How to Track Which Posts Drive Sales, Not Just Clicks

Here is the situation almost every seller and creator ends up in. You have one thing to sell and one URL for it. Over a fortnight that URL goes into your Instagram bio, a pinned comment under a Reel, a YouTube description, an X thread, a Reddit comment, and a partner's newsletter. Sales come in. Your analytics says most of your traffic was social. Your bank says you made some money. Neither one tells you which of those six placements produced the buyers.

So you guess. Usually you guess with clicks, because clicks are the number you can see. And that guess is very often wrong, because attention and intent are not the same thing.

That flip is the normal case, not a freak result. A bio link collects idle taps from people browsing your profile. A partner's newsletter reaches people who already trust the person recommending you. One produces traffic, the other produces customers, and a click count cannot tell them apart.

This guide is the method for telling them apart. It is not a tool list. It works whether you sell through Shopify, Gumroad, Stripe, or a booking form, and it works with no ad spend and no CRM. If you want the definition of the underlying idea first, the conversion tracking glossary entry covers that in a couple of minutes. This is the operating manual.

Every tracking guide on the internet starts with tagging. That is the wrong first step. Tagging tells you what to call a thing; you first have to decide what the things are.

A placement is a spot, not a platform

"Instagram" is not a placement. Your bio link is a placement. The link in the pinned comment under last Tuesday's Reel is a different placement. The one you send in DM replies is a third. All three are Instagram, all three behave completely differently, and lumping them together destroys the only insight worth having.

The test is simple: if you would consider doing more of one and less of another, they are separate placements. That usually lands you somewhere between five and fifteen, which is a manageable number of short links pointing at one destination.

Name them so they still make sense in six months

Pick a pattern and never break it. Something like product-channel-spot works for almost everyone: spr-ig-bio, spr-ig-reel12, spr-yt-desc, spr-nl-footer. You will read these in a list months from now, half-remembering what you did, so favour boring and readable over clever.

Do not forget the placements that are not posts

The ones people miss are the quiet ones: a link in your email signature, a QR code on packaging, the pinned message in a Discord server, a line in your podcast show notes. These are often the best performers precisely because nobody optimises them. Give them links too. If a placement is physical, the same logic applies to tracking QR code scans.

This is the confusion that costs people the most, and almost nothing on the internet clears it up: guides tend to pick a camp and argue for it. They are not competing answers. They answer different questions and they belong on top of each other.

What the short link is good at

It is short, so it survives being read aloud, screenshotted, or retyped. It stays yours when a platform rewrites or wraps outbound URLs. It works in places where a long tagged URL looks like spam. And it counts the click at the moment it happens, before any of the visitor's browser settings get a vote. On your own branded domain it also tends to earn more clicks than a generic one.

What the UTM is good at

UTM parameters travel into the destination, so Google Analytics, Shopify, or Stripe can attach them to a session and then to an order. That is the half a short link cannot do on its own. Build them without typos with the free UTM builder, check inherited ones with the UTM parser, and use the complete UTM guide for the naming conventions. If you are cleaning up an existing mess, the UTM audit workflow is the faster route.

Layer them

Put the UTMs on the destination URL, then shorten that. Your short link stays clean and memorable, and the tags arrive intact on the other side. Short links carry tracking parameters through the redirect, so nothing is lost in the hop.

Choosing between a short link and a UTM is like choosing between the envelope and the address. You need the envelope to get it carried and the address to get it delivered.

Placements and tags get you to "this many people came from here". The last mile is joining a click to an actual purchase, and that needs one more piece: an identifier that leaves with the click and comes back with the sale.

How the round trip works

When someone clicks a link with conversion tracking switched on, the redirect appends a random id to your destination URL. A small script on your site remembers it in first-party storage. When that visitor reaches your thank-you or order-complete page, the script reports the id back with the order value, and the sale lands on the exact link that earned it. Setup is two copy-paste lines, and there is a button that copies the full instructions if someone else manages your site.

Why this is not a third-party cookie

Nothing about this identifies a person. The id is meaningless outside your own account, it is stored on your own domain rather than a shared advertising one, and it carries no profile, no email, and no device fingerprint. It answers one question: did that click end in a sale. That is also why it keeps working while third-party cookies are being switched off around it.

The window, and the Safari caveat

A click can convert for 30 days. Safari and iOS cap script-set storage at around seven days, so a Safari visitor who buys three weeks later will show up as an untracked sale. Treat late Safari conversions as a bonus rather than a bug, and never assume your tracked total is your total revenue. It is the attributable share, and it is always a little smaller than the truth.

The method is the same everywhere, but what you do with the answer differs a lot by situation. These are the five that come up most.

1. The creator with a link in bio

You have one bio slot and you keep changing what is in it. Give each destination its own link rather than editing one, so you can compare "the course link" against "the merch link" over months instead of losing the history every time you swap. Pair it with a deliberate link-in-bio strategy, and remember that bio traffic is the highest-volume, lowest-intent placement you own. Judge it on sales, or it will always look better than it is.

