Attribution is where marketing meetings go to argue. The hoarding team says the billboard drove the sale. The performance team says their retargeting ad closed it. The content team says their blog started the whole journey. Everyone's a little right, which is exactly why attribution models exist, and exactly why people fight about them.
Let me explain the models in plain language, tell you which ones are actually worth using, and make an argument most attribution guides won't: that the whole discipline has been quietly broken for Indian brands because it left offline out entirely. That's finally changing.
Attribution is just answering one question: when a customer buys, which of your marketing touchpoints gets the credit?
If a customer saw your hoarding on Monday, clicked a Google ad on Wednesday, read your blog on Friday, and bought on Saturday, who made the sale happen? The hoarding introduced you. The ad brought them in. The blog convinced them. The final click sealed it. An attribution model is the rule you use to divide credit across those touches, and the rule you pick changes which channels look like winners and which look like dead weight.
Pick the wrong model and you'll defund the channel that's actually doing the heavy lifting. This is not academic. It decides where your money goes.
First-touch gives 100 percent of the credit to the first interaction. The hoarding introduced the customer, so the hoarding gets the sale.
Good for: understanding what drives awareness and fills the top of your funnel. If you care about discovery, first-touch tells you which channels start relationships.
The flaw: it completely ignores everything that happened after. A channel can introduce lots of people who'd never have bought without three more nudges, and first-touch will overcredit it wildly.
Last-touch gives all the credit to the final interaction before purchase. This is the default in most analytics tools, which is a big reason it's so overused.
Good for: understanding what closes deals. Simple to implement, easy to explain.
The flaw: it's arguably the most misleading model in wide use. It systematically overcredits the bottom of the funnel, usually branded search or retargeting, and gives zero credit to whatever created the demand in the first place. Last-touch will tell you to cut the very campaigns that made the customer want you, because the closing click stole all the glory.
Linear splits credit equally across every touchpoint. Four touches, twenty-five percent each.
Good for: a fairer, fuller picture than either single-touch model. It at least acknowledges the whole journey.
The flaw: not every touch matters equally, and pretending they do is its own distortion. A throwaway impression gets the same credit as the demo that closed the deal.
Time-decay gives more credit to touches closer to the purchase, less to earlier ones. A reasonable compromise: recent interactions weigh more, but early ones still count.
Good for: longer sales cycles where the recent stuff genuinely matters more but you don't want to erase the top of the funnel.
The flaw: it inherently undervalues awareness. Sometimes the touch that made the whole thing possible happened weeks earlier, and time-decay quietly discounts it.
Position-based typically gives 40 percent to the first touch, 40 percent to the last, and splits the remaining 20 among the middle. It rewards both the introduction and the close.
Good for: most B2C and considered-purchase journeys, where discovery and closing both matter more than the messy middle. This is where a lot of thoughtful teams land.
| Model | Credits | Best for | Watch out for |
|---|---|---|---|
| First-touch | First interaction | Awareness analysis | Ignores everything after |
| Last-touch | Final interaction | Closing analysis | Overcredits bottom funnel |
| Linear | All touches equally | Full-journey fairness | Treats trivial = critical |
| Time-decay | Recent touches more | Long sales cycles | Undervalues awareness |
| Position-based | First and last most | Considered purchases | Underweights the middle |
Here's my actual opinion, and it's the reason this guide exists.
Every model above assumes it can see all your touchpoints. But for years, Indian brands running hoardings, print, and in-store had a giant invisible touchpoint their attribution tools couldn't see at all. The billboard that introduced the customer, the newspaper ad that reinforced them, the standee that reminded them, none of it appeared in the model, because offline generated no data.
So what happened? First-touch and last-touch both defaulted to whatever digital touch they could actually track. Offline did the introducing and got zero credit, because it was invisible, not because it was ineffective. Entire marketing mixes got rebuilt around a model that was structurally blind to half of what was working. That's not an attribution nuance. That's a broken foundation.
You can run the most sophisticated position-based model in the world, and if it can't see the hoarding, it's confidently dividing credit among the wrong touchpoints.
The fix is to make offline touchpoints generate data, and QR scanning is what makes that possible. When someone scans a code on your hoarding or print ad, that scan becomes a real, timestamped, attributable touchpoint. Now offline sits in your attribution model alongside every digital touch, and first-touch can finally credit the billboard that actually started the journey.
To be precise, because precision is what attribution is about: a standard QR scans with any native camera, no app, production-ready today. Image-trigger ads, where the visual itself is scannable without a printed code, need our scanner or an app embedding it and are in beta. So the QR path into your attribution model works now.
Once offline is visible, the model choice finally matters for the right reasons. Run first-touch and you'll often discover the hoarding was starting far more journeys than anyone credited. That's the whole point: you can't pick a fair model until every channel is actually in it.
If you want the applied version of this, our offline advertising attribution guide walks through wiring scans into your reporting.
Attribution isn't about finding one perfect model. It's about making sure every channel that touched the customer is visible before you divide the credit. Get offline into the picture and every model gets more honest. Start free and give your billboard the credit it's been quietly earning.