Social Media Metrics That Measure Demand, Not Likes

Most businesses never read their monthly social media report. The reason is not that the report is badly prepared; it is that the numbers inside it have no counterpart in the business. Reach went up, engagement went down, followers reached such-and-such a figure — none of these answer the question "how much work came in from this channel this month".

Part of the problem comes from the measurement tools themselves: the platforms' own dashboards show reach and engagement with ease, but they do not show demand. Demand happens outside the platform — in a message, in a phone call, in a form. If measurement has not been set up, that moment is not recorded and cannot be accounted for at the end of the month.

The text below splits the social media report into three layers and sets out who each layer is shown to and which decision it is meant to prompt. At the end there is a report template that fits on a single page. Nowhere does it say "this rate is good"; the rates that circulate as sector averages rarely show what they are based on, and their meaning changes from one business to the next.

Why the report goes unread

For a report to be read, the numbers in it have to be capable of changing a decision. A reach figure on its own changes no decision: it is not clear what to do when reach rises, or what to do when it falls. So the report becomes a document that gets filed rather than read.

The second reason is that the report is written in the channel's language. "Engagement rate" is a platform concept; it has no equivalent in the business owner's everyday vocabulary. The same information is understood at once when it is written as "six people sent us a message from this post this month".

The third reason is that the report carries no responsibility. A report without a sentence along the lines of "we saw this, so we are changing that" turns into a spreadsheet. If there is no decision to make at the end of the month, the report itself is unnecessary.

All three reasons share one source: the report starts being written before measurement is set up. Without measurement, the only numbers left are the ones that can be pulled from the platform's own dashboard, and those sit in the reach and engagement layer. In other words, the report becomes unreadable not because it was badly put together, but because the column that actually matters never existed. That is why setting up measurement comes before the reporting routine.

Three layers: reach, engagement, demand

Social media measurement is not one list but three separate layers, and when they are mixed together the report loses its meaning.

Reach and engagement are production metrics: they show which content type makes people stop in the feed, and they steer content decisions. Shown to the business owner on their own they serve no purpose, because neither of them is a business outcome.

The demand layer is a business metric: messages, calls, direction requests, forms and visits to the website. The report should open with this layer at the top; reach and engagement should sit below it as explanation.

This separation also manages expectations. A month in which reach falls while demand holds steady is not a bad month; fewer people may have been reached, but the people who were reached may have had a stronger interest in the business. To be able to see that possibility, the two layers need to be written down separately.

From profile to website: no measurement without tagged links

Where a visit from social media to the website came from is not known automatically. Some referrals pass on their source and some do not; with visits that come through in-app browsers, the source information is frequently lost and the visit is recorded as "direct".

The fix is to tag the link: parameters added to the address record which channel and which post the visit came from. Without tagging, part of the traffic that social media sends stays invisible, and the report makes the channel look weaker than it is.

The limit of tagging should be stated plainly too: it records only the visits that reach the site by clicking the link. It does not record the person who saw the profile, remembered the name and searched for it directly the next day. That is not a flaw in the measurement but its nature — the same blind spot exists for traffic from AI assistants, and we treated that side as a separate channel in our article on tracking AI traffic in GA4.

UTM conventions: three naming rules

For tagging to be useful, the naming has to be disciplined. Three rules are enough.

Lowercase, no spaces. Tag values are stored case-sensitively; Instagram and instagram show up as two separate sources and the report splits in two. Pick one spelling and never change it.

A fixed vocabulary. The values that go into the source, medium and campaign fields are defined in a list in advance. In a setup where everyone writes whatever they think fits, the report becomes unreadable within three months.

A date in the campaign name. When the campaign field carries the topic name and the month, the performance of the same topic in different months separates out. This is the most practical split for showing which topic is worth producing again.

Keeping the tags in a single table and generating every new link from that table is the one habit that decides whether the report can still be read three months later.

How to record demand from DMs and comments

Most of the demand that social media generates never touches the website: it lands in the inbox. If measurement has not been set up, that demand shows up nowhere, and at the end of the month someone says "social media brings us no work".

The no-tools method. Whoever answers the message fills in five columns in a shared spreadsheet: date, channel, what the request is about, which post it came from (if known), outcome. It takes a few minutes a day, and by the end of the first month there is a table that nobody in the business has seen before. The limit of this method is human discipline; an entry that was forgotten does not come back.

The semi-automated method. You use the labelling features in the messaging inbox: labels such as price question, appointment request and complaint are defined, and every message is labelled as it is answered. At the end of the month the spread of labels gives you both the number of requests and their type.

The source question. "Where did you hear about us?", asked at the appointment or quote stage, is the oldest measurement tool there is and still the most useful. The answer is not always accurate, but it gives you information that no tool can provide on its own.

The three do not need to be used together; one method the business can keep up is better than three it cannot.

Calls, WhatsApp and directions: the limit of counting a click as demand

Tapping the call button on a profile is a click; it does not mean the person actually called. Tapping the directions button does not mean the person came. Both metrics go into the report, but they should be written down as demand indicators, not as demand.

The way to close the gap is to keep the record on the other side as well: how many calls actually came in, how many people actually walked through the door. When the two numbers sit side by side, the ratio between clicks and what really happened becomes the business's own yardstick — and that yardstick belongs to that business alone; it cannot be interpreted by looking at a ratio borrowed from somewhere else.

