Google Fetch as Googlebot - Beware!
This is a pattern is what I would like to call "fetch shock" and is especially noticable when using the "Crawl this URL and its direct links" function.
Rank is established gradually as your content gets processed by the Google algorithm. It's not done when the site is crawled. Instead, the crawler simply reports back data and the data is set aside against your domain and processed when system resources are available. The part of the algorithm assigned to building statistical profiles for your website using data heuristics builds an impression of your site, which is then compared to other sites along a complex continuum - that's called "rank".
By using Fetch as Googlebot
during or after processing of your site, you can temporarily or permanently 'upset' the current status, because you force Google to rapidly submit content via the crawler to the processing system. The assessment of that content may or may not 'agree' with where the algorithm was up to immediately before you used the Fetch command. In fact, it's possibly going to cause a sudden rush of 'unknown' data to the analysis. That is most likely to immediately drop your rank.Why does it drop?
Because the rapid submission of content creates a rush of unprocessed data. The ratio of unprocessed to processed data bulges toward unprocessed based on whatever content differs from that in Google's processed cache. This volume of your site will have a zero contribution. When you have more new data, and that data has zero processed value, your median also moves toward zero. Ergo - you lose rank.
This is not to say that using Fetch as Googlebot is a bad idea. You just need to be aware that it can cause some degree of Fetch Shock.Will you recover from it?
No. Your exact position may never be the same again. It will depend on the contributing value of the new data, once that data is processed. Your rank may improve if the balance of the data weighs positively once processed. The opposite is also possible.What should you do?
There is no do or don't in relation to using Fetch as Googlebot, but if you haven't prepared your client for the possibility that it can have a sudden and dramatic negative effect (which may recover just fine, or not), don't Fetch. Patience is a virtue. Successful Fetch will almost guarantee rank. It doesn't make any promise as to where that might be. And it could be at position 258, not #1.
Image: Fetch as Googlebot executed on the day the green "impressions" line took a dive. Partial recovery indicated at the end of the graph. Impressions and position indicated for a single keyword. Data source and image from Google Search Console. Learn about Google search Console here: https://goo.gl/8fKWSO