Bitcoin can be good for remittances but if it becomes popular in a region it will out price itself for that and other reasons.
Also almost all bitcoin services I've seen so far have saved money two ways. 1. They have offloaded the last mile cost on another service(a banks ATM network, or corner store, etc). If it becomes popular those services are going to start wanting their cut much they same as WU would pay them. 2. They ignore KYC. Again if they get popular the banking industry will definitely apply pressure to get them to start following this which will again increase costs.
A quick apples to apples comparison suggests that converting sterling to Kenyan Shilling denominated M-PESA mobile money costs around 4-5% including forex and fees via Western Union and Moneygram (unlikely to be the cheapest of many providers in the marketplace) whilst sending to M-PESA via the BBC-advertised remittance services costs a 3% forex fee plus cost of obtaining Bitcoin (surprisingly expensive if your salary is paid in GBP; I'm struggling to see any options below 1%, and the cheaper ones aren't instant either).
There might be an advantage for the Bitcoin remittance services, but perhaps not one worth the additional inconvenience, delay and counterparty risk. Downward pressure on the major remittance firms' prices is always welcome, but given the relative lack of demand to convert in the other direction from KSH to BTC, the cost bases for these cryptocurrency enabled entrants into the market might actually be higher than the major remittance firms, especially if they grow...
But your link is to compete against scenarios where "existing mechanisms" cost 1.2%, and bitcoin came close to that. While BBC is talking about scenarios that cost an average of 12.3%
In addition, the linked case is tipping the field slightly by requiring that the money start in a bank and end in a bank, while the BBC article is discussing underbanked people.