This website works best with JavaScript enabled. Learn how to enable JavaScript.

VWO Blog

A Blog Around Everything Conversion Optimization And Beyond

Never miss an update from VWO

A value for this field is required.

A lover of the written word, I plan to be the planet's first sit-down comedian. When I am not rethinking a misplaced comma, I write about conversion optimization and website usability. You can follow me @mohitanagpal

Comments (52)

Leave a Comment
  1. Interesting finding! I’ve seen similar results, but mainly with the difference that for New Visitors social buttons had increased conversion, but for returning visitors (that were probably more goal oriented and only distracted by the SM buttons) it had a decreased conversion effect.

    Do you see similar differences between new and returning visitors?

  2. In my opinion the reasons that are stated against
    social sharing buttons are worth testing themselves:

    1. Negative social proof: Test Social Sharing Buttons with Counter against those without to verify the assumption.

    2.Distraction from main goal: Test different positions for the social sharing buttons and more particularly try to display them in one row, maybe use a frame around them, so the visitos recognizes them as a special “social sharing area”.

    All though I find this case studies very interesting, I would like to have some insight into the actual test-charts and specific data like conv./visitors. Because success like getting statistical significance after one week is great, but hard to really judge without further information.

  3. @Aiden,

    This test ran on more than 10,000 visitors and the result was statistically significant (95%). We don’t always have the liberty to share specific data points. Some users allow it, most don’t, so sorry about that.


    We had some issues with images not being propagated across multiple servers. That’s been resolved and you should see all images.

  4. Very interesting case study Mohita.

    I’d be interested to see a similar test run across multiple verticals.

    Since different customers in different verticals can behave in a different way it would be interesting to see this.

    It would also be good to see different data points you have tested, understand that you can’t always do this as it depends what the customer allows you to reveal etc.

  5. The only issue is that though this test may prove true, removing social media buttons may decrease the inflow of new customers to the page. So whole you’re widening the bottom of the funnel, you may be narrowing the top of the funneld. I’d like to see the front end traffic statistics alongside these numbers as well before judging the relevancy of this particular A/B test…

  6. @Al,
    I used Open Site Explorer to check Taloon’s number of social shares. The numbers are extremely low which prove that the website doesn’t get a lot of visits through social shares. So removing these buttons doesn’t really make any difference to the inflow of new customers. You can check the report here

  7. Very interesting post!

    It had not occurred to me before that the zero-shares social share buttons were acting like a negative social signal, but now that you bring that up it makes sense.

    I do have a number of social share buttons on the Additive Analytics blog. (That said, every post is shared at least by me.) I figure that that content is very shareable, and I’m bound to get more pageviews if folks are sharing the posts to their own networks!

  8. Going against the norm can sometimes yield better results. As online shoppers, we are naturally more reluctant to purchase products that we feel are not trusted… hence, no social signals may indicate the lack in trust in the product.

    Interesting post.


  9. Mohita,

    I’d be very interested to know if the change held up over time. On our own (very large) website we often see statistically significant shifts in behavior when we change even small things, as users are ‘reacting’ to the change. We have found that the period of a week is not long enough to gauge statistical significance — in our case we’ve found four weeks to be a much more reliable measure. What we tend to see is a spike in changed behavior, which slowly moves back to the norm.

    Do you know what happened over time here?

  10. @Jesse: I suggest to always remove the first few days of data of a test anyway to remove users from the test that respond to the change itself instead of the new variant. It also helps if you only show the testvariants to new users only.

    Also: Strictly speaking you can perfectly reach statistical significance by testing one hour. On the other side you can just as wel don’t reach statistical significance by testing over 4 weeks. It all depends on your traffic and conversion percentage. But the question of course is: What do the results mean, what is the effect size and how useful is it over the long run.

    We tend to always measure at least 2 weeks because we often see a difference between weekday/weekend so we can compare that as well. If you have a lot of traffic you can scale down the percentage of total users participating in the test (we aim for 150 conversions per testvariant).

  11. @Jesse

    Thanks for asking the question. I have asked the person who ran the test to come here and offer some perspective.

