How to Set an Attribution Lookback Window Properly

To set an attribution lookback window, find your median time from first touch to payment, then set the window comfortably longer than it. Never keep the tool's default. A window shorter than your sales cycle quietly erases the start of every customer journey.

How to set an attribution lookback window that fits your business

Your tool already has one. You didn't choose it. Someone else's average customer did.

That default is the most expensive setting you've never looked at. A lookback window is how far back your system searches for touches that count toward a sale. Anything older than the window is treated as if it never happened.

Think about what a short attribution lookback window means. A reader finds you through an article. Weeks later they're back through a search for your name. Then they buy. If your window is too short, the article is gone. The sale belongs to search. The article gets no credit, and you cut it.

That's the failure. It isn't dramatic. Nothing breaks. The dashboard stays green and keeps telling you a confident story that's missing its first chapter.

Why a short window punishes the work you're proudest of

A short window doesn't treat every channel equally. It punishes the slow ones.

Some marketing pays back in days. Some pays back in quarters. Content, a podcast, showing up in a community: all of them work long after the touch, usually because someone eventually types your name into a search box. A window that's too short only sees the last step of that walk.

So the channels that look strongest are the ones that harvest. The ones that quietly built the demand look weak. And you, being a sensible person with limited hours, cut the weak ones.

This is how a solo founder destroys their best long-term asset while feeling disciplined. It's worth reading that sentence twice. You didn't get lazy. You got a wrong number and acted on it.

If you've been wondering why first touch and last touch disagree so badly in your numbers, the window is the first thing to check. A tight window drags the two models apart in the cheapest way possible, by deleting the early evidence.

Free guide · 26 pagesStop paying for what does not work.The Attribution Playbook: how to find which marketing produces money.Get the free guide →

Step one: find your real sales cycle

You need one number. The median time between a customer's first touch and their first payment.

Median, not average. One customer who took forever, or one who bought in an afternoon, shouldn't move it.

Here's how to get it, in the order I'd do it:

1. List your last batch of paying customers. Not leads. Money.
2. For each, find the date of their first recorded visit. Your tracker holds it, if the form carried the history across.
3. Find the date they paid. Your payment system holds that.
4. Work out the gap for each one. Line them up from shortest to longest.
5. Take the middle one. That's your median.

This only works if the join exists. If you can't connect a visit to a payment, you can't compute this, and that's your finding. The guide is blunt about it: not knowing your cycle length is itself the first result, and a valuable one. It tells you the identity handoff is broken, which is a bigger problem than any window. The fix is covered in how to track where your leads come from.

Small business, few customers? Then the number is rough. Good. Rough and honest beats precise and borrowed.

Step two: set your attribution lookback window longer than the cycle

Now the rule. The window goes comfortably past your median. Not equal to it. Past it.

Why? Because a median means half your customers took longer. If the window equals the median, you erase the early touches for half of everyone who ever bought. Comfortably longer covers the slow half too.

Don't agonise over how much longer. Say your median is a few weeks. A window of a couple of months protects the slow half. The exact figure matters less than being on the right side of the line.

What does it cost to be generous? A little noise. Old touches that mattered less get counted. That's a fair price. The opposite error costs you a channel.

Here's how the three possibilities for your attribution lookback window behave:

One row of that table is safe. It's the last one.

What to do if your tool won't let you change it

Some tools lock the window. Don't panic, and don't buy a new tool because of it.

First, check whether the tool stores raw touches anyway. Many do, and only the report applies the window. If you can export or query the touches, you can compute your own view over a longer period.

Second, if you truly can't widen it, treat the report as a last-touch lens only. Say that out loud. "This view shows what closed people, not what found them." Then judge discovery from another source: first-touch data, or what people tell you when you ask.

Third, and this is the underrated one, ask. A single optional question on your main form, "how did you hear about us?", reaches back further than any window. People remember the podcast they heard three months ago. Your tracker doesn't. The method for pairing the two is in how to track untaggable marketing channels.

What you shouldn't do is carry on trusting a window you've just learned is too short. You know better now.

Free assessmentHow trapped are you, really?Take the Corporate Suffocation Index, the free solopreneur readiness assessment.Take the assessment →

Measure the lag, then use it as a trial length

Your window has a second use. Once you know the gap between first touch and money for each channel, you know how long a fair trial of that channel really is.

Content, say, might take far longer to pay than a referral. If you judge both after the same month, content loses every time and deserves to lose none of them.

So for outcomes attributed to a slow channel, ask one question: how long between first touch and money? Write down the answer per channel. That's the lag.

Then you can agree your trial lengths before you start, instead of arguing with yourself every month. The guide makes the point sharply: deciding the trial length in advance is the only reliable defence against cutting something in month four that pays back in month seven. It's also the groundwork for how to kill a marketing channel on a rule instead of a mood.

Revisit it when your business changes

A window isn't set once and forgotten. Your cycle drifts.

Change your price and the cycle usually stretches. Add a cheaper entry product and part of it shrinks. Move from selling to individuals to selling to small teams and it can change completely. Each of those shifts means your median has moved while the window stayed where you left it.

So put it on the calendar. Once a quarter, recompute the median from your most recent customers and check that the window still sits comfortably past it. Ten minutes. If it has moved, move the window.

Do this alongside your monthly numbers, not instead of them. The routine is laid out in how to review marketing results every month.

Free guide · 26 pagesStop paying for what does not work.The Attribution Playbook: how to find which marketing produces money.Get the free guide →

A Monday version you can finish before lunch

You don't need a project. You need an hour.

• Open your payment system. List your recent paying customers.
• Match each to their first visit. If you can't, write down why. That's the real task.
• Work out the median gap. Round it up.
• Look at your tool's current window. Is it comfortably longer? If not, change it today.
• Write the number on a note next to your naming convention. Future you will thank you.

That's it. One number and one setting for your attribution lookback window. It will change what your channels look like, and the channels that gain are usually the ones you built with the most patience.

Be ready for an uncomfortable result. Widen the window and a channel you'd written off can climb. Treat that the way you'd treat any surprising number. Check the tagging and the join first. Once it survives that, believe it.

Free assessmentHow trapped are you, really?Take the Corporate Suffocation Index, the free solopreneur readiness assessment.Take the assessment →

Frequently asked questions

What is an attribution lookback window?

It's how far back your system searches for touches that can receive credit for a sale. Anything older than the window is ignored. If it's shorter than your sales cycle, the early part of each customer journey disappears from your reports.

How long should my lookback window be?

Longer than your median time from first touch to payment, and comfortably so. Work out that median from your own customers rather than trusting the tool's default. Equal to the median still erases the start of the journey for the slower half of buyers.

What if I don't know my sales cycle?

Then finding it is your first job, and it's a useful result. It usually means visits and payments aren't connected, so you can't yet say how long anyone took to buy. Fix the handoff between your forms and your records first.

Can a window be too long?

Slightly. A very long window lets old, weak touches share credit and adds some noise. For a solo operator comparing channels, that's a much smaller problem than losing the first touch of a customer who took their time.

The full method, with the checklists, is in The Attribution Playbook, a free 26-page playbook.