2. The seller posting the same product six ways

This is the case in the chart at the top. One product, six placements, and your instinct says to keep making more of whatever gets the most clicks. Once you can see revenue per placement, the instruction is usually the opposite: quietly stop doing the thing that gets attention and do more of the thing that got you paid. If two placements look genuinely close, split test them at the redirect layer instead of arguing about it.

3. The affiliate sending traffic to someone else's checkout

You cannot put a script on a merchant's thank-you page, so the browser method is unavailable to you. What you can do is pass an identifier into the network's sub-id field and have the network report the sale back to you server-side when it is approved. That is a different mechanism with its own vocabulary, and it is covered end to end in the postback URL guide. The strategy layer, sub-ids, disclosure, and cloaking, lives in the affiliate link tracking guide.

4. The newsletter and community operator

Your email platform reports clicks per campaign, which dies with the campaign. A short link outlives it, and it keeps counting when a subscriber forwards the issue or pastes your link into a Slack channel. That is exactly the traffic your email platform cannot see. There is a full walkthrough in newsletter click tracking.

5. The placements you simply cannot instrument

Some destinations will never accept a script: a Gumroad or Stripe payment link, a marketplace listing, an app store page, a partner's own site. Do not fake it. Use one of the two honest fallbacks instead. Give each placement its own discount code and read the code breakdown in your payment processor, or point each placement at a page you do control that then forwards to the uninstrumentable destination. Either way, say out loud which placements are measured and which are estimated, so you never quietly compare the two.

Once the data arrives, the failure mode changes. It stops being "I have no idea" and becomes "I am confidently wrong", which is worse. Three habits prevent most of it.

Divide by human clicks, not raw clicks

Conversion rate is sales divided by clicks, and the denominator decides whether the number means anything. Automated traffic inflates it: scanners, prefetchers, and preview bots all register as clicks. Flyn separates human clicks from bot clicks on every link for free, and the rate you should quote yourself is the one computed against humans. The click fraud guide goes deeper on what the automated share is made of.

Know how many clicks make a finding

Nobody publishes this and it matters enormously. Comparing two placements on 40 clicks each is reading tea leaves. A rough working rule:

If a placement never reaches 100 human clicks, that is itself the finding: it is too small to matter, and no amount of analysis will change that.

Measure your own conversion lag

People rarely buy on the first visit. Look at the gap between click and purchase across your last few dozen sales and find the middle value. If your median lag is six days, then judging a placement 24 hours after posting is guaranteed to be wrong, and slow-burn placements like show notes and YouTube descriptions will always look worse than they are. Set your judging window to at least twice your median lag.

The moment you have three sources of truth you will notice they disagree, and the usual reaction is to assume something is broken. Usually nothing is. Each number counts a different event.

Why clicks exceed analytics sessions

The big one is link preview bots, and it is the single most common cause of "hundreds of clicks and zero sales". Slack, Discord, iMessage, WhatsApp, Telegram, X, and Facebook all fetch a URL to build the preview card when someone shares it. One paste into a busy Slack workspace can produce a burst of fetches that no human made. Browsers also prefetch links, ad blockers stop analytics scripts from ever running, and some visitors leave before the page finishes loading.

Why sessions exceed orders

Because most people who visit a page do not buy, which is normal and not a tracking failure. What is worth watching is the shape over time. If sessions hold steady and orders fall, that is a page or pricing problem. If both fall together, it is a traffic problem, and you should be looking at the placement, not the page.

What a healthy gap looks like

There is no universal number, so establish your own baseline over a few weeks and watch for movement instead of chasing a target. A consistent 20 to 30 percent drop from clicks to sessions is unremarkable. A jump to 70 percent in a week says something changed: a new placement in a chat-heavy channel, a broken redirect, or a script that stopped loading. The direction of the change is the signal, not the absolute number.

Measurement that does not change behaviour is a hobby. Once you can see revenue per placement, three decisions follow, and they are worth making on a schedule rather than a whim.

Cut the ones that only look busy

The placement with lots of clicks and no sales is not free. It costs you the time you spend making content for it. Cut it, or change what it points at: sometimes a high-click, low-sale placement is fine and simply pointed at the wrong destination, and sending it somewhere earlier in your funnel fixes it.

Repeat what actually paid

If a partner newsletter produced 41 sales from 380 clicks, the obvious action is more partner newsletters, not more posts. This is where most of the value of the whole exercise sits, and it is almost always an instruction to do less of something loud and more of something quiet.

Re-test on a schedule

Placements decay. Feeds change, audiences move, and a bio link that carried you last year may be dead now. Re-read the scoreboard monthly rather than continuously, because continuous watching is how you end up reacting to noise. Monthly is often enough to catch a real shift and rare enough that you are looking at meaningful sample sizes.

None of this requires ad spend, a data team, or a new stack. It requires deciding what your placements are, giving each one its own link, connecting the click to the order, and being honest about how much of what you see is real. If you want to run it on Flyn, conversion tracking is included on Pro, click counts and the human-versus-bot split are free on every link, and you can create your first tracked link in under a minute. Creators can start from the creator setup, and stores from the ecommerce setup.