How does a long purchase cycle (several days between click and purchase) affect tROAS calculations?

What to check before you touch anything

• What's the real median gap between click and purchase for this business (the Time Lag report in Google Ads)? Without this number, evaluating "recent" periods will be systematically skewed on volume and distorted on ROAS.
• Is the conversion window set wide enough to capture the real purchase cycle? If the cycle is longer than the window, some value is physically lost and never factored into ROAS.
• Are you evaluating recent days directly? Fresh periods will look worse than they really are, simply because some high-value conversions haven't happened yet or haven't been recorded yet.
• How often are decisions being made about the campaign (bid, target, budget changes) relative to the length of the purchase cycle? If decisions happen more often than the ROAS data can mature, any conclusions will be based on incomplete data.
• Does the purchase cycle differ across segments (say, expensive products with a long consideration period versus impulse, low-cost purchases)? A blended ROAS across the whole campaign could mask very different behavior in different segments.
• Are there any automated rules or scripts making decisions (pausing, adjusting bids) based on the last 1–2 days of data? It's worth explicitly checking their logic against the real conversion delay so they aren't reacting to structurally incomplete numbers.

A quick primer: how conversion delay actually affects bidding

Smart Bidding (including tROAS) is calibrated on historical conversion data — based on past clicks, the system estimates the probability and expected value of a future click. If the real cycle from click to purchase runs several days, the system simply doesn't have a full set of "fresh" conversions for the last few days at any given moment — they'll show up in the data later, retroactively.

This creates two distinct but related effects:

1. Recent-day reporting looks worse than it really is. ROAS for "yesterday" or "the day before" will look lower than it really is, since some clicks from that period haven't converted or been recorded yet. Google explicitly addresses this with a dedicated, official feature — conversion delay estimates: the system estimates how many conversions from recent clicks are still "in transit" and factors that into bid strategy simulations and impact estimates, rather than showing structurally incomplete numbers as if they were final.
2. The bidding logic itself can also fluctuate temporarily, not just the reporting. If the conversion cycle is long (say, 5–7 days), the system genuinely sees fewer confirmed conversions at any given moment than will actually materialize — and it can temporarily under- or over-value a given segment until enough "matured" data builds up. This is one reason Google recommends waiting 2–3 weeks after meaningful changes (a new target, a new budget, a new structure): the algorithm needs not just to gather data, but to wait for that data to stop being distorted by the lag.

It's worth not confusing two similar-sounding terms: conversion cycle is the real time between click and purchase (user behavior), while conversion delay is how long it takes the platform itself to find out about that conversion (which can be even longer than the conversion cycle if there are delays on the offline-conversion-import or CRM-sync side). In practice, both factors stack up and determine how "incomplete" recent data looks.

Possible approaches

• Standard practice is to exclude the most recent days/weeks from analysis (based on the median and "tail" lag from the Time Lag report) and compare more "settled" periods, rather than judging by the last week.
• Google recommends setting the conversion window based on the real purchase cycle, rather than leaving it at the default — if products are expensive and the decision isn't instant, a narrow window will systematically understate ROAS.
• After meaningful changes (target, budget, structure), it's worth budgeting at least 2–3 weeks for evaluating results — this is a general Google recommendation specifically because of the combined effect of learning and conversion lag, not just "a habit of waiting."
• If decisions about the campaign happen frequently while the purchase cycle is long, you can track faster leading indicators alongside final ROAS (add-to-cart events, checkout starts) as a directional signal, understanding this is a proxy, not a replacement for the final ROAS number.
• For segments with meaningfully different purchase cycle lengths (fast, cheap items vs. slower, considered purchases), it can make sense to evaluate ROAS separately for each rather than using one blended figure that hides different underlying dynamics.
• If a meaningful share of valuable purchases happen outside a reasonable conversion window (a very long cycle — say, for expensive products bought with financing or after a consultation), consider offline/delayed conversion import as a complement to Google Ads' automatic attribution window.
• If your account has automated rules reacting to daily metrics, either exclude the most recent 2–3 days from their logic (mirroring manual analysis) or move them to longer rolling windows (7+ days), so they don't make decisions based on structurally incomplete data.