Pickup lag
Pickup lag is how long it takes for a change to a web page to show up in what the AI platforms say.
Pickup lag is the time it takes for a change on the web to show up in AI answers. You publish a new page, fix a wrong price, or get named in a review, and the answers carry on as before for a while. The gap between the change and the answers catching up is the pickup lag.
It matters because it decides when you can judge your own work. Look too soon and a change that's working looks like it failed.
Where the lag comes from
A few things have to happen before an answer can use your change. A crawler has to visit the page again. The platform has to update what it has stored about it. And then the page has to come up when the platform searches for the questions your customers ask.
Some of what a model says comes from what it learned in training, too. That part only changes when the model is trained again, which can take months. A wrong fact that comes from training can outlast the page that caused it.
The lag also differs from one platform to the next, and from one page to the next. A news site that's crawled every hour gets picked up faster than a page deep in a small website.
An example
Say Harrow & Pine starts delivering across the country and updates its delivery page on 1 March. A week later, ChatGPT still says it only delivers within London.
That doesn't mean the update failed. Maybe the page hasn't been read again yet, or other pages that still say London are outweighing it. Checking again in two or three weeks tells you more than checking the next morning.
Pickup lag in AnswerOrbit
When you mark a recommendation done, we take two readings of its numbers, one two weeks later and one six weeks later, to allow for the lag. For any other change, Before and after a date in Recommendations compares your numbers before and after the day you made it. See Recommendations.