Attribution Windows: Choosing 1-Day, 7-Day, or Lifetime

Switch your Meta Ads attribution window from 7-day click to 1-day click and watch reported conversions drop by 30-50% overnight, with zero actual change in sales. Nothing about your business changed — you just stopped counting conversions that happened outside a shorter time box. This is the part of attribution windows that trips up even experienced marketers: the window isn’t a measurement of truth, it’s a policy decision, and different policies make wildly different channels look good or bad.

What an attribution window actually controls

An attribution window is the maximum time allowed between a touchpoint (an ad click, a view, an email open) and a conversion for that touchpoint to receive credit. A 7-day click window means: if someone clicks your ad and converts within 7 days, the ad gets credit; if they convert on day 8, it doesn’t, even if that ad was the actual reason they came back and bought.

This single setting changes reported ROAS more than almost any other lever in your measurement stack, and it changes it differently by channel. Paid search captures a lot of already-decided-to-buy intent, so shortening its window barely dents its numbers — someone searching your brand name today usually converts today. Paid social and display, which mostly plant awareness rather than capture existing intent, lose a disproportionate share of their credited conversions when you shorten the window, because their value shows up as a nudge three days later, not an instant click-to-purchase.

This means comparing ROAS across channels with the same window setting isn’t actually apples to apples unless the channels have similar buying-cycle shapes — and they almost never do.

Matching window length to your actual sales cycle

The right window isn’t a platform default, it’s a reflection of how long your buyers actually take to decide. Pull your CRM or analytics data and look at the actual distribution of time-to-conversion for closed deals or completed purchases over the last 6-12 months. Don’t estimate this — measure it. Most teams are surprised by what they find.

For a low-consideration e-commerce purchase (say, a $40 consumer product), the bulk of conversions genuinely do happen within 1-3 days of the triggering touchpoint, and a 1-day or 7-day click window reasonably captures the real buying behavior. For a B2B SaaS product with a sales cycle averaging 45 days, a 7-day window is nearly useless — it will systematically undercount every channel that contributes early in a long cycle (top-of-funnel content, first-touch paid social, a webinar attendance) and overweight whatever channel happens to be active in the final week before close, usually branded search or a sales-driven email sequence.

A rough rule that holds up across a lot of businesses: set your primary attribution window at roughly the 70th-80th percentile of your actual observed conversion-time distribution. If 75% of your conversions happen within 21 days of first touch, a 21-30 day window captures the real behavior without diluting credit across an unreasonably long tail. Going shorter systematically punishes upper-funnel channels; going much longer starts crediting touchpoints that had no real causal relationship to the eventual conversion.

Click vs. view attribution — a different axis entirely

Window length is one dimension; click-through vs. view-through is another, and conflating the two causes a lot of confusion. Click-through attribution only credits a touchpoint if the user actually clicked. View-through credits an impression even if the user never clicked but saw the ad and later converted through some other path (direct visit, organic search, etc.).

View-through attribution is where the most inflated numbers hide. A display or video campaign can show enormous “view-through conversions” simply because it was shown to broad, high-intent-adjacent audiences who were going to convert anyway — the ad gets credit for a conversion it had no real influence on. Meta and Google’s default reporting often blends view-through into headline ROAS numbers, which is part of why paid social so often looks more efficient in-platform than it does once you check it against your actual CRM revenue.

A defensible approach: use click-through attribution as your primary decision-making metric for budget allocation, and treat view-through as a directional, secondary signal you sanity-check but don’t optimize spend against directly. If a channel’s performance story only holds up when you include heavy view-through credit, that’s a signal to investigate rather than a green light to increase budget.

The multi-touch problem windows don’t solve

Even with a well-calibrated window, single-touch attribution (whether first-click, last-click, or last-non-direct-click) still assigns 100% of credit to one touchpoint in what’s usually a multi-touch journey. A buyer might see a LinkedIn ad, later click a retargeting ad, then convert via a branded search two weeks later. Last-click attribution gives 100% credit to branded search — the channel that was, in reality, just closing out a journey that display and retargeting built.

This is why window length and attribution model need to be considered together, not separately. A short window paired with last-click attribution is the most aggressive possible way to strip credit from upper-funnel channels — it’s a combination that will always make paid search and email look disproportionately effective and paid social/display look disproportionately weak, regardless of their actual contribution. If your reporting stack only supports single-touch models, lean toward a longer window and a linear or position-based model rather than last-click, to avoid systematically starving the channels that build the pipeline your bottom-funnel channels later close.

