The Annual WordPress Hosting Survey - Kevin Ohashi/ Review Signal

Pretty comprehensive, much discussed (and commented upon/ maligned) in the WP hosting community

Some usual names here - Clourways, GD, Siteground... but some new ones I came across too.

Squats are the new Push-ups

Thanked by (1)bikegremlin


  • Man, all the advertised options are so expensive...

    Thanked by (1)Ympker
  • vyasvyas OG
    edited July 2022

    @Unixfy said:
    Man, all the advertised options are so expensive...

    Ha ha

    Not just that if you see the priciest tiers… the higher you pay, the less optimal the performance.

    Squats are the new Push-ups

  • YmpkerYmpker OGContent Writer

    @Unixfy said:
    Man, all the advertised options are so expensive...

    Yeah, often enough such reviews only include hosts with a generous affiliate program (usually the big players and not LE/more affordable hosts) to generate revenue (if they are not being paid for their benchmarking/testing like e.g. @bikegremlin would be).

    Thanked by (2)bikegremlin Unixfy
  • Thanks for mentioning my work :) Hopefully it's not too maligned.

    I can address a few of the things mentioned here.

    Prices: yeah, they are quite different than the lowend community. I imagine even at the <$25/month tier people might think they are too expensive. It's targeted towards a different market segment, and goes up to most expensive enterprise segment.

    Price and Performance, not sure how you come to the conclusion of less optimal performance with paying more. If I look at <25 and enterprise tier I see pretty major differences. The slowest enterprise tier WebPageTest is .84 which would be a little better than average for <25. While 6/8 enterprise plans would be 3rd or faster compared to 24 plans in <25 tier. In the Load Storm test the top 6/8 enterprise plans would all be second fastest p95 compared to 24 plans in <25 tier (while handling 5-6x the number of requests). Some of the benchmark scores like queries per second drop because of architecture changes. When you need to scale and build high availability systems you start separating things like database and web servers. That introduces a significant amount of latency versus the same machine, but that's the trade off for scale/reliability. They can certainly deliver content to the users faster on average. I can also share some research I did a while ago from older benchmarks trying to answer the question if price and performance are correlated (tl;dr: yes - price and performance positively correlate) :

    As far as hosting reviews and high paying affiliates. It's a pretty garbage space in general and I will be the first to say it. I do have affiliate agreements with basically any company that has a program. For these benchmarks the companies pay a nominal ($100 for cheapest tier and $500 for Enterprise tier) fee to participate. I've worked hard to try and create a brand of transparency because I think it helps with trust. These tests have everything documented in the Methodology page. The fee to participate at every tier, any extra fees paid for additional tests, the process of each test including a copy of the test site for download, all the k6 scripts used to run the benchmarks are open source on GitHub and I am happy to answer any further questions. Yeah, the hosting review market is scummy. Also, I am trying to change that and part of that is coming to places like this and hopefully constructively engaging with skeptics. I do what I do because I was a skeptic, a motivated and annoyed one who is trying change things.

    Thanked by (4)bikegremlin vyas _MS_ Ympker
  • @reviewsignal
    Thanks for the awesome work and the above explanation.
    Without getting too much into rationale behind my comment on optimal performance (or not) - here is my take: the BM results for some of the providers in the results indicated lesser (PHP bench for eg, stating as an example only, not an absolute).

    In light of the link you posted - , which I had not perused earlier,
    I am happy to revisit my statement .

    Re; "Maligned" - that was tongue in cheek- to quell any skepticism even before it came up - I am sure you receive plenty of it, some for good reasons, others not so much. I leave it at that.

    My biggest takeaway from the results is - something I have stated on another forum:
    "What can we learn from the top providers? What stack do they use? What best practices can be replicated?"
    Many providers here list offers that are way below the lowest tier you have published, but that does not mean they (or others) do not have beefier plans- your response(s) in that context are appreciated.


    Thanked by (1)bikegremlin

    Squats are the new Push-ups

  • Thank you again and I'm happy to explain and engage.

    You're right on the PHP/WP Bench scores. Some of the lower priced companies on single server setups can really blow away the numbers since they are single threaded tests. It's also why I treat PHP/WP bench scores as a non-impacting metric. They are interesting and can help compare some things, but they don't actually tell the story about real world performance because architectures matter as I explained above.

    No worries on the tongue in cheek statement, I've certainly had the occasional drama here and there. I'm always willing to talk about them, I'm not infallible, every year I try to improve the process and do better. So I appreciate when people bring new ideas, weaknesses in my current process, etc to me. I keep a changelog in the methodology to help show how the tests evolve over time too.

    I hope the results really do add value and help consumers and hosting companies. There's surely a lot to be learned in all the details. If you have any questions I'll do my best to answer (publicly or privately).

    Thanked by (2)vyas bikegremlin
Sign In or Register to comment.