  12. Is it possible the social sharing was slowing down the page load time? I know sometimes the social icons can take longer to load, causing the page to seem like its still loading, therefore turning users off.

    I would also be interested to see some extended referral data. For example, if you turn off the social sharing icons, do your social referral stats decrease over time?

    This is a much more difficult test to run, because of other features the icons affect. I’ve run a very similar test and found similar results, but I’m not concluding its a loss just yet. Keeping an eye on the other factors.

    Thanks for sharing!

  13. @Megan
    A similar point was made by Al in the comments and I had checked OpenSiteExplorer to find out that Taloon had very low social shares which meant that hardly any visitors were coming to the website through social shares. Here’s the report if you want to take a look.

    Had the number of social shares been high, then it would been a different story. About social sharing buttons slowing down the page, that’s a possibility. There’s at least one case study that says these icons increase site loading time (for mobile devices)

    I will be very interesting in knowing more about the similar test results you mention.

    Keep the valuable comments coming 🙂


  14. It’s a shame this article has such a misleading title. Sending an e-mail news letter claiming that “Removing Social Sharing Buttons Increases Conversions” based on 1 case study is just wrong, especially for a company that creates tools for a/b testing, something that should help people to NOT base decisions on one factor only. I’m disappointed.

  15. Nice post, and nicer to see you follow up on questions Mohita.

    I don’t have a pre-disposed opinion on the subject but I shared the same initial thoughts of some others who wondered about a potential narrower funnel. But your data made it clear that wasn’t a concern here.

    Still I wonder how this applies to different scenarios. For example, Talloon is a reseller with a wide assortment of products, like a Best Buy or Walmart or even Amazon. Given the breadth of their product line and the fact they are not the originators of most (any?) of their offered products, it would seem significant social sharing to be unlikely.

    But what about companies that have a single product or a small product line for which they are the originating vendor, like Nest and it’s thermostat and smoke detectors, or Wingify itself? Would social sharing hurt or help total sales?

    And compare them to companies that have huge product lines like Ikea and Samsung where they are the originator of the product; would social sharing help or hurt there? Most (all?) of those types of companies use resellers so maybe their goal is brand awareness which isn’t a directed goal but maybe their main CTA is to locate a reseller?

    Of course it will be hard for you to test for those latter companies unless you happen to get them as a client *and* have enough clout to get them to change their websites to test it, but maybe you could test this on the site to see how it works for you?

    1. @ Mike
      Thank you so much for liking the post and sorry for getting back to you a little late. You are right, we can’t really say if an Amazon and Ikea would also get similar results if they remove their social sharing buttons from the product pages. There are many factors at play here and we don’t encourage anyone to blindly make changes without factoring their own goals and circumstances. As for VWO, we don’t have social sharing buttons on our funnel pages as it could be a distraction. But we have them on our blogs and that really works for us.

  16. With regards to negative social proof, the sounds about right, but here is my dilemma:

    If you put Like/+1 buttons on your page, initially you will have zero likes/shares/etc. In this case, it does not help, as was mentioned above, it could be that people do not trust the product or the quality of the product.

    In this case study, it had a positive effect to remove those sharing buttons. However, something else to consider is what if there were thousands of likes/shares/+1’s? The assumption is that it will boost sales (but this should obviously be tested).

    The dilemma is that if you do not put those social sharing buttons on the site, you will never get those tens, hundreds, or thousands of likes/shares, especially if your product is great and has good reviews. I believe that in this case, you (or the company) has to figure out some way to utilize social media in conjunction with the product.

    Anyone have any thoughts on this?

  17. I really agree with the Negative Social Proof point. 0 likes = no one likes this?
    Why does no one like this?
    Does it suck?

    People who are just about to buy something a vey unlikely to want to start endorsing it on FB before they even have it. Getting people back to your site once purchased so they can click like is very un-likely to occur. Marketers should take their heads out of the sand and stop thinking that what works for mega sized ecommerce sites that have a tiny % of social junkies that equate to a few dozen item likes, will work for lower volume sites.

    Really, do you think a person who is buying a big snow plow machine is going to “like” it on FB?

