Before product-market fit, most social media metrics measure the platform, not the business. Follower counts and impressions rise and fall for reasons unconnected to the only question that matters at this stage: are the specific people you want as customers noticing, and are any of them talking to you?
Four numbers answer that question, and all four are countable by hand: replies from people who match your customer profile, clicks from your profile or posts to your site, DM conversations with target customers, and waitlist or trial signups you can trace back to social. Everything else can wait until the sample sizes stop being a joke.
Why follower count and impressions mislead at small scale
At small scale, variance swamps signal. When one post gets 300 impressions and the next gets 3,000, the difference is usually a feed test, a timing accident, or one larger account replying, not a verdict on the content. Reading strategy lessons out of week-to-week impression swings on a 200-follower account is astrology with a dashboard.
Follower count has a subtler problem: composition. Early followers skew toward adjacent audiences, other founders, friends, and people farming follow-backs. An account can grow to 2,000 followers with fewer than 50 potential customers in the audience. Optimizing for the total actively selects for content the adjacent audience likes, which pulls you away from the narrower posts your buyers would notice.
Impressions add optimization pressure in the same direction. Feeds reward what feeds reward, and broad relatable content will always out-impress a specific post about the workflow your product fixes. If impressions are the score, you will drift toward a general audience that will never buy anything from you.
The one useful thing to do with your follower list is an occasional composition check. Scan the most recent 50 followers and count how many plausibly match your customer profile. That ratio, not the total, tells you whether your content is finding buyers or just finding company.
The four metrics that predict anything
Each of the four shares one property: it names a person you can go talk to. That is the tell for a useful pre-launch metric, because social at this stage is customer development with distribution attached, and a metric that cannot lead to a conversation is decoration.
- Target-customer replies: replies from people who match the profile you sell to. Total reply count flatters; three replies from actual prospects beat thirty from other founders. Keep a running list of names.
- Profile-to-site clicks: clicks from your bio link and post links to your site, tagged so your analytics can attribute them. A click is a cost the reader chose to pay; a like is not.
- DM conversations: new conversations with target customers, in either direction. These convert to calls, design partners, and first customers at a rate no public metric approaches.
- Qualified waitlist signups: signups that match your customer profile and trace back to a post or a conversation. Unqualified totals are the vanity metric wearing a conversion costume.
The weekly scorecard
Five minutes on Friday, counted by hand. At this scale, manual counting is a feature rather than a chore, because it forces you to look at who replied instead of how many did.
| Metric | Where to count it | What counts | What to write down |
|---|---|---|---|
| Target-customer replies | Notifications, by hand | A reply from someone matching your customer profile | Name, which post, what they said |
| Profile-to-site clicks | Site analytics, tagged links | Sessions arriving from your tagged social links | Count, and which post or bio drove them |
| DM conversations | Inbox | A new conversation with a target customer, either direction | Name and topic |
| Qualified signups | Waitlist or signup form | A signup matching the profile, traceable to social | Count and the source post |
A scorecard that teaches
Week of Aug 17: 4 target replies (2 on the pricing-mistakes post), 11 tagged clicks, 1 DM about migrating off a spreadsheet, 2 qualified signups. Pricing content is finding buyers; ship the follow-up post.
A scorecard that flatters
Week of Aug 17: 9,400 impressions, up 40 percent. Momentum!
Calibrate what a good week looks like
Absolute numbers stay small for a long time, and that is fine. Three target-customer replies and one real DM conversation is a good week at 150 followers; it means the right people can find you and consider you worth answering. Compare month over month rather than week over week, and watch the names as much as the counts. The same prospect replying three times across a month is a stronger signal than three strangers replying once.
Attribution needs one piece of light plumbing: tag the links you post so profile-to-site clicks are visible in your analytics instead of dissolving into direct traffic. That is a fifteen-minute setup job, covered in the UTM guide linked below.
The scorecard sharpens further if each week carries one written hypothesis. Before posting, note what you expect: 'the migration post should draw replies from ops leads.' Friday's count then confirms or kills a specific belief about your audience instead of drifting into general score-keeping, and a killed hypothesis is worth more than a flattering week.
When to ignore analytics entirely
- Your first 90 days of posting. The only goal is the publishing habit, and any dashboard will only tell you that you are small.
- Any single week. One week is one sample; act on patterns that survive a month.
- Launch weeks. A spike tells you about the launch, not about your baseline content.
- When a platform visibly reshuffles its feed; wait for the new normal before reading trends.
- Whenever measuring takes longer than a scorecard. If the review outgrows five minutes, the process has become procrastination with charts.
Close the loop
The scorecard only matters if it feeds decisions. Let the posts that produced conversations set next month's topics, route the click data through proper link tagging, and revisit which account should be doing the talking as the numbers shift between your personal profile and the company's.