TikTok ads, measured where the sale actually happens

TikTok counts what happens in its app and on your website, within a window it lets you choose. A business that closes by phone or in person needs a different number: TikTok leads matched to closed sales by phone and email, reported as a floor, with the platform's own figure kept beside it as a signal about the creative.

TikTok has become a channel that local and sales-led businesses take seriously. Roofers film tear-offs, dentists explain aligners, car dealers walk the lot on camera, and some of it brings in customers. The question every owner eventually asks is the obvious one: how many, and what were they worth?

The platform will answer that question with confidence. Its answer is built from the conversions it can see, and for a business whose sale happens on the phone, at a counter or in a quote meeting, the conversion it can see is rarely the sale. This post sets out what TikTok actually counts, where that count goes wrong in both directions, and the route to a revenue figure you can defend.

What TikTok counts as a conversion

A TikTok conversion is an event the platform observed or was sent — a form submitted in the app, a pixel event, an Events API event — within an attribution window after somebody clicked or watched an ad. For web advertisers the default is seven days after a click and one day after a view.

Every advertising platform measures itself this way, and TikTok is more transparent than most about the settings. The settings still matter more than most advertisers realize, because they decide which sales belong to the ad before any sale has happened. We have written about choosing a window you can defend; TikTok's defaults are a starting point for its inventory, not a fact about your customers.

View-through credit deserves particular attention on a platform built on autoplay video. Nobody chooses to see a TikTok ad; it arrives in the feed. A one-day view window therefore credits the ad with any conversion from anybody who scrolled past it the day before, whether or not it registered with them. TikTok has added a stricter variant that only counts views with some attention behind them.

Six seconds of attention is a better threshold than none, and it is still not a sale. It tells you that somebody watched and later did the thing you were tracking. It cannot tell you whether they would have done it anyway, which is a question for a holdout test, not for any attribution setting.

A TikTok conversion is a rule applied to the events TikTok can see. Change the window or the view setting and the same month produces a different number. Report the rule beside the figure, every time.

Why a business that closes offline sees the wrong number

For a business that sells by phone, in person or through a sales team, TikTok's number is wrong in both directions at once. It misses the customers who acted somewhere TikTok cannot see, and it counts leads and view-throughs that never became revenue. The two errors do not cancel; they just make the figure impossible to reconcile.

Consider the paths a TikTok viewer actually takes to becoming a customer of a home services company or a med spa:

• They submit a form inside the app. TikTok sees it and counts a lead. Whether it became a $9,000 roof or a number that never answers, TikTok does not know.
• They call the number in the video. TikTok sees nothing, unless the number is a tracking number and somebody connects its calls back.
• They remember the name and search for it later. The sale arrives as branded or organic search, and TikTok gets no credit anywhere — the same pattern we described for YouTube and other video.
• They walk in. Nobody records how they heard about you unless the front desk asks.
• They were going to buy anyway. They scrolled past the ad yesterday, filled in a form on your site today, and TikTok's view window claims them.

The first and last paths inflate TikTok's number; the middle three deflate it. Which effect dominates depends on the business, the creative and the offer, and nothing inside Ads Manager will tell you. The only way to find out is to look at the sales.

Give TikTok a door you can see through

The fix is not a better pixel. It is a response route that belongs to TikTok alone and leaves a phone number or an email behind: an instant form, a phone number used nowhere else, or a landing page that writes its source onto the lead. Whatever comes through that door can be matched to a closed sale exactly.

Instant forms

TikTok's lead generation forms open inside the app and can pre-fill contact details for logged-in users, which is why they convert well. The same convenience means the details belong to the TikTok account rather than to the purchase. As with Meta's lead forms, a meaningful share of emails will be addresses the customer does not use when they buy, so make the phone number a required field. Phones match across systems far more reliably than personal email addresses.

Download the leads at least weekly, or connect them to your CRM. TikTok's leads center is not where your sales team works, and a lead that sits there for four days is a lead somebody else has already called.

A phone number that appears nowhere else

If a video says "call us", the number on screen should exist only in TikTok creative. Put it on your website as well and its calls stop meaning anything. A tracking number from your existing call tracking provider is enough; one per platform is plenty for most local businesses, and one per campaign is only worth it if you will actually act on the difference.

A landing page that records where the lead came from

For ads that send people to your site, use campaign parameters on every link and make the form capture them into hidden fields, so the source is written onto the lead record at the moment it is created. A source recorded on the lead survives a CRM migration, a cookie banner and a change of analytics tool. A source reconstructed later from session data usually does not.