    1. @Adam

      I think one way to make people click on that Like button after they have made their purchase is by optimizing the Thank you pages. Instead of a simple ‘thank you’ note, the customer can be directed to a Thank You page. You can have a social share widget there and ask people to post a comment or tweet about their new purchase.

  18. Very interesting! But I guess its a chicken-n-egg situation..You talk about negative social proof, but unless those sharing buttons are not there, how are you going to GET any social proof. What would be better is to understand from analytics that which Social network is driving more traffic to your website & having a sharing button ONLY of that network. I think that would make the difference.

    Another ideal test would be to compare conversions on a product page with high social proof v/s conversions on the same page without social proof.

    1. @ Jitendra
      You are right! Analytics is the key here. In this case, Taloon was hardly getting any traffic from social networks. So removing the buttons didn’t really hurt. And that’s an interesting test you suggest. Do let me know if you plan to run it sometime.

  19. This is a ridiculous blog and post.

    How can you make a statement like that? On the fact if social sharing buttons increase or decrease CR, it’s individual to site, traffic channel, product page layout and a thousand other things.

    You can’t just say that “removing social sharing buttons on product pages increases conversions”. It’s INDIVIDUAL. Learn that word.

    Please, get it right. You can’t generalize like this. Stop it.

    1. @ Albin
      I have already used the expression ‘Wow’ in the comments once, so I will refrain from using it again. But seriously. Wow! ‘Ridiculous’ is a really strong word for a headline whose only fault is that it chose to be in the present tense. We never-ever tell anyone to make significant changes like these without first testing them. It’s absolutely ‘individual’ and specific to their business and customers. So yeah, we totally understand that. We are an A/B test company. Remember?

      Having said that, I get your point about the post sounding generalized. Thanks for pointing that out. In the future, we shall keep the headline in the past tense to make it clear that it is a specific case and cannot be applied in all situations.

  20. Interesting article. I used to be of the opinion to have one clear call to action on each page and in the early days of Facebook i did not have any mention of it on our site so we it would not take a customer off site. Now i feel they are more necessary for brand exposure.

    I do feel that zero likes/tweets/pins on a product can have a negative impact. I have seen a site ( that the like button on each product will like the brand page rather than the product. So each product looks like it has 10K likes. What your opinion on this?


    1. @ Fabio
      That’s a neat trick that could help in improving social proof. But I would still say that should be A/B tested.

  21. NO surprise to me, in fact obvious!

    1) Procrastination kills “sales”, in fact any attention grabbing/distracting feature, not just social buttons do, especially during check out processes.

    2) depending on how little and how many “shares” your visitor gets to see, will already devalue your offer or at least start getting them think, Hmmmmm… which brings as back to point 1.

    3) depending on the integrity of your users, how can you expect WITHOUT them knowing more about the product/delivery/customer service of the seller, to recommend it/them to friends or in fact anyone else. This is as if standing in front of a new brick and mortar shop, never stepped in, nor interacted, not having bought anything… but calling all your friends with your cellphone or sending text messages, they’d have to check it out?! A person acting as such would be considered… INSANE! and soon lose his “friends”

    But of course knowing the exact numbers for your own site is mandatory to know HOW much you LOSE.

    THANKS, as always to the team at VO for these specific tests/numbers.


  22. @alex rucker
    Wholeheartedly disagree with your passion that this test is “obvious.”

    Specifically, you’re third point misses a large population. You’re completely disregarding the customers who have shopped with the brand or store before. For example, it is not insane for someone to share a toy they’ve bought for their child with the message “Great toy for children who love Elmo!” while purchasing the toy for someone else. Or if someone is restocking shampoo that they use everyday, they can clearly speak to the quality of the product.

    Sharing also doesn’t have to be a direct endorsement. Someone could share a product and say “I’ve been looking for something like this!” or “I can’t wait to receive my ____.” None of these circumstances would lose their friends.

  23. @megan

    this was not meant to offend you or anyone else.

    I wrote it’s mandatory to test it out on your own site, because there are so many different goals a site can be set up for, as you know.

    Depending on that site has returning customers percentage, not merely visitors, they could offer the possibility to share and not only “this is what “I” was looking for”, that doesn’t really help friends or anyone, much.

