Destroy the like button: what to measure on social instead

There is a line on our homepage that reads "Our mission: destroy the like button." Below it, at the very bottom of the page, there is a like button. It does nothing. People press it anyway, which is the point.

We put it there because the like is the most misleading number in marketing, and because every brand we have ever met has, at some stage, judged a month's work by it. This is the argument for stopping, and for what to count instead. It is written from the agency side, where we see the monthly report land on a founder's desk and watch their eyes go straight to the likes.

The short answer

A like is the cheapest gesture a person can make towards a brand, so it tells you the least. Measure the actions that cost the viewer something: a save, a share, a visit to your profile, a click to your website, a message, an enquiry, a booking. Give every post one job and one metric. Wire social to the business through a small number of tracked routes (a tagged link, a lead event, a code, a phone number) so that once a month you can say what the channel brought in, in rupees or in patients or in bookings. Everything else in the report is context.

What a like measures

A like measures that a thumb moved. It takes less than a second, costs nothing, commits the person to nothing, and is visible to nobody who matters. People like posts out of politeness, out of habit, because the picture was pretty, because a friend made it, because they were half-asleep. None of those reasons has anything to do with buying.

Instagram knows this. In May 2021 it gave everyone the option to hide like counts, on their own posts and on everybody else's, and explained that the idea was to let people "focus on the photos and videos being shared, instead of how many likes posts get." The platform that invented the double tap decided five years ago that the number was a distraction. Brands are still putting it at the top of the report.

The likes a brand collects also come overwhelmingly from the people it already has. Followers like. Strangers scroll. A post with 400 likes from 400 followers has reached the people who were already convinced, and a report that celebrates it is a report on how polite your existing customers are.

The ladder of cost

Every action a viewer can take has a cost to the viewer, and the cost is the signal. Here is the ladder as we use it, from the cheapest rung to the most expensive.

A like costs a thumb twitch. A comment costs a few seconds and a little exposure. A save costs nothing visible to anyone, which is exactly why it is honest: nobody saves a post to be polite. People save a post because they intend to come back to it, and intention is what you are selling. A share or a send costs reputation, because the person is attaching their name to your content in front of someone they know. A profile visit costs attention and curiosity. A click to your website costs leaving the app, which people hate doing. A message or an enquiry costs identity; the person has stepped out of the crowd and told you who they are. A booking or a purchase costs money.

Read a month's work against that ladder and the picture changes. A post with 80 likes and 300 saves has done something. A post with 3,000 likes and 4 saves has been decorative. A post that produced eleven messages, nine of them asking about price, has done more for the business than anything else that month, and it may well have had the fewest likes of the lot.

Healthcare taught us this faster than any other category. People do not like a post about infertility or breastfeeding or a child's surgery in public, because a like is public and the subject is private. They save it, and they send it to one person, and they message. A fertility clinic that judges its content by likes will conclude that nobody cares. The saves and the inbox say otherwise.

Views are not impressions, either

The breastfeeding series we made for Motherhood Hospitals, #CareInEveryLatch, shows 1.66 million impressions and 14.8 million views. People sometimes read those two numbers as a contradiction. They are not. Impressions count the times a post appeared in front of someone. Views count the times it was played. Fourteen million plays from under two million appearances means people were playing it again. For a series whose job was to answer new mothers' questions, repeat watching was the whole result. A like count would have told us nothing about whether the content worked. The replays did.

The lesson generalises. Watch time, completion rate and replays are measures of whether the content did its job. Likes are a measure of whether it was pleasant.

One post, one job, one metric

The discipline that fixes most reporting is embarrassingly simple: before a post goes out, write down what it is for, and name the single number that will show whether it did it.

A post meant to reach new people is judged on reach among non-followers and on shares. A post meant to teach is judged on saves and watch time. A post meant to start conversations is judged on comments and replies, and on their quality rather than their count. A post meant to sell is judged on clicks, enquiries and whatever happens after. A post meant to make the brand feel like itself, which is a legitimate job, is judged on whether people who know the brand recognised it, and you will see that in the comments and the sends.

