Three Goals, Three Scoreboards for Social

Most social media reports fail for one reason: they report everything. Reach, impressions, likes, follower growth, link clicks, engagement rate, saves, story replies — twelve numbers in a slide, none of them tied to a decision you were trying to make. You read it, you nod, you file it, and next month the same twelve numbers arrive with slightly different values.

The fix is not a better dashboard. It is deciding, before the campaign runs, which of three jobs it is doing. A post that exists to make people in Laxmi Nagar aware that your clinic exists is not the same as a post meant to make a shortlisted buyer finally choose you. They work differently, they take different amounts of time, and they leave completely different traces in your analytics. Judging one by the other's numbers is how good work gets killed and bad work gets renewed.

Awareness: did the right people see it, enough times?

Awareness work has one job — putting your name in front of people who did not know it, often enough that it sticks. Nobody buys from the first exposure. So the honest awareness questions are about coverage and repetition, not about response.

What to read:

• Reach, not impressions. Impressions count showings. Reach counts people. If you reached 8,000 people with 40,000 impressions, each person saw you five times. That is a frequency of five, and whether it is useful or wasteful depends entirely on how long your buying cycle is.
• Frequency. This is the number most agencies skip because it tends to embarrass the media plan. A small local audience plus a decent budget plus a month of running gets you frequencies in the teens very fast, at which point you are paying to annoy the same 3,000 people.
• Video completion, defined honestly. Meta's own documentation defines a ThruPlay as a video watched to the end, or for at least 15 seconds if it is longer. A "video view" on most platforms triggers in a couple of seconds and tells you almost nothing. Ask for ThruPlay or equivalent, and ask for the drop-off curve — where people leave tells you whether the first three seconds are working.
• Branded search volume. This is the one genuinely external check on whether awareness is landing. Open Google Search Console, filter queries to your brand name and its common misspellings, and watch the impressions line over months, not weeks. If people are starting to search for you by name, something upstream is working.
• Direct and unattributed traffic. A WhatsApp forward usually arrives in Google Analytics as direct traffic with no referrer. So does someone typing your name into the address bar. Rising direct traffic in a month when you ran heavy awareness work is weak evidence, but it is evidence.

What not to buy here: follower growth as a deliverable, and giveaway contests run to inflate it. A giveaway reliably produces followers who want a free blender. They cost you nothing at signup and quietly depress every engagement figure afterwards, because the platform now shows your posts first to an audience that has no interest in them. Agencies sell follower targets because they are easy to hit. They are the single least informative number on social.

Consideration: is anyone leaning in?

Consideration is the middle. Someone knows you exist and is now deciding whether you are worth their time. The signals here are small, deliberate actions that cost the viewer something — a tap, a save, a message, a forward.

Read these:

• Saves and shares, above likes. A like costs nothing. A save means "I want this later". A share means "I am willing to put my name next to this". For a content studio, a furniture showroom, a CA practice — saves are the closest thing social gives you to a bookmark.
• Profile visits and link taps from a specific post. Most platforms will break this down per post. A post with modest reach and a high profile-visit rate is doing more for you than a viral one with none.
• Conversations started. Direct messages, WhatsApp click-throughs, story replies, comment questions with real content in them. For a lot of Indian small businesses the actual funnel runs through WhatsApp, so a Click-to-WhatsApp ad's message count is the consideration metric, full stop.
• Behaviour after the click. If social sends someone to your site, what do they do? Pages per session and time on page for social traffic specifically, segmented from search traffic. Social visitors who bounce in four seconds were never considering anything; you bought a tap.
• Returning visitors from social. Someone who comes back a second and third time is deciding.

Consideration is also where your site has to hold up. A reel can do everything right and still lose the person if the landing page takes six seconds on a 4G connection. Google's published Core Web Vitals thresholds give you a usable bar: Largest Contentful Paint under 2.5 seconds, Interaction to Next Paint under 200 milliseconds, Cumulative Layout Shift under 0.1. Test on mobile, on mobile data, not on your office Wi-Fi. If you are failing those, fixing them is better value than more ad spend — our note on auditing your own website in an afternoon and our web development work both start in the same place.

Conversion, and why the number will never be clean

Conversion is the only stage where the metric is a business outcome: an enquiry, a booking, a sale, a signed quote. It is also the stage where measurement is hardest, and where you should be most suspicious of confident numbers.

Here is the mechanism, plainly. Platforms report conversions they can see. Meta sees a conversion if the person clicked the ad and the Meta pixel — a small piece of tracking code on your site — fires on your thank-you page within the attribution window. Google Analytics 4 attributes by last non-direct click by default, so if the person saw your reel on Tuesday, searched your name on Thursday and bought on Saturday, Analytics credits search and social gets nothing. Meanwhile Meta may credit itself for that same sale. Both reports are internally consistent. Both are partial.

Then add India-specific leakage. Apple's App Tracking Transparency prompt means a large share of iPhone users are not trackable across apps at all. WhatsApp forwards strip the referrer. Someone sees your ad, calls the number on your Google Business Profile, and the entire transaction happens on a phone call with no digital trace.

So what do you actually do?

1. Tag every link you post with UTM parameters. These are the ?utm_source=instagram&utm_campaign=diwali bits on the end of a URL. They cost nothing, they survive most of the mess, and without them your Analytics will lump half your social traffic under "direct".
2. Ask the customer. One field on the enquiry form — "how did you hear about us" — with a short dropdown. It is imperfect, people misremember, and it is still the best cross-check you have against platform self-reporting. Push the answer into your CRM so it sits next to the deal value, otherwise it is a number nobody ever reads.
3. Run a holdout when the budget justifies it. Switch the campaign off in one city for three weeks and watch total enquiries, not attributed enquiries. A clumsy test on real totals beats an elegant model on partial data.
4. Measure cost per qualified enquiry, not cost per lead. Ninety leads at ₹40 is worse than twenty at ₹180 if eighteen of the twenty can afford you.

What not to buy: a report that adds Meta-attributed conversions to Google-attributed conversions to arrive at a total. That total is double-counted and meaningless. Also be wary of anyone who presents conversion attribution as settled. The honest position is that you have several partial views and you triangulate.

The one-page version

Two more rules worth holding to. First, pick one goal per campaign and say so out loud before it runs. A campaign asked to do awareness and conversion simultaneously will be judged on conversion and cancelled. Second, write the target down. "Reach 20,000 people in East Delhi at a frequency under four" is a brief you can grade. "Increase brand visibility" is not.

What to do this week

Open your last three monthly social reports side by side. For each number on them, ask: if this went up 30%, what would I do differently? Delete every number that has no answer. You will usually be left with three or four, and those are your actual scoreboard.

Then fix the plumbing — UTM tags on every link you post, a "how did you hear about us" field on the enquiry form, and the answers flowing into wherever you track deals. That takes an afternoon and improves every report you will ever read afterwards.

If you want a second pair of eyes on what your current reporting is actually telling you, or help wiring enquiry sources through to a system you will genuinely use, have a look at our social media work and automation, or just get in touch and describe what you are trying to achieve.

Questions we get asked