We’ve seen one simple mistake ruin SMS surveys again and again. It happens only with text messages, but if you plan a text survey, we’d love to help you avoid this mistake and improve your response rate.

Let’s use a mobile operator in the Philippines as an example. They might want to conduct a mobile survey to their loyal customer base about their multi-channel behavior. Instead of using an online survey tool like Survey Monkey, they go with an SMS Survey because it’s important to reach everyone, regardless whether that customer is online or not.

Here is the first SMS template they may be sending out:

Lara, we love that you’ve been with us for over a year.
Thank you for trusting us!
Would you do us a favor and answer a few simple questions about when you use your mobile phone?
If you complete the survey, we’ll send you immediately a load top up of P20.
Reply with “yes” if you want to take the survey.

The basics are fine: The message is only two SMS parts long, which our UI will readily tell you. It’s also plain 7-bit GSM 0.38, which means there is no chance of funny characters showing up.

The company sent it from a virtual phone number, so that customers can reply. It also starts on a positive note and then promises a strong incentive, an instant airtime reward. Finally, the call to action is clear. Presented with this invitation, plenty of content customers feel compelled to participate in the SMS survey and reply. Great!

And then the entire SMS customer survey goes down the drain with a message like this:

Thanks so much for participating, Lara!
Here’s the first question:
How often do you use your phone while watching TV?

This question is almost guaranteed to ruin the text message survey. Worse than that, it leaves the company with unhappy survey participants—which reduces the chance they’ll be helpful next time.

But wait, what’s so bad about this question?

For a start it hurts the survey because it makes it near impossible for a mobile survey tool to figure out if someone actually answered the question.

Let’s look at some example responses inspired by what we’ve seen:

I don’t watch TV, but sometimes my brother.

Two times only yesterday, after 5 pm.

I don’t have a TV.

Who is this? Why are you sending me messages?

What you want from a bigger survey are two things.

First, you want the incentive to be paid out quickly to every participant. To make that happen, you need to know if people actually answered the question. For example the last reply isn’t actually an answer. But how would you know?

Second, you want automatic and real-time reporting. For example, the mobile operator might want to know what percentage use both TV and phone “never”, “sometimes”, “most of the time” or “always”. But what people answered isn’t even related to those ideas. With the way the question was asked, both reward payouts and instant reporting aren’t possible.

Instead, the mobile operator must tell its customers exactly what the valid answers are. It can’t be ambiguous.

Here is the question rewritten:

Thanks so much for participating, Lara!
Here’s the first question:
How often do you use your phone while watching TV?
Reply with “never”, “sometimes”, “often” or “always”.

To make sure participants understand that only the four words are to be sent, you might uppercase them and remove the quotes:

Reply with NEVER, SOMETIMES, OFTEN or ALWAYS.

This is good—but it’s not great yet! To reply, survey participants have to type “a lot”. The more they have to type, the less likely they are to respond. If it looks like work … you know how it is. So you don’t want to make it hard to respond, you want to make it super easy. And that means single character replies are your best option.

Another problem with words is typos. “nveer” is not “never”. With single-character responses you avoid those problems, too. Not convinced? Read how Unilever verified this (among other findings) in their recent Indonesia campaign.

To sum it all up: what would be even better is to make the answers numeric like so:

Reply with a number:

1 if “never”
2 if “sometimes”
3 if “often”
4 if “always”

Now we have a question message that is designed to deliver automated reports and payouts. But, we’re not done yet.

You can further increase the chances to get valid replies by using an “error message”. An error message is a feature of the engageSPARK survey system. In case a survey participant replies with something invalid, a special message is sent to the participant explaining in more detail how to reply.

For example, the mobile operator could add this error message to the survey:

Sorry, that’s not a valid reply.
Please only reply with a number
You must reply with one of these numbers:

1, 2, 3, 4

Finally, the engageSPARK text survey system helps detect valid replies by ignoring casing of words and special characters and it also allows aliases or synonyms.

For example, instead of only “3“ for “often”, you could make the survey accept all these messages as basically the same reply:

• OFTEN
• many
• many time!!
• many times
• mostly
• most times

Refining questions, using error replies and advanced matching techniques are important tools to conduct SMS polls successfully. This then allows you to use automatic, instant rewards and real-time reporting.

Have more questions about engageSPARK SMS Surveys? Chat with us now!