How to tell if social media is bringing you customers

Your most popular post this month was a joke. People liked it, shared it and tagged their friends. A quieter post explaining how to book got much less attention. Which one did more for the business?

You cannot answer that from the like count. But you do not need an expensive reporting tool to start finding out. A small business can learn a lot by connecting three things: what it posted, what people did next and what the business actually recorded.

The aim is not to prove that every post made a sale. Some posts help people discover you. Others answer a question, build familiarity or help someone finally book. The useful question is whether your social media is helping enough of the right people move towards the outcome you need.

This guide uses a fictional salon and florist, with made-up numbers, to show how that review can work. The examples are not benchmarks or customer results. You can copy the method without expecting the same numbers.

Start with the outcome, not the report

Before opening Instagram, finish one sentence: “This month, social media should help us get more…” Be specific. A salon might want more first appointments on quiet weekdays. A florist might want more local collection orders. A workshop might need enquiries from people who are free on a particular date.

“More engagement” is a different kind of aim. It describes activity around the content, not necessarily progress for the business. It can be useful, but it leaves an important question unanswered: engagement from whom, about what?

For our fictional salon, Elm Studio, the goal is not any message from anyone. It is a new customer asking about an appointment the team can actually provide. A question about opening hours is useful, but a message with a service, preferred date and interest in booking is a stronger buying signal. Keep both if helpful, but do not count them as the same thing.

Agree on what counts before you review the results. For example: an enquiry is one new conversation about a possible booking; a booking is one confirmed appointment; a completed visit is an appointment that actually happened. Multiple messages in the same conversation are not multiple enquiries. A cancellation should not remain in the completed-visit count.

This sounds like admin because it is. A small amount of clear admin often makes the numbers more useful than a very polished dashboard with unclear labels.

Keep attention, interest and results separate

A view, a save and a booking tell different stories. Put them into a simple order so you can see where the evidence gets stronger. This is a way to read the numbers, not a claim that every customer follows a neat straight line.

Even platforms do not treat one number as the whole story. Meta's explanation of content ranking describes using several predictions together, including whether someone may share a post. That explains why social activity has several useful signals. It does not mean that any one of them proves a sale.

Read the definition of each metric in the tool you use. Check whether it counts repeat exposure or distinct accounts, and which time period it covers. A count of accounts is not a count of paying customers. Avoid comparing two labels just because they look similar.

For a local business, location matters too. A widely shared joke may reach people who will never be able to visit. That does not make it a bad post. It means you should not use the total audience as proof that the local booking goal is being met.

Do not add views, saves, website visits and enquiries into one “results” total. The same person may appear in all four. Keeping them separate lets you see what happened instead of making the total look impressive.

Give interested people a path you can check

A useful post needs a next step that you can follow through the business. For Elm Studio, that might be a consultation booking link. For our fictional florist, Stem House, it could be a collection-order form. Choose the path that fits how customers already buy, rather than forcing everyone into a new process just to make tracking easier.

If you use Google Analytics on your website, a tagged link can identify traffic from a particular source. Google's campaign URL guidance explains how the source, medium and campaign parameters appear in acquisition reporting. For an Instagram profile link, a simple example is:

https://example.com/book
?utm_source=instagram
&utm_medium=social
&utm_campaign=first_visit
&utm_content=profile_link

The line breaks are only for reading. The real link is one continuous URL. Replace the example domain and path with your working booking page. Keep naming consistent so the same campaign does not split into several report rows.

A tag identifies the link someone used. It does not automatically record a completed booking, and a shared profile link does not tell you which post caused the visit. If you want separate link-level information, give separately placed links different content labels. Do not attach personal customer details to tracking parameters.

If the booking finishes on another company's site, check what that provider actually records and whether your measurement follows the handoff. Do not assume your website analytics can see a confirmation it never receives. You may need the booking tool's report or a simple manual record instead.

Test the path yourself. Can you get from the post's instruction to the right service? Does the page work on a phone? Are prices and available dates clear? If interested people reach a confusing page, making more posts may simply send more people to the same problem.

Phone calls and walk-ins need a lighter approach. Ask “Where did you first hear about us?” and, when natural, “Was there anything that helped you decide to book?” Those are different questions. A friend may have introduced the business while Instagram helped someone feel comfortable enough to visit.

Write down the answer without forcing it into the channel you hoped to hear. “Friend, then Instagram” is more useful than quietly changing it to “Instagram.” Customer memory is imperfect, so treat self-reported answers as useful evidence with limits, not exact tracking.

Compare two posts without fooling yourself

Imagine Elm Studio publishes two posts. One is a broad joke about bad hair days. The other explains what happens at a first consultation. The team looks at each post after the same seven-day window and checks the booking record for enquiries it could connect to that post.

