How attribution works
Hardly anyone buys on a first visit. Someone sees a Facebook ad, searches your name a week later, reads a guide, and only then clicks a Google ad and sends the enquiry form. Attribution is the set of rules that decides which of those touches gets credit for that enquiry, and how much of it each one keeps.
Two things have to happen before any rule can run. Each visit has to be identified as coming from somewhere — usually through a click identifier the ad platform adds to the link, or UTM parameters you add yourself. Then the conversion has to be recorded and matched back to those earlier visits, within a fixed period called the lookback window. Touches outside that period, or made on a device the browser will not let the site recognise again, never enter the calculation at all.
Only then does the model do its work. One rule hands everything to the final click, another to the earliest, and the shared models spread the credit across the journey. The model is a choice, not a fact: the same sale can be credited to three different channels depending on which rule the report is running.
Why attribution matters
Budget decisions are made from attributed numbers. If your reporting quietly gives every sale to the last click, the campaigns that create demand — video, social, awareness content — look like they do nothing, and the cheapest thing to cut is the thing that was feeding everything else. Cut it, and the efficient branded search you kept starts shrinking too.
It matters more in a market like Nepal than most dashboards admit. A large share of enquiries arrive by phone call, Viber or WhatsApp, none of which a browser-based tool can see on its own. If those are missing, the channels that drive conversations will always look weaker than the ones that drive form fills, however good they actually are.
Common mistakes with attribution
The commonest is adding platform reports together. Google Ads, Meta and your analytics each count conversions their own way, over their own windows, and each claims the ones it helped with. Summing them produces more sales than the business actually made.
The second is treating a model change as a performance change. Switching the reporting model moves credit between channels; it does not move a single real order. Comparing this month with last month after a change is comparing two different rulers.
How to act on it
Name one system as the source of truth, usually your analytics or your CRM, and hold every channel to it. Write down which model and lookback window that system uses, and note the date whenever either changes. Keep the ad platforms for decisions made inside each platform, where their own view of their own clicks is genuinely the most useful thing available.
Then sanity-check against reality. Count the enquiries that actually reached your inbox, phone and messaging apps over a month and compare the total with what the reports claim. The gap tells you how much of your marketing is invisible, and closing it — through a sound analytics and tracking setup and by asking every caller how they found you — is worth more than any argument about models.
