We recently came across an account that had an overall delivery rate of 40%. Now we pride ourselves on having a 98-99% delivery rate, so this stuck out like a sore thumb. When we dived a bit deeper into the reasons why the delivery rate was so low, we found one key reason - the phone number database was of low quality.
Wish to know more about messaging through APIs? Read our ultimate guide to SMS APIs!
This meant that while paying for 100% of the messages, only 40% reached the recipient. The bigger impact was on missed revenue by way of mitigated conversion. If only 40% of messages reach the customers with a conversion rate of 5%, then out of a campaign of 1000 messages, only 20 converts, instead of a potential 50. So, if each conversion had a value of 5 EUR, that would mean a total of 150 EUR in missed revenue. But how can you avoid that?
What is a mobile number database?
First, let’s start by explaining what a mobile number database is. A phone number database is any sort of file, repository or page that contains a set of phone numbers that a customer has provided you for various purposes. Be it for marketing, transactional messages or anything else, the data consists of the recipient’s information (name, surname, etc.), the phone number itself and the country code.
The latter is quite important yet often overlooked. If it is missing and when the system that is used for sending messages does not provide an automatic add-on option, then the database will become rather useless. That's because the messages will be sent out without the country code to the first number that happens to match. The system pics the first numbers that match any country code and then attempt to deliver the message.
What affects the mobile number database quality?
Several factors affect the quality of mobile number databases. A great indicator is the industry the organization (i.e., our customer) is in. For example, there are obviously more invalid numbers in the fast loan business (e.g. payday loan companies). Also, SMS marketing databases typically have significantly more invalid mobile numbers.
On the flipside, other industries have a very low amount of invalid phone numbers. Like for example, finance, where users are highly motivated to receive verification PIN codes, so they provide correct information. Whatever the use-case though, there are still errors in any dataset.
The usual error rate is 5-10% of erroneous numbers and 3% of roaming numbers making the total out to 8-12% of the dataset. And it’s in your best interest to know which numbers can be reached before sending the message because mobile operators charge you regardless.
Now, let’s get to the point. How can you make sure that your database is of high quality?
There are quite a few things to check:
- When the number is given, have the option of selecting the right country code without the customer inserting it themselves. This can help clean up the dataset and have it be more reliably formatted when exporting or importing.
- Always check the mapping and data fields for possible errors before importing a file for sending purposes. Make sure the country codes are correct, that no additional characters have entered the data fields or that no numbers are missing. If unsure, rather keep them out than send to a number for which the message was not intended.
- If possible, cut the data into smaller sets to make checking easier. For example, check if the uploaded data matches the data in your original dataset.
For larger datasets that are nearly impossible to handle manually or if an SMS API connection is used constantly, we offer a tool for our customers that can help - The Number Lookup API. It’s a simple tool that pings mobile numbers without sending an SMS.
The ping checks if:
1) the number is valid
2) if it is roaming
3) which operator it belongs to
4) and if it has been ported from another operator
Looking up mobile numbers through the Number Lookup API is usually much less expensive per request than sending an actual SMS and hoping to get a delivery report (in most countries). This means that you can eliminate the erroneous numbers from the database at a fraction of the cost of sending those messages.
Say you clear 100 numbers from the 1000 number dataset with the Number Lookup API at a cost of 0.01 EUR (in most cases an HLR request to check the number is one-tenth of the cost of an SMS). Sending an SMS to each of those numbers and having it undelivered would cost 1 EUR, as the cost of an SMS is most often 0.01 cents. This means that the difference between using the Number Lookup API and just sending an SMS is 99 cents in this case or 99 messages worth of credit. And this escalates by orders of magnitude once the traffic increases.
Timely maintenance is key
It’s important to keep the database up to date always when a campaign is sent out. If your database is constantly evolving and the sending process is daily, running through the dataset at regular intervals is key.
There is no absolute way of saying what the best specific interval is for any company as use cases vary from the size of the dataset and frequency of use to re-occurring numbers in the database.
There are some key indicators to look for though:
- When was the database last checked and how many numbers have been added since that point, if any?
If the dataset has remained the same and is used on a monthly basis, then cleaning it on a monthly basis will do as well.
- How was the dataset gathered and how important is the content that the recipient gets?
As mentioned before the difference between receiving marketing messages and PIN codes.
- Is the dataset for frequent use or infrequent use?
If the dataset is used to deliver transactional confirmations daily, then the validity is much higher than for a marketing database used 2-3 times a year. In addition, the recipient is expecting the messages and is less likely to have themselves blacklisted from the list.
- How big is the dataset?
If the dataset is large enough, say more than 100000 numbers, there will be inevitable issues in it. People change operators, return their numbers, changes numbers, etc.
Following these steps will help avoid going over budget and save a fair bit of money down the line. Timely maintenance of your mobile number database will give your communication efforts higher cost efficiency as well as more bang for your buck.