Stores credit Meta with a quarter of their orders. Meta reports twice as many.
Two years of Meta ad accounts and store orders from online stores in Bulgaria: what they spend, what a new customer costs, what Meta reports and what the store records.
- Online stores in Bulgaria that connected Meta to Releva, October 2024 to September 2026
- Meta's reports for ads shown in Bulgaria, next to each store's own orders
- A statistically significant sample, indexes and ratios, no store named, no prices
The short version
Online stores in Bulgaria, October 2024 to September 2026. Store figures: October 2024 to June 2025 and May to September 2026, when orders carried their source reliably (new customers: May to September 2026). Journeys: orders from 15 January to 31 March 2026, traced through recorded ad visits, which were reliable in those months. Last touch, as each store records it. Not a controlled test.
Meta matters, and reports more than the store records. At the median store, the store credited Meta with 25% of orders and 24% of revenue. Meta's own reports counted 2.3 times as many purchases, more than the store in almost every account.
Two returns for the same euro. In the same accounts and months, Meta reported about 5.7 euros of revenue for every euro at the median account. Counted as orders in the store, it was 3.8. The two count differently: Meta by each account's attribution settings, the store by the last click.
A new customer cost 44% of their first order. At the median store, Meta was credited with about a third of first orders from May to September 2026. Meta spend per new customer credited to Meta came to 44% of their first order's value. Whether that pays depends on the store's margin and on how many order again.
Costs held, Black Friday didn't. Year on year, the cost of impressions and clicks showed no clear change at the median store. In November 2025 these advertisers almost doubled spend against their average month, and 1,000 impressions cost 38% more.
Ads and messages share the sale. About half the orders email and Viber closed followed a visit from a Meta or Google ad in the 30 days before.
In November 2025, spend almost doubled and impressions cost 38% more.
Meta spend and the cost of 1,000 impressions in Bulgaria, month by month, October 2024 to September 2026. Both are indexed, so the average month = 100.
Online stores in Bulgaria that connected their Meta ad account to Releva, a statistically significant sample. Meta's own reports, restricted to ads shown in Bulgaria. Only stores that advertised in every one of the 24 months are included, so stores joining later don't move the pattern. The largest advertisers pushed hardest: the typical store spent about a quarter more in November 2025 than in its average month, and paid about a fifth more per 1,000 impressions. In November 2024 the rise in spend was much smaller.
Read the full report
Enter your work email to open the rest of The Meta Ads Report 2026 on this page. Inside: what Meta reports against what the store records, and what a new customer costs. Then costs year on year, where the money goes, and how ads and messages share the sale.
Thanks. The full report is open below.
We use your email to know who reads our research. Privacy policy
Meta counts about two purchases for every one the store gives it.
We set each store's Meta reports, for ads shown in Bulgaria, against the orders the store itself credits to Meta in the same months. The store credits an order to Meta when the shopper's last visit came from a tagged Facebook or Instagram ad.
At the median account, Meta reported 2.3 times as many purchases. Meta reported more than the store in almost every account, and the difference is statistically significant. In accounts where the store credited Meta with a larger share of orders, the gap was smaller, about 1.8 times.
The two count differently. Meta's reports follow each ad account's attribution settings. The store gives each order to one source, the last click, and ads that arrive untagged look like direct visits. For budgets, the store's view is the one that ties to revenue.
| Median account | As Meta reports | Counted in the store |
|---|---|---|
| Purchases credited to Meta | 2.3x | 1x |
| Revenue for every euro spent | 5.7 | 3.8 |
| Share of the store's orders | not reported | 25% |
| Share of the store's revenue | not reported | 24% |
Online stores in Bulgaria with their Meta account connected to Releva. Only months with Meta spend and store orders when stores' source tracking worked: October 2024 to June 2025 and May to September 2026. Return figures use the same accounts and only months whose purchase values match the store's own orders. Revenue, not profit. The store credited Google Ads with 12% of orders at the median store, counted the same way.
A new customer credited to Meta cost 44% of the first order.
At the median store, Meta was credited with about a third of first orders from May to September 2026, and 22% across all stores together. Stores differ a lot: in many, Meta was credited with a third or more of first orders, in others under a tenth, which can happen when ad links are not tagged.
We divided each store's Meta spend in Bulgaria by the new customers whose first order was credited to a Meta ad, and set it against the value of their first order. At the median store it came to 44%. In about one store in six it was more than the whole first order. Whether that pays depends on the store's margin and on how many of those customers order again.
| New customers, May to September 2026 | |
|---|---|
| First orders credited to Meta (median store) | about 1 in 3 |
| First orders credited to Meta, all stores together | 22% |
| Meta spend per new customer credited to Meta, against their first order (median store) | 44% |
| All Meta spend per new customer from any channel, against their first order (median store) | 11% |
A new customer is a shopper whose first order in at least 15 months came between May and September 2026, after stores' source tracking recovered. Months with no recorded order sources are left out. Meta spend is for ads shown in Bulgaria in the same months, prospecting and retargeting together.
More spend, and no clear change in price.
We compared the same stores over two years: October 2024 to September 2025 against October 2025 to September 2026, using Meta's own reports.
