How a data gap forms
Somebody sees your ad on a phone, thinks about it, and buys three days later on a laptop after refusing cookies and blocking scripts. That sale happened. Your ad platform has no way to connect it to the ad, so its report is short by one. Multiply that across a month and you have a data gap: the difference between commercial reality and the part of it your tracking managed to capture.
The causes stack up. Browsers limit how long tracking cookies survive. Phone operating systems ask users whether an app may follow them across other companies’ apps and sites, and many say no. Ad blockers stop the script loading at all. People switch devices mid-decision. Some buy by phone call, over WhatsApp or in a shop, where no pixel exists to fire. None of these are faults in your setup; they are how the modern web works.
Platforms fill part of the hole with modelled conversions — estimates built from the users they can still observe. That helps the totals look sensible, but a modelled figure is not a receipt, and different platforms model differently, which is why two dashboards rarely agree.
Why data gaps matter
Because budget follows measurement. A channel that is easy to track looks efficient; a channel that is hard to track looks wasteful, whatever they actually contribute. Brand search, referrals and word of mouth are systematically flattered because the click nearest the sale is the one that gets recorded, while the ad that started the search often disappears.
In Nepal the gap widens further, because a large share of enquiries finish over the phone or on WhatsApp rather than through a form. If only the form is measured, half the pipeline is invisible and the campaigns generating calls look like failures.
Where data gaps go wrong
The first mistake is treating the gap as a bug to be eliminated. It can be narrowed, never closed, and chasing perfect capture burns weeks for very little. The second is patching it by loosening every privacy control, which trades a reporting problem for a legal one.
The third and most expensive is the opposite reaction: distrusting all the data and going back to instinct. Incomplete numbers still contain a great deal of signal as long as you know which direction they lean, and a consistent under-count is far easier to work with than no measurement at all.
How to act on it
Start by measuring the gap rather than guessing at it. Compare platform-reported conversions with your own order or enquiry records for the same period, month after month, and watch the ratio. A stable ratio means the platform is a usable steering wheel even if the absolute number is wrong.
Then narrow it where the effort pays: a server-side connection such as the Conversions API recovers events that a browser blocks, and capturing phone and chat enquiries as recorded actions brings offline demand into view. For the part that remains, use experiments — pause a channel in a controlled way and watch total sales — because a geo holdout test measures impact without needing to see the individual user at all.
