Every Platform Takes Credit: How to Find the Sales Your Ads Actually Drove

Build good DIY testing habits

Test with intention. Write down the metric, the comparison and what result would change your plan, before you see the data.

Match rigor to stakes. A small, reversible budget shift needs less certainty than cutting a channel or reporting results to your board.

Measure where the sales are. Look at whole channels and full campaign windows, not tiny audience segments with a handful of purchases.

Treat “no clear result” as inconclusive. With small numbers, a test that shows no difference often just didn’t have enough data to detect one.

Look for agreement. When platform data, your ticketing system, survey answers, and a holdout test point the same way, act on it.

Where to start: your first 90 days

Month 1: Standardize UTMs, configure GA4 ecommerce (if it isn’t already) so your website data is complete and reliable, and add a survey question.

Month 2: Build one dashboard with MER and cost per new buyer.

Month 3: Run your first holdout test. An email holdout is the easiest place to begin.

After that, aim for one test per quarter. Each one makes the next budget conversation a little more grounded in evidence and a little less in guesswork.

What your data can tell you

Most organizations already have several sources of data. Each answers a different question, and each has blind spots.

Google Analytics

What it tells you: which channels and campaigns bring people to your website, and which of those visits end in a purchase.

What to keep in mind:

It only sees clicks. Someone who sees your Instagram ad, doesn’t click, and later searches your name gets credited to search.

It doesn’t follow people across devices. A patron who browses on their phone and buys later on their laptop looks like two different people.

Its memory is short. GA recognizes people by browser cookies, which get cleared, blocked or expire, so returning patrons often look brand new.

It misses offline sales. Phone and box office orders never reach it.

It relies on consistent UTM tracking. Learn more about UTMs.

It defaults to “last click” attribution. The primary reports focus on first and last click attribution. But even the data driven attribution reports are skewed by the factors above. Learn more about GA4’s various advertising reports.

Together, these gaps tilt credit toward organic and paid search, which usually come right before a purchase. To make GA as reliable as possible, standardize your UTMs and set up cross-domain tracking to your ticketing checkout so the journey doesn’t break at the moment of purchase.

Audience surveys

What it tells you: a “How did you hear about us?” question at checkout or in your post-show survey catches what no tracking can see: a friend’s recommendation, a radio spot, a group text.

What to keep in mind:

Memory is imperfect. People tend to name the most memorable touchpoint, not everything that influenced them.

Not everyone answers. The patrons who respond may not represent all of your buyers.

Consistency matters. Keep the answer list short and unchanged so you can compare season to season.

Ad platform results

What it tells you: Meta, Google and other platforms report the purchases they attribute to your ads. This is the most detailed view of what’s working within each platform: which creative, audience or placement performs best.

What to keep in mind:

Each platform grades its own homework. Many times, people see ads across a variety of channels, and they will all take credit for the resulting purchase. Aggregated ad platform results will then usually add up to more than your actual sales.

View-through credit is generous. Platforms can count people who saw an ad without clicking, including people who would have bought anyway. Retargeting is especially prone to this.

Some results are estimated. Privacy changes mean platforms model a share of conversions rather than observe them directly.

Use platform results to compare campaigns, targeting tactics, and creative performance within a platform, not to compare platforms or prove total impact. Sending purchase data from your ticketing system back to the platforms, through built-in integrations or offline uploads, makes reporting more accurate.

None of these sources can tell you what would have happened without your ads. That’s what testing is for.

Get closer to cause and effect

Marketing mix modeling estimates each channel’s contribution from your historical spend and sales. Google’s Meridian and Meta’s Robyn are free and open source, but they require coding in Python or R, plus roughly two years of weekly data. With small audiences and changing programming, these models’ insights often come with wide ranges of uncertainty. Instead, these are the habits that matter more.

Track blended KPIs

Why it matters: This is one of the simplest ways to see whether your marketing is paying off overall. Your ticketing system holds every order, online, by phone, and at the box office. Compare those totals against total marketing spend over time to get numbers that don’t depend on which platform gets credit:

• Marketing efficiency ratio (MER): total ticket revenue ÷ total marketing spend.
• Cost per new buyer: total spend ÷ first-time ticket buyers.

What to keep in mind:

It shows a relationship, not a cause. Spend usually rises when demand does, around on-sales and openings, so sales would climb anyway.

Productions differ. A well-known title sells differently than a world premiere, so compare like with like.

Outside events matter. Keep a change log of announcements, price changes, press and weather so you can explain jumps.

Run Tests

Why it matters: Testing is where you get closer to cause and effect. A common incrementality method is the geo test: run ads in some cities and not others, then compare. But if your audience lives within an hour of your venue, you don’t have spare markets to hold out. The good news is that you can split people instead of places.

Platform lift studies. Meta and Google both run randomized lift studies that show ads to one group of people, hold back a control group, and report the difference in conversions. They’re the most rigorous option, but they come with high minimum spends and minimum conversion volumes to generate a significant result, so they can be out of reach for smaller organizations. 

That doesn’t mean you’re out of options, though.

Do-it-yourself holdouts. These use the same logic on a smaller scale:

Email holdouts. Randomly hold back 10–20% of a list from a promotion, then compare purchase rates for one group vs. another in your ticketing data.

Branded search pause. Turn off ads on your own organization’s name for two to four weeks and watch whether total sales or site traffic dips.

On/off pulses. Alternate weeks with and without a given channel across a long-running campaign, like a subscription push, and compare totals.