Numbers differ from Facebook or Google

The numbers gas shows come straight from the interfaces of the individual networks. Even so, it happens that a value in gas does not match what you see in Facebook, in Business Manager, in Google Ads, in Sklik or in Google Analytics. Most such differences have an understandable cause, and you can recognise it from the way the difference looks. Some differences, however, mean an error that we have to fix.

This article walks you through both cases. We are not claiming that the numbers in gas are always correct. We want you to be able to tell an expected difference from one that somebody needs to check.

Before you start

A difference is only worth investigating once you are really comparing the same thing. Get four pieces of information ready:

• the value from gas and the value from the network, both copied down exactly,
• the period each of them covers,
• the name of the profile, page or ad account it belongs to,
• the name of the metric exactly as it is written in both tools.

Without these four pieces of information the cause is hard to find, and we will not be able to find it either when you write to us.

Procedure

Go through five checks in this order. Most differences are explained during them.

What happens next

After these five checks one of two situations remains.

The difference is expected when it matches one of the descriptions in the Common problems section: a different metric definition, a different time zone, fresh data, currency conversion or a discontinued metric. In that case you do not have to do anything, both numbers are correct in their own way, they just measure something different. If you need one binding value for a client, state which tool it comes from.

The difference is suspicious when any of these three statements applies:

• the difference is an order of magnitude, meaning gas shows ten, a hundred or a thousand times less or more than the network,
• gas shows zero and the network shows a non-zero value (or the other way round),
• the difference is consistently the same over time and does not disappear after several days or after changing the period.

Common problems

Reach and impressions look swapped

This is by far the most common cause. gas uses two different words for two different things:

• Reach is the number of people the content was shown to. Each person is counted once, even if they saw the content several times.
• Impressions is the number of times the content was displayed. If one person sees a post three times, three are counted.

In its Czech interface Meta translates the English term Reach as "Dosah", that is, as the number of people. So when you compare Dosah from Business Manager with Impressions in gas, you are comparing a number of people against a number of displays, and the number in gas comes out higher. The correct counterpart of Business Manager's Dosah is the Reach item in gas. You will find the same warning in the help text of the Reach card on the Overview.

Reach for a longer period is far higher in gas than in the network

Reach counts individual people, and therefore it cannot be added up across days: someone who saw the content on Monday and again on Wednesday is one person for the whole week, but appears twice in the two daily values. Networks can calculate the real reach without duplicates for a chosen period, whereas gas stores values by day and adds them up for the chosen period. The longer the period you select and the more loyal your audience is, the bigger the difference. For a single day the two numbers usually match, for a month gas can be noticeably higher.

Exception: in the detail of a single ad in the Meta Ads section, gas shows the real reach without duplicates for the whole downloaded period and points this out right next to the number.

Yesterday's number changed a few days later

This is normal and it happens with all paid networks. Platforms assign results to the day on which the user saw the ad, but they may record a conversion or interaction several days later. Google states that ordinary Google Ads statistics have a delay of up to three hours, whereas conversions counted by a model other than last click are added with a delay of up to fifteen hours, and processing the conversions themselves can take several tens of hours. For the same reason Meta keeps rewriting numbers inside its attribution window.

gas therefore downloads recent days repeatedly and overwrites them with new values. The practical consequence: treat yesterday's number as preliminary. If you are building a report for a client, expect the data for a finished month to settle only after several days.

Data for today is missing or low

The preset periods in gas end yesterday, so in most views you will not see today at all. Even if you select today using a custom range, the network is still calculating it and gas downloads it only a few times a day, so the value will be lower than in the network's live interface. For following a day in progress, the network's own interface is more suitable.

The daily breakdown does not match, but the period total does

This is almost always the time zone. Every platform decides for itself when its day ends, and gas takes its division over unchanged:

• Meta Ads reports numbers in the time zone set on the ad account. It is set when the account is created and never changes afterwards.
• A Facebook page has its daily statistics counted in Pacific time, so the day ends in our morning hours, not at our midnight.
• YouTube counts days from midnight to midnight Pacific time, so again not according to our calendar.
• Pinterest reports in UTC, which is two hours in summer and one hour in winter away from our time.
• Google Ads and Google Analytics follow the time zone set on the ad account, or on the measurement property respectively.

When the account time zone is not the Czech one, part of the evening traffic falls into a different day in gas than you would expect. A total for a longer period smooths the difference out, a daily breakdown does not. The time zone is always set on the platform side only, gas neither determines nor changes it.

The numbers for Facebook and Instagram differ by a slightly different amount every time

Meta calculates some metrics by estimation or modelling and itself labels them as estimated, or as still in development. For such metrics it explicitly states that they serve to follow a trend and are not suitable for precise historical comparison, because the calculation methodology may change. A small difference between gas and Business Manager for these metrics is therefore a property of the data, not a transfer error.

The post impressions chart ends in mid-June 2026

Meta discontinued this metric and offers no replacement for it, so after mid-June 2026 no new data arrives. The chart ends at this point and will not continue. Older periods remain complete and usable.