Auditing what your ad platforms are actually reporting

Every platform (Meta, Google, TikTok, LinkedIn) sets its own default window and lets you adjust it independently, and most accounts are still running whatever default was set when the account was created — often a 7-day click, 1-day view window nobody consciously chose. Audit this quarterly:

• Pull the attribution window setting from every ad platform you run
• Compare each one against your measured actual conversion-time distribution
• Standardize all platforms to the same window where the sales cycle is comparable, so cross-channel comparisons in your reporting are actually valid
• Document any platform where you intentionally use a different window (e.g., a shorter window for a flash-sale campaign vs. your always-on brand campaigns) so nobody mistakes it for an oversight later

This audit regularly turns up accounts running a 1-day click window for a business with a 30-day sales cycle, which means the platform’s in-dash ROAS number has been dramatically understating that channel’s real contribution for months, sometimes years, without anyone questioning why a channel “never seems to convert” in-platform despite anecdotal evidence it’s working.

Reconciling platform numbers against your source of truth

Ad platforms grade their own homework — each one attributes conversions using its own window and model, on its own data, with an obvious incentive to look effective. The only reliable check is reconciling platform-reported conversions against your actual CRM or backend revenue data, using a consistent methodology across channels rather than trusting each platform’s self-reported number independently.

Set up a monthly reconciliation: total revenue/conversions your CRM shows as coming from a channel (via UTM-tagged, first-touch or multi-touch modeled data) versus what that channel’s own ad platform claims. Discrepancies of 10-20% are normal and expected due to methodology differences. Discrepancies of 50%+ mean either your window/model settings are badly mismatched to reality, or there’s a tracking gap (a broken pixel, missing UTMs, cross-device journeys your stack can’t stitch together) inflating or deflating one side of the comparison.

A worked example: recalibrating a window after moving upmarket

A company selling project management software started as a self-serve, SMB-focused product with a median time-to-purchase of 4 days, and its Meta and Google accounts were both set to a 7-day click window from day one — a reasonable match at the time. Over 18 months, the company shifted its ICP toward mid-market teams of 50-200 seats, added a sales-assisted motion for larger deals, and its median time-to-purchase stretched to 34 days, with a meaningful tail of deals closing past 60 days.

Nobody revisited the attribution window during that shift. The result: paid social, which had been introducing mid-market prospects early in a now much longer cycle, was showing a reported ROAS around 0.6 — apparently unprofitable — while branded search, which mostly caught prospects in the final week before close, showed a ROAS above 4. Based on those numbers, the team nearly cut paid social entirely. Before doing so, someone finally pulled the actual conversion-time distribution from the CRM and found the 75th percentile now sat at 38 days, not 7. Widening the window to 30 days and switching from last-click to a position-based model changed paid social’s attributed ROAS to roughly 1.8 — still not the strongest channel, but clearly not the channel to cut first. The near-miss here is common: a metric that looks decisively bad is sometimes just measuring the wrong window for a sales cycle that quietly changed underneath it.

The common failure mode: adjusting the window mid-quarter to make a channel look better

A window recalibration should be driven by the underlying conversion-time data changing, not by a channel’s numbers looking disappointing in a given reporting period. A frequent and damaging pattern: a channel is underperforming against its budget, someone notices that widening the attribution window would make its numbers look better, and the window gets quietly extended right before a board update or budget review — with no underlying change in actual buyer behavior to justify it.

This erodes the credibility of the entire measurement system once discovered, and it usually is discovered, because a sudden, unexplained jump in a channel’s reported performance the same week its budget was under scrutiny is not a subtle pattern to an analytically-minded stakeholder. Keep window changes tied strictly to a documented recalibration process — reviewed on a fixed schedule (twice a year, as noted below) using actual conversion-time data pulled fresh each time — and treat any off-schedule change request as a signal to dig into why a channel’s numbers are under pressure, not a reason to adjust the measurement until the pressure goes away.

Sequencing a window audit when you’re starting from scratch

Teams inheriting a measurement stack with no documented window rationale — the common state at most companies — shouldn’t try to fix every platform simultaneously. Start with whichever channel currently receives the largest share of budget, since a misconfigured window there has the largest dollar impact if it’s silently misleading a reallocation decision. Pull that channel’s actual conversion-time distribution from the CRM first, set its window to match, and only then move to the next-largest channel. This also builds institutional confidence in the process — a clear, well-reasoned recalibration of the biggest channel, with the resulting numbers making more intuitive sense to stakeholders who felt something was off, makes it much easier to get buy-in for auditing the smaller channels afterward, rather than asking for a wholesale measurement overhaul upfront before anyone’s seen it pay off.

Making the window a deliberate, revisited decision

Treat your attribution window as a setting you review at least twice a year, not a one-time configuration. Sales cycles shift as your product, pricing, and ICP evolve — a company that moves upmarket from SMB to mid-market customers will see its conversion-time distribution stretch out, and a window calibrated for the old buyer will start systematically misattributing credit for the new one. The businesses that get attribution right aren’t the ones with the fanciest multi-touch model; they’re the ones who actually check their window and model choices against real conversion-time data on a recurring schedule, rather than trusting whatever default a platform shipped with three years ago.