The same distinction becomes even more critical on the advertising side; we wrote separately about how the consent and modelling layers work beyond the click in our article on consent, GCLID and modelling in ad measurement.

A return-by-content-type table

The decision at the end of the month is made by looking at a single table: what each content type brought in.

The table is built like this. The rows are content types — not individual posts. Process walkthroughs, answers to frequently asked questions, team introductions, examples of work, announcements and so on. The columns carry the three layers: how many pieces were published, average reach and engagement, how many requests came in.

Looking post by post is misleading, because the performance of any single post is open to chance. Looking by type, on the other hand, forces a decision: if a type brings in no demand two months running, it comes off the calendar.

The hardest column in the table is the demand column, and whether it can be filled depends on the logging routine described in the previous section. No log, no table; that is why measurement has to be set up before content production starts.

Where Google Business Profile items sit in the report

For businesses that serve a local area, the most-read page is often not the website itself but the map listing. That is why activity on the map listing belongs inside the social media report, not outside it.

The items that go into the report are few, and all of them sit in the demand layer: calls made from the profile, direction requests, clicks through to the website and the number of new reviews. View counts belong to the reach layer and stay on the lower lines.

The most critical distinction here is this: when the information on the map listing contradicts what is on the website and the social profile, it is usually the listing that people read. That is why consistency of business details goes into the report as a measurement item on a line of its own — how many fields are inconsistent across how many channels, and how many of them were fixed that month.

Monthly report template: nine lines, one page

The longer a report gets, the less likely it is to be read. The nine lines below fit on a single page, and every one of them can be tied to a decision.

1. Total number of requests this month and the split by channel.
2. Type of request: price, appointment, information, complaint.
3. Number of requests that turned into an outcome.
4. Visits from the profiles to the website, and which link they came through.
5. Map listing: calls, direction requests, website clicks, new reviews.
6. The return-by-content-type table.
7. Number of pieces published and how closely the calendar was followed.
8. Reach and engagement (lower line, as explanation).
9. What we are changing this month: three items at most.

The ninth line is the only mandatory line in the report. A report that proposes no change is an archive document. Capping it at three items is deliberate as well: a longer list cannot be carried out within the month, and the same items end up written again the following month. A recommendation that never gets implemented is the fastest route to a report nobody reads.

When measurement changes a content decision

The purpose of measurement is not to push the numbers up but to correct the calendar. The decision is made in three situations.

If a content type has brought in no demand two months running, that type comes off the calendar and one more piece of a type that does bring demand takes its place. What makes this decision hard is the effort already invested; what makes it easy is having the criterion written down at the start of the month.

If a topic brings in demand on both channels, a second and a third piece are produced around that topic. A topic that brings in demand is a seam you struck by chance, and it deserves to be mined deeper.

If reach falls while demand holds steady, nothing gets changed. This is the most frequently misread pattern: the drop in reach is treated as a problem by reflex, and a content type that works is swapped out for no reason.

The limits of measurement have to be accepted as well. Part of the effect of social media is delayed and indirect: someone who sees the profile today searches for the brand name three weeks later. Ways of reading delayed signals like these are a separate discussion, and we covered them in our article on measuring demand in the zero-click era. To follow what happens to demand after it arrives — repeat customers, churn, customer value — see our article on customer data metrics for small businesses.

If you would like to rebuild your report around these three layers and fit demand logging into the way your business already works, take a look at our social media management service or write to us directly.

Frequently Asked Questions

Does follower count not matter at all?

Follower count is not an outcome but an accumulation; it cannot serve as a business metric on its own, because it can be bought, inflated and can grow without ever turning into demand. The one place it means something is when comparing the same content type over time: if more demand comes in with the same number of followers, the content has got better. If it goes into the report at all, it belongs on the lower line next to the reach and engagement layer; it should not be the first number the business owner sees.

After how many months does meaningful data build up?

It is more accurate to give a criterion than a number: before you can decide about a content type, several examples of that type need to have been published, and the demand each of them produced needs to have been recorded. In a business that publishes only a few pieces a month, this naturally takes longer. Deciding on the strength of a single post's performance is premature in every case; looking by type and across at least two consecutive periods is the simplest way to filter out chance.

Can we measure without buying a paid tool?

Yes, and for most small businesses that is the right place to start. The three things you need are free: the habit of tagging links, an analytics tool installed on the website, and a shared spreadsheet where demand arriving through messages is written down. Paid tools make these three easier and speed up reporting, but they do not create a logging routine that is not there. Confirming that the logging routine can actually be kept up for a month before you buy a tool saves you from an unnecessary subscription.

If the same customer came from both an ad and an organic post, which one gets the credit?

There is no single right answer; what matters is that the rule is written down at the outset and not changed from month to month. There are two common approaches: credit the last touchpoint the request came through, or mark both channels as having "contributed" and keep that in a separate column so the total does not get inflated. At small volumes the second approach teaches you more, because it shows how many touches the customer really needed before deciding. When the rule changes, comparison with earlier months breaks down; that is why reviewing the rule once a year is enough.

Reach dropped but demand stayed the same: is something wrong?

On its own this is not a sign of a problem, and it is often a good sign: fewer people may have been reached, but the people who were reached may have had a stronger interest in the business. Changing the content type by reflex is the most damaging move you can make here, because a setup that works gets broken by a misread measurement. The right response is to keep watching the demand layer and to step in only if the drop in reach starts to show up in demand as well. Reach is not a target to be chased for its own sake but an explanatory item to look at when demand falls.