Page reach changed abruptly in mid-June 2026

At the same time Meta also discontinued the original calculation of page reach and replaced it with a different one. The new calculation is stricter (it counts only content that was actually rendered on screen), so the values tend to be lower. gas draws both series in a single chart because it is the same metric, but the break in mid-June is not a drop in your page's performance, it is a change of measurement on Meta's side. So do not compare the series before and after that date directly.

Instead of a chart, the report shows the note Metric was discontinued

When the report period extends beyond the day on which the platform discontinued a metric, and no data has arrived after that day, gas does not draw a chart ending in zeros but a note saying Metric was discontinued, stating the platform and the date. This is intentional: an empty chart would look like a gap in your communication. As soon as a replacement metric starts arriving, the note disappears by itself and the chart returns.

Spend is in koruna, but the ad account is kept in euros or dollars

For Meta and Google Ads ad accounts, gas converts spend into koruna using the Czech National Bank rate for the day on which the spend occurred. The ad account shows you the original currency. Both numbers are correct, they are just in a different currency, and you will not get the total for a longer period by multiplying by a single rate, because every day has its own.

The same spend comes out differently in two places in gas

Conversion into koruna is not yet consistent across the whole application. If you come across two different numbers for the same period on an account in a foreign currency, this is most likely why:

• The Meta Ads section, including the Campaigns and Ads tables, converts spend using the daily Czech National Bank rate.
• The Google Ads section converts using the daily rate on the Overview and on the Spend page. The Ad groups table uses a single current rate for the whole period and the Campaigns table does not convert spend at all.
• Pinterest Ads does not convert spend, the value stays in the ad account currency.
• Meta's accompanying metrics CPC, CPP and CPM stay in the account currency.
• Sklik bills in koruna, so no conversion arises there.
• In the Campaigns module the budget breakdown by network is added up without conversion.

All of these amounts have Kč written next to them, even where no conversion took place, so the symbol alone will not confirm the currency for you. We are working on unifying this. If you need certainty about which value to rely on, write to us through the contact form and we will tell you which number applies to your account.

Campaign spend does not match the total in the network section

The Campaigns module adds spend from several networks into a single amount based on the campaign tag. Besides the currency conversion described above, the difference is also affected by the fact that only ad campaigns whose name contains your campaign tag are included in the total, and only days falling within its period. The network section, on the other hand, shows the whole ad account. So the difference is expected every time something is also running on the account that does not belong to the campaign in gas.

One day in Google Analytics shows zero or too low a number

Google Analytics processes data for up to two days, and during that time the numbers change even in its own reports. gas stores a day once it has been downloaded, so that it does not have to fetch it again every time the page is opened, and it does not overwrite the stored value by itself. So if someone looked at a given day before Google finished calculating it, an incomplete number can stay in gas permanently.

This will not fix itself. If you see a day with zero or with an obviously low value in the Websites section while the value in Google Analytics is higher, write to us through the contact form and state which property and which days are involved. We can load the data again.

Conversions in the Google Ads section arrived only after several days

Google Ads assigns a conversion to the day of the click, not to the day the conversion happened. The number for an older day therefore keeps rising long after that day has ended. For this reason gas regularly reloads conversions for roughly three months back. If Google adds a conversion even later, it will no longer show up in gas. So if you see conversions in Google Ads for a period older than three months and not in gas, write to us through the contact form and we will add them.

An ad is missing in gas even though it is in Business Manager

Meta provides statistics at the level of individual ads for at most 37 months back. For older periods the numbers are available only at campaign level, which gas also states right next to the ads list. This is not a missing transfer, it is a limit set by Meta.

The profile was linked recently and older periods are empty

After a profile is linked, gas pulls the history retrospectively, but only as far back as the given network provides it, and the range differs from network to network. Empty history before the date of linking is therefore not a difference in numbers but a missing input. Filling in history after restoring interrupted access, and the limits of that, are described in the article Network access has expired.

The number in the PDF report differs from the one on the page in the application

A report is calculated at the moment it is generated and does not change from then on, it is a snapshot of the state. The page in the application, on the other hand, shows current data. If the numbers were recalculated after the report was generated (see the situation about retrospective recalculation above), the report and the application will diverge. This is intentional: a report sent to a client must not change behind your back.

The difference is an order of magnitude, for example tenfold

An order of magnitude difference no longer has any of the causes described above. It means an error in unit conversion or in currency conversion, and we have to fix it. Write to us through the contact form and state the network, the ad account or profile, the period and both numbers.

gas shows zero, the network shows a non-zero value

Zero against a non-zero value is never an expected difference. First check that you have the right profile and the right period selected in gas and that the account does not show the Paused label (the article Network access has expired describes what to do about it). If everything is in order and the zero remains, write to us through the contact form.

The difference stays the same and does not change from day to day

Differences caused by recalculation get smaller over time and disappear. A difference that stays the same for weeks is therefore systematic: either you are comparing a different metric, a different profile or a different period (go through the five checks above again), or it is an error on our side. If you pass all five checks and the difference persists, write to us through the contact form.