If the goal is first appointments, the guide gives the team a more promising lead to follow. It helped generate conversations about the service. The right next move might be another clear first-visit post, this time answering the question that came up most often in those conversations.

It would still be too much to conclude that guides always beat jokes. These are tiny numbers, not a reliable general rule. The posts may have gone out on different days, reached different people or benefited from an offer elsewhere. Someone who booked after the guide may already have noticed the salon through the joke.

Keep the public number and the business outcome side by side. Do not turn three bookings into a grand success story, and do not throw away the less direct post because you cannot connect it to a sale. A sensible content mix can include discovery, useful explanations, proof of work and a clear invitation to book.

A fairer review compares posts with similar jobs, formats, age and support. Separate paid reach from unpaid reach. Note a public holiday, a local event or a change in availability. A two-week-old post has had more time to produce enquiries than one published yesterday.

When the count is small, write down the actual count as well as any percentage. Going from one enquiry to two is a 100% increase, but “two enquiries instead of one” is a much clearer basis for a decision. The point is to learn, not to make the month look dramatic.

Use business records to check the story

You do not need a spreadsheet with forty columns. Start with the week, post or topic, the action you asked for, relevant enquiries, confirmed orders or bookings, and a short note on the evidence. Use a private order or enquiry reference if you need to avoid counting the same person twice. Do not put sensitive customer messages into a public content report.

For Stem House, the weekly note might read: “The collection guide led to four messages about Friday flowers. Two became orders. One needed a delivery area we do not cover. One has not replied.” That is an invented example, but it shows how useful a short note can be. The next post might explain collection times or delivery boundaries, rather than just repeat the most colourful bouquet photo.

Keep an “unknown” category. Some customers will not remember the post. Some will see an image in a group chat. Some will find you through several routes. An honest gap is better than a confident but false attribution.

Google Business Profile can add another view of local interest. Google defines its call metric as clicks on the call button, not completed conversations. Direction requests are not confirmed visits, and website clicks are not orders. The available measures depend on the business. Use them as clues alongside your records, and do not label a rise in Google activity as an Instagram result without evidence connecting the two.

Also distinguish a confirmed booking from useful business. Did the customer attend? Was the order cancelled? Did a discount attract work the business could not deliver well? You do not need to answer every question every week, but check them before deciding that the highest booking count is automatically the best outcome.

Finally, recorded credit is not proof that the sale would not have happened without social media. A returning customer may click an Instagram link because it is convenient. That is still a real order, but it is different from a new customer discovering the business through a post. Keep those stories separate where your records allow it.

Make one small decision each week

Pick a quiet time for a short review. Start with the goal, then look at the posts old enough to judge. Check useful actions and business records before looking for your most-liked image. Finish by writing one sentence about what you will keep, change or test.

If posts get attention but few people visit the profile, check whether the subject relates clearly to what you sell. If people visit the profile but do not take the next step, check the offer, essential information and link. If enquiries arrive but bookings do not, look at the actual reasons: price, availability, location, unclear service fit or a slow response.

These are possible explanations, not automatic diagnoses. Ask what evidence would distinguish them. If three people asked the same booking question, answer it in the next post and on the booking page. If you only have one vague signal, collect more information before changing the whole plan.

Keep changes small enough to learn from. You cannot tell what helped if you change the subject, format, price, posting schedule and booking page all at once. On the other hand, do not leave a broken link in place just to keep a test tidy. Fix clear problems first.

For businesses with a longer buying cycle, a weekly review should focus on progress rather than instant sales. A wedding enquiry can take much longer to become a booking than a coffee order. Follow that enquiry over time instead of declaring the original post a failure after a few days.

You can start this review with the tools and records you already have. Only pay for a reporting tool once you know which missing information would change a decision. A tool that saves time joining reliable records may be useful. Another dashboard of the same likes may not be.

Give the next content plan something to learn from

The most useful result of measurement is a better next post. “The consultation guide brought questions about upkeep” is something you can act on. “Engagement rose” often is not.

Unstudio helps research your business, plan content and make draft posts. Give it the business context behind the numbers: which questions became bookings, which services have room for more customers and what people still misunderstand. Those details are often more useful for the next brief than a screenshot of a large view count.

Do not assume that connecting a social account also connects your till or booking records. Keep checking real orders in the systems that hold them. Unstudio can help with the content work; a claim that a post caused a sale still needs evidence.

If your review finds that people are interested but still unsure about visiting, start with the five questions a good Instagram feed should answer. If you know what to improve but need fresh subjects, try our guide to posting when nothing is new. Or bring your website to Unstudio and start turning what you know about your customers into a useful plan.

Elm Studio and Stem House are fictional businesses. All example views, likes, enquiries and bookings are invented to explain the method, not typical results or Unstudio performance claims. The cover is an AI-created editorial image. Metrics and reporting options can change; the linked platform documentation defines the measures discussed here.