Measure TikTok from your side of the wall. A form, a number or a page that only TikTok uses gives you leads with phone numbers and emails attached, and those can be matched to closed sales without anybody's model in between.

From reported conversions to matched revenue: one month, worked through

The matched figure starts from the same spend as TikTok's figure and ends somewhere very different. Walking one month through both, line by line, shows where the gap comes from and why only one of the two numbers can be reconciled with the books.

Suppose a remodeling company spends $6,000 on TikTok in a month. Ads Manager reports 140 conversions: 102 instant-form submissions and 38 view-through conversions on the website form. The company also gave its videos a dedicated phone number, which took 22 calls that TikTok never saw. Average closed job value is $2,400.

Three things stand out. First, the 38 view-through conversions are not in the matched figure at all, because those people became leads through the website form and are credited to whatever source that form recorded. If some of them were genuinely persuaded by TikTok, the matched figure undercounts; that is the trade for a number that does not guess.

Second, the 22 calls are in the matched figure and nowhere in TikTok's. In a business where people prefer to call, those can easily be the best leads the channel produces, and the platform report has no idea they exist.

Third, the two cost figures describe different things. $42.86 is what a TikTok event cost. $666.67 is what a TikTok customer cost. Only the second can be compared with your margin, and with the cost of a customer from every other channel.

Cost per conversion tells you what TikTok charged for an event. Cost per matched sale tells you what it charged for a customer. A budget decision made on the first is a decision about a number your accountant has never seen.

Sending the sales back to TikTok

Once you have matched sales, you can return them to TikTok as offline events, so its delivery system learns which viewers become customers rather than which ones fill in forms. TikTok accepts offline events keyed on email or phone number, by API, through a partner, or as a manual CSV upload.

This is the same loop we described for Google and Meta's offline conversions, and the same cautions apply. Upload only what you have matched with confidence, upload it promptly so it falls inside the window you set, and hash the identifiers unless you have a reason not to. You are sending your customers' contact details to an advertising platform; your privacy notice should say so.

Do not mistake the returned figure for an independent measurement. Once TikTok has your sales, its report will show purchases, and those purchases are your own data passed through its attribution rules. The report is better for optimization. It is not a second opinion.

What the matched number misses, and how to see the rest

Matched revenue is a floor. It captures everybody who came through a TikTok-only door and bought, and it misses everybody who saw a video and arrived some other way. The size of that second group can only be estimated by comparison: across time, across regions, or with a deliberate holdout.

Short-form video is built to be remembered rather than clicked. A homeowner who watched three of your roofing videos over a month and then searched your name when the storm came is a TikTok customer in every sense except the one an attribution report can see. That sale will be credited to branded search, which is accurate about the click and silent about the cause.

• Watch branded search and direct calls. If TikTok is working at the top, the volume of people searching your name and calling your main line should move when it runs and when it stops.
• Change one thing at a time. Launch TikTok in a month when nothing else changes, or stagger it against your other channels, so a comparison can say something.
• Use geography if you have it. A business serving several areas can run TikTok in some and not others. That is the strongest evidence available without a platform's help, and it uses your closed sales rather than anybody's conversions.

Report these as evidence, not attribution. A table that adds a matched $21,600 to an estimated lift and calls the total "TikTok revenue" has mixed a fact with an argument, and the first person to question the argument will stop believing the fact.

CloseRev matches a closed-sales export to a lead export on email and phone number, so TikTok's form leads and the calls to its number become rows in your lead file with "TikTok" in the source column. Each matched sale is credited to the earliest matching lead, and sales it cannot match stay in Direct / Unknown — it does not connect to TikTok's API, and it never moves view-through claims into revenue.

Is TikTok worth it for a business that sells offline?

It can be, and you can find out for your own business in two or three months. Give TikTok its own door, match what comes through it to closed sales, and compare the matched cost per customer with every other channel. If the floor alone pays for the spend, the case is made.

The mistake in both directions is deciding without that number. Owners who judge TikTok by Ads Manager tend to keep it too long, because the cost per conversion looks cheap. Owners who judge it by their CRM's source field tend to cut it too soon, because the customers it influenced arrived through search. Neither is looking at the sales.

Matched revenue is not the whole truth about TikTok, and it does not pretend to be. It is the part you can prove, which makes it the part worth starting from.

Judge TikTok on customers, not conversions. A channel whose matched floor covers its spend is earning its place before anybody argues about the rest; one that cannot cover its spend on matched sales needs a better argument than a dashboard.