Trying to make every post do every job is how feeds become beige. Giving each post one job is how a month's content starts to look like a plan. The test we apply to our own reports is the one we suggested in How to choose a social media marketing agency in Bangalore: which number would we expect to be fired over? If the honest answer is none of them, the report is decoration.

The plumbing

None of this works without connecting social to the business, and the connections are cheaper than brands assume. Five of them cover almost every case.

1. Tagged links. Every link from a bio, a story or an ad carries parameters that tell your analytics where the visitor came from. This is ten minutes of setup and it separates "traffic" into "traffic from this post".
2. A lead event. When someone completes the enquiry form, books an appointment or starts a WhatsApp conversation from the site, your analytics records an event. On our own website the number we care about is completed briefs, because nothing else on the site is a sale.
3. A route that only social knows about. A code, a landing page, a dedicated phone or WhatsApp number used nowhere else. When someone uses it, you know.
4. One question at the point of sale. "How did you hear about us?" asked by the receptionist, the sales team or the order form, written down every time. Imperfect, and still the best cross-check there is.
5. A monthly reconciliation. Sit the social report next to the sales or bookings data and look at the same month in both. Most of the value of social shows up here, as a pattern across months, rather than in any one post's numbers.

For a clinic, the final metric is patients who booked. For a café, covers. For a D2C brand, orders and the cost of each one. For a B2B firm, qualified conversations. Work backwards from that figure to the ladder, and the report writes itself.

What this does to the work

When a brand stops asking for likes, the content changes within a month. Nobody makes a bland inspirational quote for saves; nobody makes a stock-photo festival greeting to get sent. The team starts making things that are useful, specific and a little risky, because those are the things people keep and pass on. Our best healthcare work has never been likeable in the Instagram sense. A courtroom drama about fertility food myths is not a pretty post. It was watched 1.1 million times.

The like has one honest use left: as an early read in the first hour, when a post is either going out to new people or not. Treat it like a smoke detector. It tells you something is happening. It cannot tell you what.

The button at the bottom of our homepage

It still does nothing. People keep pressing it, which we find funny and slightly damning. If you want to argue about it, or you want a report that puts the enquiries at the top and the likes somewhere near the footer, the brief takes two minutes.

Niraj Patadiya is a co-founder of Social Saga.

Frequently asked questions

What are vanity metrics in social media?

Vanity metrics are numbers that look like progress but do not connect to a business outcome: likes, follower counts, raw impressions and total engagement. They rise with pleasant content and existing audiences, and they can be bought. The alternative is to measure actions that cost the viewer something (saves, shares, clicks, messages, enquiries, purchases) and to track them through to sales.

Which Instagram metrics matter most for a business?

Reach among non-followers for growth, saves and watch time for content that teaches, shares and sends for word of mouth, profile visits and link clicks for intent, and messages, enquiries and bookings for revenue. Impressions and likes belong at the bottom of the report as context.

How do you measure social media ROI?

Connect social to the business through tagged links, a lead event in your analytics, a code or phone number used only on social, and a "how did you hear about us" question at the point of sale, then reconcile the social report against sales or bookings every month. Divide what the channel brought in by what it cost, including the agency fee, production and ad spend.

Do likes affect the Instagram algorithm?

Platforms weigh many signals, and a like is among the weakest of them. Actions that take more effort, such as watching a video to the end, saving, sharing or sending it, say more about whether content held attention. In our experience posts that get saved and sent travel further than posts that only get liked, and Instagram itself has let users hide like counts since 2021.

How often should a social media report be produced?

Monthly, with a short weekly check on anything live. The report should name each post's job, show the one metric that measured it, and end with what changes next month. A report that only shows totals going up is describing the weather.

Is a high engagement rate always good?

No. Engagement rate counts likes and comments as equal to saves and shares, so a post with thousands of emoji comments and no saves can score higher than a post that sent fifty people to your website. Look at which engagements you are getting and what each one costs the viewer; the total hides that.