any version that has spent its share with no enquiries gets paused. Next month you repeat the routine and compare. The florist now has a reason behind every ad, and a record of what it learned.

The same routine works for a tuition centre in Puchong. Past results might show that parent testimonials beat teacher photos. The libraries might show every centre promising better grades. Parent DMs might keep asking about class size and make-up lessons, which becomes the angle nobody else is running.

Where Marvis fits

You can run this routine with a notebook and a spreadsheet. Marvis is built for the drafting end of it. In our hero demo, Marvis turns a product photo, a brand kit and market research into four scored ads for Instagram, Facebook, TikTok and LinkedIn. The scores in that demo are illustrative and do not predict results, so treat them as a way to compare drafts, not a forecast. Our tagline is "Marvis AI creates, analyzes and optimizes your ads", and you stay in charge of which ones to approve and what to put money behind. If you want to try it, start at /sign-up.

Whatever tool you use, this month's task is the same: put your last three months of ads in order, from best to worst, and write down one sentence on why.