Will TBB speedtest give correct results on loadbalanced sys?

I'm trying out a load balanced/failover config on a Draytek 2860.

Two lines
A) BT VDSL with 7 Mbps down, 579 kbps upload (but unstable, generally poor quality line, more bad than good days)
B) BT 4G via a Netgear LTE Modem Model LB2120 (consistently more than 24 Mbps down)

B) is configured as
Failover, Load balance, Traffic threshold 128 k, down 1M
Load balance mode "According to line speed"
When users surf to websites (say video download, ie sufficiently high demand), Line B) becomes active.
In other words, on the slightest suspicion of real traffic, the faster line should take over.

Will TBB speedtest (or any speedtest) give me an accurate result, or can't it handle such configurations. I can imagine that it won't really work, but am I right?

Your help is very much appreciated.

Two sources are used for main download test, so may be split across the two or it may not.

Other testers that split between sources will behave similarly but if higher numbers of sources the chance of using both links will improve.

Thanks for these details. On balance I think that aggregated links like this can't really be speedtested by conventional means.

This router config is now IP based (target IP), I can't see how the other possible method (session based) will be any different.

Thanks again for some insight.

I have a 3220 draytek with 3 wan's and session based load balancing and all the usual speed tests combine the 3 wans.
As long as the speed tests are multi session.
Ive found the 3220 to be excellent at balancing out the connections.
The three wans are all mobile, 2 x unlimited and one fixed quota - one is on a external device which gets the best speed (max 60Mbps), the other two are on internal modems and their speed varies - ive had speed tests as high as 105Mbps at times, which is way more than the external can do on its own.