Argo making image response time slow, adding 600ms delay

After enabling Argo on a couple of our servers to try it our we noticed a big jump in image load times in Google Search Console.

I am talking about the “Settings - Crawl Stats” section, viewing the “Image” / “Average response time” data.

In both cases the average response times jumped from around 400ms to 1000ms.

My main problem now is that if we disable Argo our image response time jumps to around 1800ms !

I have disabled Argo on one account and have seen this is now being reflected in the GSC. I am hoping this issue may resolve itself back to normal and our response time will return to 400ms soon.

I have set up an AWS server in the USA to ping some test images and can confirm the same results as Google is reporting.
Our servers are in Sydney and our local response times within Australia remain good, around 150-400ms as expected for local traffic.
Here I am talking about the “total_time” which included the content download.
Images downloaded from the USA are reporting a “starttransfer_time” of around 1200ms.

I opened a ticket with support and was devastated to hear back that a ‘1200ms’ connection time from USA to AU was acceptable. After 2 weeks back and forth I could not get them to understand the situation, and that we were getting 400ms.

Has anyone else noticed Argo affecting their site response times like this or seen a huge negative affect turning Argo off?

This shows one account where we have left Argo enabled:
Capture_x

This shows the other account where we switched it on around the same date but switched it off again a few days ago:
Capture_x2

  • Scott

Depends on how you’re measuring Argo’s benefits. Argo improves the server response time metric. Closest you can get to measuring that with Google Analytics is the Server response time metric. You can see my Cloudflare Argo experience and measures via Google Analytics/Google Datastudio and Cloudflare Argo analytics API custom script I wrote at WebPerf - Cloudflare - PageSpeed - Cloudflare Argo Smart Routing In Action | Centmin Mod Community Support Forums

Thanks for your reply.

We are measuring the “server response time” manually and looking at the overall download time from the Google crawl statistics.
Both of these became much worse when we enabled Argo, and unfortunately worse again when we switched it off.
All we really hope for now is to return to where we were a couple of months ago.

In case anyone is following this thread I will keep updating my findings.

Google Search Console just downgraded our Core Web Vitals.
We were at 100% but we now have 50% of our pages failing Largest Contentful Paint (LCP).
LCP are images which are taking too long to load so I assume that’s related.

I am upgrading one of these servers to the Business Plan today to see if that helps although we have servers on the Business plan which also have long connection times but we are desperate to fix the issue as Google is about to start using Core Web Vitals in their search algorithms next month so this couldn’t have happened at a worse time.

If it fluctuates even with Argo disabled, then that suggests the TTFB slowness is origin server backend related if you are not caching dynamic HTML requests as Cloudflare by default only CDN caches static files and not dynamic HTML unless you tell it to do so via Cache Everything + Edge Cache TTL page rules i.e. common one is slow MySQL database queuing up PHP requests which slow down server response times.

Thank you again for the reply.
I am only showing the Google report for images in these graphs. I understand anything else can be greatly affected by server side processing.
I wish I could get back to where I was before switching on Argo.
-Scott

Latest update.
The server where I disabled Argo again seems to be on a TTFB rollercoaster, not recovering to where I was Pre-Argo.
(The first line is where we enabled Argo, the second is where we turned it off again in the hope of returning to where we were before with TTFB.

The second server I have tried a new approach to recover to pre-argo delays, on the server we upgraded to the Business Plan however the delays have not improved, if anything things seem to be getting a little worse.

If anyone had had a similar experience with enabling Argo and has managed to recover I’d really like to hear form you. As the old saying goes, if it ain’t broke don’t fix it.

  • Scott

Another update.
The server which has Argo still enabled is getting worse, it’s approaching a 1500ms delay now.
Google is rolling back the crawl budget, we have seen a 50% reduction in Google crawling images over the past week. Hopefully this rectifies itself before Google gives up on us altogether.

The server which I have switched argo off: It never recovered to pre-argo days.
The delays for Google are now so bad I’m worried Google will stop crawling this site soon so I’m switching Argo back on today. I’ll keep you updated.

  • Scott