In short
- Is: the days, and touchpoints, from an ad interaction to the conversion. From the first touch it sizes the journey; from the last click it says when a day's results are complete
- Makes recent days look worse: on click date, yesterday's spend is complete but its orders are still coming in, so its ROAS keeps rising for days or weeks
- Platforms: Google Ads dates conversions by the click in its standard columns and suggests date ranges that ended at least 30 days ago. Meta's longest click window is 7 days
- Has a long tail: in our example, a quarter of what a Monday's clicks bring arrives after day 7, too late for any 7-day click window
- Shortened by Safari: if 40% of that late quarter comes from Safari buyers who hadn't visited for a week, a tracker with a script-written ID loses €240 of it to Direct and measures a shorter lag
How time to conversion is measured
Time to conversion is the time between an ad interaction and the conversion it led to, read per conversion and then across all of them. Where you start the clock decides what the number tells you:
- From the first touchpoint: how long the whole customer journey takes. This is the lag your attribution window has to cover.
- From the last click: how long buyers take after the touch that brought them back. This lag decides when a day’s results are complete.
Touchpoints count too: a 13-day journey with four touchpoints and a single click 13 days before the order are different problems. Google Ads’ path metrics report shows both, days and interactions to conversion, measured from the first or the last ad interaction, but only for the keywords and ads in your Google Ads account (Google).
“Would average days to conversion and touchpoints per journey help?” a customer asked us in a call this year. The days do, read as a distribution rather than an average. Lags are skewed: some orders come within a day of the click, others weeks later (Google), and one slow buyer drags an average up. Read the median and the day by which 80% or 90% of revenue has arrived, and split both by new and returning customers and by channel: prospecting and brand search have different lags by design. Touchpoints per journey say more about your tracking than about your buyers, as the touchpoint entry’s section on touchpoints per journey shows.
What conversion lag does to recent numbers
In a video we recorded for a customer this year, the explanation fit in one sentence: for that shop, today’s results came from spend made about two weeks earlier. Every report has to put each order on one day, and whichever day it picks, the lag shows.
On click date, the order goes to the day of the click that led to it, so recent days are unfinished: their spend is complete, their orders are not. Google Ads dates conversions this way in its standard columns (Google). Google warns that recent performance can look weaker because some people who clicked your ads haven’t converted yet, suggests date ranges that ended at least 30 days ago for complete conversion data, longer with a longer conversion window, and offers the segment “Days to conversion” to see how long your buyers take (Google). Smart Bidding’s bid strategy reports mark the dates for which more conversions may still come in.
On conversion date, each day is complete, but its revenue comes from clicks of earlier days. The lag shows up whenever spend changes: a budget increase looks worse at first, a cut looks better, and a paused campaign keeps earning for a while. Which date to use for what: click date vs. conversion date. Refunds and platform corrections move past days too: data freshness.
Platforms also cap the lag they can credit at their window. Meta’s longest click window has been 7 days since it dropped 28-day click in 2021 (Meta), so a purchase 9 days after a click is no click conversion in Ads Manager, whatever your own tracking shows. Google Ads counts up to 90 days after a click, 30 by default (Google).
A worked example: one day of clicks, 28 days of orders
On a Monday, a shop spends €1,000 on Meta prospecting. Over the next four weeks, that Monday’s clicks bring orders worth €2,400 net of VAT, a ROAS of 2.40, but not at once. With an assumed lag curve, which you would replace with your own, Monday looks like this by click date:
| Days after the click | Share of the final revenue | Revenue of Monday’s clicks | ROAS |
|---|---|---|---|
| Same day | 30% | €720 | 0.72 |
| 1 | 45% | €1,080 | 1.08 |
| 3 | 60% | €1,440 | 1.44 |
| 7 | 75% | €1,800 | 1.80 |
| 14 | 88% | €2,112 | 2.11 |
| 28 | 100% | €2,400 | 2.40 |
Four readings:
- Yesterday is unfinished, not bad. On Tuesday morning, Monday shows a ROAS of 0.72, less than a third of what it will reach. Compare days at the same age, each day after 7 days against other days after 7 days, or divide by the share that has usually arrived by then: after 1 day, 1.08 ÷ 45% = 2.40.
- A 7-day click window sees three quarters. Meta’s longest click window can count at most €1,800 of this revenue as click conversions; the €600 after day 7 is outside it.
- Safari shortens the tail. Say 40% of the revenue after day 7 comes from buyers on an iPhone or in Safari who hadn’t visited for a week. A tracker whose visitor ID is written by JavaScript has forgotten them: €240 lands in direct traffic, Monday tops out at 2.16, and the measured lag looks shorter than it is, because the late buyers it lost never show up as late.
- Conversion date lags the budget. With €1,000 every day, each day collects €2,400 from four weeks of click days, a ROAS of 2.40. On the day the budget doubles to €2,000, it reads 1.56: the new clicks have brought 30% of their €4,800 so far (€1,440), the older days €1,680. Nothing got worse; the new budget’s revenue is still on its way.
Choosing an attribution window from your lag
Set the window to cover the first-touch lag of your new customers, the day by which most of their journeys, say 90%, have converted, as the attribution window entry’s section on which window fits explains.
Measure that lag with identifiers that outlive it. A tool whose visitor ID a script writes forgets every buyer who stays away for more than 7 days, on 24% of web traffic (ITP, WebKit), so it can’t observe a lag with a quiet week in it: its data suggests a short window and hides the journeys that would prove it wrong. A platform’s numbers have the same blind spot: they never show a lag longer than its window.
How adtribute handles lag
- The window is part of every attribution model: a Custom window of the last 7 days leaves Monday’s clicks out of the journeys behind the late €600, and Unlimited keeps them in (the attribution window entry’s section on how adtribute sets the window, docs: Attribution models in adtribute).
- On click date, the last days fill in late: your whole history updates every night, so recent days keep growing as their late orders arrive (docs: When your data updates). How to switch a board or analysis to click date is in the click date vs. conversion date entry’s section on how adtribute dates conversions.
Our view: two lags, two jobs
Our view: measure two lags and give each one job. The first-touch lag of new customers sets the attribution window; the last-click lag of all orders sets how long you wait before you read a day. Mixing them up leads to two mistakes: a window chosen from how fast returning buyers order, too short for the journeys that bring new customers, and a campaign cut after one day because its click-date ROAS looked bad. Until a day has matured, compare it only with days of the same age, and if you have to decide sooner, project it with your own lag curve instead of reading it raw.
FAQ about Time to Conversion
How long is our customer journey really?
Measure it rather than assume it: the days from the first tracked touchpoint to the order, for new customers, with tracking whose identifiers survive longer than the lag you are looking for. A tracker that loses Safari visitors after a quiet week can’t show a longer journey, so a short measured lag can be an artifact of the tracking.
Why does yesterday’s ROAS keep rising?
Because it is read by click date, which books every order on the day of its click: yesterday keeps collecting the orders its clicks bring today and in the weeks after, as the click date vs. conversion date entry’s comparison of the two dates explains.
How long should we wait before judging a campaign?
Until most of its conversions have arrived. Google Ads suggests date ranges that ended at least 30 days ago, longer with a longer conversion window, for complete conversion data. Before that, compare days of the same age or project them with your own lag curve.