At the median store, spend rose 27% and clicks per impression 19%. The cost of 1,000 impressions, of a click and of a purchase moved by less than 10%. None of these changes at the median store is statistically significant. All stores together, which weights the largest accounts most, 1,000 impressions cost 14% more and a click 12% less.
| Same stores, year on year | Median store | All stores together |
|---|---|---|
| Spend | +27% | +33% |
| Cost of 1,000 impressions | +4% | +14% |
| Cost of a click | −6.5% | −12% |
| Clicks per impression | +19% | +29% |
| Cost per purchase, as Meta reports | +5% | −4% |
Online stores in Bulgaria that advertised in both years, ads shown in Bulgaria, Meta's own reports. All stores together weights the largest accounts most, so we lead with the median store. No change at the median store is statistically significant.
Sales campaigns take almost all of it.
Campaign details were available for about three quarters of spend, October 2024 to September 2026. There, 86% went to campaigns with Meta's Sales objective, 89% counting older catalog and conversion objectives. 17% went to campaigns named Advantage+ and 12% to catalog or dynamic campaigns. 8% went to campaigns named as retargeting or remarketing, and 5% to campaigns named as prospecting. Most campaign names don't say.
Retargeting is where Meta and messages meet: both reach shoppers who already know the store. Below, we look at how the two share the sale.
Half the sales that email and Viber close followed an ad visit.
We followed orders placed from 15 January to 31 March 2026 back 30 days: which ad visits and which message clicks came before each order. We use these months because visits from ads were recorded reliably then.
Of the orders email and Viber closed after a click, 53% came after a visit from a Meta or Google ad in the 30 days before. That was 23% from a Meta ad and 40% from Google Ads. For orders none of these channels was credited with, it was 31%. Messages were ahead of that baseline in most stores, a statistically significant difference.
Many orders that messages closed followed a visit from an ad. Judging either channel alone misses part of the path to the sale.
| Orders closed by | Visited from a Meta ad in the 30 days before | From Google Ads | Either |
|---|---|---|---|
| Email or Viber, after a click | 23% | 40% | 53% |
| Recommendations and search on the site | 21% | 27% | 46% |
| None of these channels credited | 13% | 20% | 31% |
Orders 15 January to 31 March 2026 in online stores in Bulgaria, last touch. When an order carries more than one source, messages are counted before ads. A visit counts as from a Meta ad when its link was tagged as Facebook, Instagram or Meta. The method found the ad visit for about 9 in 10 orders the stores credit to Meta. Untagged ads look like direct visits, so these shares are a floor. In these months some orders lost their recorded source, so the last row includes some orders that came from ads, which makes the gap if anything smaller.
Different jobs, and no consistent winner.
In the stores that used both, in the same months, we set Viber against Meta, both counted as orders in the store. Neither returned more per euro consistently.
The kind of Viber message matters. At the median store, Viber automations were credited with more revenue per euro than Meta ads, and one-off blasts with less (different stores and months). Meta ads can reach shoppers who don't know the store yet. Viber reaches people who already gave their number. Many stores use both, and a holdout group shows what each adds. Read The Viber Report 2026
What Meta reports depends on what the store sends it.
A bedding and sleepwear brand, European and not part of this sample, switched Releva off for a test in summer 2026. Six weeks later, the return its Meta account reported was 34% lower, while the cost of impressions and clicks per impression held. So part of the drop may be in what Meta could count, not only in sales. It is one store and a before-and-after test, not a controlled one. Releva sends Meta each order from the server, with its value. Read the case study
Five changes for the next budget
Judge Meta on store orders
Use Meta's reports to run the ads and the store's orders to set the budget. Meta counted about two purchases for every one the store gave it.
Budget for the second order
A new customer credited to Meta cost 44% of their first order at the median store. Plan the flows that bring them back before you scale.
Plan Black Friday early
In November 2025, 1,000 impressions cost these advertisers 38% more than in an average month. Set budgets and audiences in October.
Let messages close
Half the orders email and Viber closed followed an ad visit. Connect ad audiences to your flows.
Measure with a holdout
Both views count what happened. A holdout group shows what the ads and messages added.
How we counted
Stores. Online stores in Bulgaria that connected their Meta ad account to Releva and gave read access, a statistically significant sample. Accounts without read access are left out.
Meta data. Meta's own monthly reports, read-only, restricted to ads shown in Bulgaria, October 2024 to September 2026, with each account's own attribution settings. Return figures use only months whose average purchase value matches the store's own orders.
Store data. Orders recorded on each store's site. An order counts for Meta or Google Ads when the shopper's last visit came from a tagged ad (last touch). From July 2025 to April 2026, far fewer orders had a recorded source, for Meta and Google alike. So store figures use only October 2024 to June 2025 and May to September 2026. Accounts where Meta counted more purchases than the store recorded orders in all are left out of the comparison.
New customers. A shopper's first order in at least 15 months, placed from May to September 2026. Months with no recorded order sources are left out.
Journeys. Orders from 15 January to 31 March 2026, with the ad visits and message clicks in the 30 days before each order. An order counts as closed by email or Viber only after a click on it. Journeys use recorded visits from ads, not the order's own source. Those visits were recorded reliably in these months, while order sources were not.
Money. Leva converted at the fixed rate of 1.95583, other currencies at fixed approximate rates. We show indexes, ratios and shares, never prices or store revenue, and no store is named.
Who made this. Releva sends the email, Viber and push messages in this study and runs some Meta audiences for these stores. Almost all the Meta spend we looked at is the stores' own campaigns.
Not a controlled test. These figures show what happened, not what each channel added on its own.
See what Meta really brings your store
Thirty minutes with your own data: what Meta reports, what your store records, and which flows turn its first orders into second ones.
Book 30 minutes
You'll talk to one of us.