How we measure
Many posts on this site say "in Enalitica's data" and then give a number. This page is the one place that says what that phrase means: whose orders, which window, what is left out, how a figure is checked before it is printed, and which figures we decided not to publish. Every post that quotes our data links here.
What the data is
Enalitica syncs real orders from WooCommerce, Shopify, Shopamine and custom shops, daily ad cost and the conversions each platform reports from Google Ads, Meta, Microsoft, LinkedIn and ChatGPT ads, GA4 sessions, and the leads of service businesses. A blog figure is an aggregate across the accounts that have the relevant connection active. "Stores with both Meta and Google Ads connected" means exactly that: the comparison is only fair where both platforms report on the same orders.
Orders are counted from the store's own records, not from analytics events. Cancelled, refunded and test orders are excluded. Revenue is net of VAT where the store records it that way. Ad cost is the daily spend the platform's API returns, in the account currency, and platform-reported conversions are taken as the platform states them, including their view-through and modeled share, because the whole point of the comparison is to show that number next to real orders.
The window and the sample
Most published figures come from two measurement runs over the last 90 days: one closing on 25 September 2026 and one on 28 September 2026. Each post names the run it uses. The 28 September run covered nearly 12,000 orders, more than 6,500 order journeys, nearly 8,000 leads and more than 1 million GA4 sessions; the customer figures use the 12 months to the same date and nearly 30,000 customers. We refresh the runs when a post is refreshed, and the date in the sentence tells you which run you are reading.
How a figure is checked before it is printed
- Per-store spread. A pooled number that one large store carries is not published as a general finding. Where the spread matters we print it ("median store 68.5%, range 47% to 95%").
- Coverage. When a field is known for only part of the orders (the device, the keyword), the sentence says so and the share is computed on the orders where it is known.
- Artifacts. Fields that our own system refreshes on a schedule are not used to measure timing, because they would measure the schedule.
- Our own rollout. When a capability was switched on during the window, we report the direction and say so, not a curve.
- Platform behaviour. We never claim a platform changed its rules from our trend alone. What the platform documents and what we measure are shown side by side, each with its source.
- Anonymity. No client names, no order ids, no customer data. Tilen's own store, Opremisidom, may be named; everyone else is "a furniture store" or "a pharmacy", and only with permission.
Figures currently in use
The numbers quoted across the blog, with the run they come from. A post may round them or quote a subset.
| Figure | What it compares | Run |
|---|---|---|
| Platforms claim as many conversions as the stores had orders | Stores with both Meta and Google Ads connected: the two platforms together reported about as many conversions as the stores had orders in total, worth about three quarters of all revenue, while the stores also sold through email, organic and direct. | 90 days to 25 and 28 September 2026 |
| Meta reports 4.6 conversions per order carrying a Meta click | Orders with a Meta click id against conversions Meta reported in the same stores and window. | 90 days to 28 September 2026 |
| GA4 sees 76.6% of real orders | GA4 purchase events against store orders in stores with both connected; median store 68.5%, best 95%, weakest 47%. | 90 days to 28 September 2026 |
| 26.4% of orders carry a paid click id | 14.7% gclid, 12.4% fbclid, 2.4% srsltid, across nearly 12,000 orders. AI assistants were the referrer on about 0.5% of orders. | 90 days to 28 September 2026 |
| Last touch by channel | Organic 24.8%, direct 24.4%, Google Ads 20.4%, Meta 11.8%, referral 6.4%, email 3.3%, across more than 6,500 order journeys. 25.5% of journeys touched two or more channels; 23.2% had a different first and last touch. | 90 days to 28 September 2026 |
| Brand keywords are 40% of Google Ads orders with a known keyword | Keyword known for 54.3% of Google Ads orders; median store about 3 in 10. Performance Max was 30.5% of Google Ads orders. | 90 days to 28 September 2026 |
| Repeat customers | Across nearly 30,000 customers with an order in the last 12 months, 14.7% ordered again within those 12 months (per store 8% to 21%) and 23.4% of all orders were repeat orders. | 12 months to 28 September 2026 |
| Leads (service businesses) | Organic 34%, Google Ads 27%, direct 23%, Meta 12%, AI assistants 1.6%, across nearly 8,000 leads; 44% of lead journeys had an ad click. | 90 days to 28 September 2026 |
What we do not publish, and why
- Average order value by channel. It reflects which stores sell what, not what a channel does.
- Cost-price coverage and refund rates. Too few stores load cost prices or sync refunds for the number to mean anything beyond those stores.
- Consent decline rates. They depend on the banner, not on the market.
- Lead response time. It is recorded by the CRM, not by us, for most clients.
- The number of stores. It changes monthly and would not make a figure more or less true; the per-store spread is the honest substitute.
Check it on your own store
None of this requires Enalitica. Take last month, write down the purchases Meta reports and the purchase conversions Google reports, add them and compare with the orders in your store. Then count GA4's purchase events against the same orders. If the platforms' sum lands near your total orders, and GA4 lands well below it, you have reproduced the two figures we quote most, on your own data. The attribution overlap calculator does the first comparison for you and says how many orders were counted twice at the least. The conversion tracking audit walks through the five checks in order.
Frequently Asked Questions
Whose data is "Enalitica's data"?
The orders, ad costs, sessions and leads of the stores and service businesses that run on Enalitica and have the relevant connection active. Every figure is an aggregate across those accounts; no client name, order or customer ever appears in a post, and a single client is named only with their written permission.
Why do you not say how many stores are behind a number?
Because the count changes month to month as clients join and because a figure that one large store carries would be misleading whatever the count. Instead we check the per-store spread before publishing and state it when it matters ("median store 68.5%, range 47% to 95%").
How can I check these numbers on my own store?
Take last month. Write down purchases from Meta Ads Manager and purchase conversions from Google Ads, add them up and compare with the orders in your store. Then count GA4 purchase events for the same month against the same orders. Those two comparisons are the two figures we quote most, and you can run them in ten minutes without any tool.