5+5 Tips to ROCK Onsite Surveys - #MeasureCampV

MeasureCamp V is over, but I must say It's been the best of the 3 I've been so far, mainly to the quality and quantity of new topics covered during this great "un-conference".

For these who don't know MeasureCamp it's an analytics event that happens twice a year with a very unique format: There are 8 rooms, 8 time slots and a board, and anyone can lead put a card on the board and lead a 20 minutes session.

MeasureCamp V Board Pic from https://twitter.com/D1PZ/status/513357079344852992

MeasureCamp V Board Pic from https://twitter.com/D1PZ/status/513357079344852992

This year I wasn't planing to drive a talk, and definitely I didn't have anything ready, but I was surprised that except a session from Dan Barker there was nothing on VOC analysis, which I believe is a very important part of my job as Analyst and Conversion Rate Optimiser.

SO, I decided to prepare something quick and simple:
5 + 5 Tips to ROCK on-site surveys (5 from me and 5 from you)


The first thing to call out is that in this session we talked about a very specific type of Onsite Surveys: the short ones you usually see on the bottom right of your screen. We are not talking about long off site surveys (like the ones you do with SurveyMonkey) nor the ones with an invitation message to answer later like 4q or userZoom.

We mean these type of surveys that we use mostly in CRO:

That being said, to nail them and avoid as many mistakes as possible, these are the 5 + 5 tips:

TIP 1: Never ask for the "future"
As tempting as it might be, never ask the user what would they do "if you offered them something". In clickstream analytics we already see WHAT people do, what we usually want to do in VOC is to understand the WHY. Let's see an example on a exit survey for users leaving the checkout:

Don't ask: Would you buy if we offered a discount (you can track that)

Ask: WHY are you abandoning the checkout process?

The difference is huge.

TIP 2: Always have an open field

Try not to limit the number of possible answers to a list, no matter how sure you are about it. For instance, if you ask Visitor Intent: "Why did you come to visit this site today?", you'll have as possibilities: 

  • I'm just browsing
  • I'm Researching about a product
  • I want to purchase a product
  • I am managing my account

If you don't add an open field you'll be missing the best insights: those that you don't know. Any time you learn something that you din't know that you don't know, that's value for the business. To unlock this, use open fields in your short surveys.


From the audience: "Use these learnings to review and iterate your survey question and answers."

TIP 3: Segment! 

Asking in your most visited landing is tempting, as we will have more answers earlier. But the results will be difficult to understand. In the other hand, by segmenting who sees the survey you can get much better results. Segment by content (page visited), user type (new / returning), or behavior.

From the audience: "Use GTM to segment by event fired, wohoo!"

TIP 4: "Analyse the survey results as an Analyst, Report like a marketeer"

The tip from my side was to be careful on the reporting and face the biases that often these surveys fall in. But during the discussion this tip was so improved by the attendants in the room that we ended up with this:

Martijn van Vreeden encouraged us to treat and report them as Market Researches do. Pro tip.


My last tip was to push analyst to do more of this. Sometimes we overthink tracking and tagging when we could use surveys to understand better the behaviors on the site. They are cheap (or free), low risk (just make sure they don't block an important CTA or message) and the results are game changers.

(The next 5 tips came from the people in the room)

TIP 6: Keep them running

Tip from Martijn. Don't just use them for CRO. If you identify relevant metrics for your dashboards in form of VOC (task completion, visitor intent, satisfaction, etc) keep them running to a sample of your audience (it can be as little as 5%!) and bring VOC to your dashboards.

TIP 7: Test, test, test

I belive it was Charles who recommended us to test them! Test the question, the answer, the placcement, the segment, test until you max out your completion rates and have data as accurate as possible! + 1000!

TIP 8:  Try them first OffLine

I loved this one. Related to the "test, test, test" try your questions offline with the people you have around. You'll immediately know if the question is clear, the answers are right, it's engaging and the tone is correct. This will save you lots of headaches!

TIP 9:  Inform if you will be answering the open fields 

Very often users will complain or ask support using open fields in surveys (there's a fine line between these surveys and feedback tools and not everyone gets it). To prevent someone from posting a question in the survey that might not be replied, alert your users if you will be responding or not.

TIP 10:  Last but not least.... Integrate your survey with your Analytics tool! 

This will allow you get much more out of the analysis of the responses. Segmentation, conversion, testing... you name it. And be creative with the use of GTM. There's a whole universe there to explore.

Let me finish this post by thanking the 15+ attendants to the sessions, those who participated and shared these tips specially, and Peter O'neill and the fantastic organizers of MeasureCamp.


Do you have any other tips? Share them in the comments!


In case you are wondering, MeasureCamp returns on March 14. Meanwhile, checkout some more amazing CRO posts in my blog.