    But as long as it is still not possible to show sharing buttons ONLY to returning customers, it will lower conversions, unless if at all (test) you have far more than from your visitors anticipated likes, etc.

    Next BIG problem… your “social proof” MUST exceed the expectations of your visitors… and you don’t know what those are, being dependent on many factors. However still be distracting even on 2nd interaction (purchase/call to action, etc.).

    BTW I personally don’t share/recommend any source as long I have not purchased, used the product and experienced their customer service as well. A blog post/article on a topic I have enough knowledge in to “judge”, being the only exception.

    That quality of sharing would really be valuable not only but specifically to people we care about.

    Remove sharing on your site(s)…(A/B test) and then report back to VWO your case study how much better your site converted and you can be sure it will.

    “INSANE” was only applicable to the example offline I had mentioned, not one-to-one to online sharing.


    What would be really interesting how different sites with different goals improve their conversion removing social share and how much in comparison, not only e-commerce “add to cart” type sites. Maybe you can reach out and get such invaluable test results for your prospects/clients? Thanks in advance.


  24. @Midas
    Wow! That’s a tall claim and a very strong position to take. What makes you thing so?

    Sorry to disappoint you. Mislead is the last thing we want to do. The purpose of publishing these case studies is to put real success stories out in the public domain. We have always encouraged people to run tests depending on their own product, target audience and goals.

  25. If you go back to the website they use as an example, they have social share icons on their product pages still. They’ve just moved them lower and made them gray.

  26. kinda ridiculous headline considering the single, very specific case study and small sample size. a company dedicated to optimization should know better… then again that headline got me to this page 😉

    1. Hey Sean, You are right. We should have ensured the headline indicated that it’s just one case study. We will take care of it in the future!

  27. Interesting findings. Honestly, there are too many comments to read them all, so maybe someone mentioned this already, but:

    Have you considered testing two different social sharing icon sets? If the negative social proof is indeed the cause of the decreased conversions, couldn’t you get around this by simply using sharing buttons that don’t show the number of shares? Maybe test that and see if that affects conversions.

    If the negative social proof hypothesis is correct, you can get the best of both worlds by using different icons. You could still have sharing buttons to help attract new customers while having minimal if any effect on conversions.

    Another possibility is to only provide sharing buttons if they have already purchased the product. They buy the product, then come back to leave a review, but this time your call to action is to share the page.

    Food for thought.


  28. I agree with the concept of not showing the social sharing buttons when counts are actually very low.

    As we are aware of two different type of social sharing buttons one that has counts mention next to it while other dont have, two.
    The best work around for this is that you don’t have to show the shared count on a particular social network as you don’t have enough yet. Whenever you have enough social shared count then you can display it.

  29. I wonder if this conversion rate lift had more to do with an improvement in page load speed than it did with removing the social sharing buttons. Social sharing widgets are notorious for slowing down a page. Did measurement of page load speed factor into this test? If not, it would be very interesting to see if there was a correlation.

  30. I am not a big fan of social marketing but it seems even the search engines love this stuff. I have found many Pinterest pages on first page of search results. After reading your article on this subject
    believe you make great points on how this may be hurting actual results. This agrees with my original
    feelings on social marketing and that so much of the
    content is diluting the results. I will most likely be removing my social buttons that just added not to long ago.

  31. Would be interesting to test the social media buttons with a counter vs. the social media buttons alone as well. Since the counter was causing the negative social proof and the icons were creating the additional exit points on the page – it seems like these are two separate things to test. As for the counter number – I’d also wonder if there is a sweet spot for the number of shares it takes before it’s of value to display it.

    Another consideration for the outcome of this test may be that these hardware products simply aren’t items people would like to share, and especially people in the target audience. I mean, who would share a link to an expensive snow-blower to their family and friends?

    Thanks for getting us thinking!

    1. Exactly Dom. We share stuff that makes up appear cool on social media. We don’t exactly see people flaunting their plumbing and construction purchases on Facebook. The test you mention is mandatory before completely giving up on social sharing buttons. And I am not sure if there is an acceptable number of shares. All we know is that the more, the merrier 🙂

Leave Comment

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes : <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

Contact Us / Login