A lot depends on whether you are talking about player vs AI strategy or player vs player run empire, or versus AI but with RP considerations.
The AI isn't capable of deliberate deceptions, or restraint for the purpose of luring someone out of position to attack. The AI also isn't really capable of being bluffed out when its scouts are shut down.
I generally go with a spectrum for my scouts, from 1.6 HS minimalists which are fast, expendable to 6 HS fighters with 1 HS sized sensors and a lot of speed, to 10 HS fighters with a 2.5-3 HS sensor suite. 1000 ton fast sensor boats are pretty easy to customize, as retooling such small shipyards is very fast.
The idea is that with the different sizes, I am more likely to find a window that allows me to track the enemy from beyond the range that they can target my scout. Maybe they can detect it, but they can't catch it or shoot it with missiles without closing.
But using small scouts for sensor security, you need to be much more proactive with them, sending them billions of km ahead of the fleet in some cases. As far as 'rules of thumb', I could go into a LOT on my scouting philosophy.
No one ship can do all scouting missions.
Basic scout roles:
Fleet scout Fast, endurance between 2 weeks and 6 months, sensor payload between 15% and 25% HS. Speed keeps it alive, and everything is stripped down to reduce its sensor profile. A scout pared down to the absolute minimum size may be able to track without being tracked, but might not be able to stay on station long enough to do its mission.
Rock checker, LONG endurance, perhaps with a single ship jump engine, has a res 1 active sensor capable of detecting the smallest enemy presence at 10,000 km, but not really great beyond that. 60-100 billion km range. Emphasis is on economical exploration. Survivability of the ship is minimal, but because it operates with other ships, survivability of the DATA is high. Rock checkers are MUCH cheaper than survey ships, and go in ahead of the survey fleet, looking for enemy presence.
Picket, long crew and maintenance endurance, but perhaps with a very short range. Role is to monitor a jump point or a planet of interest, and shadow contacts. A survey support carrier can drop it off at a jump point, and pick it up several years later.
Fleet security, MASSIVE sensor. The biggest you can research and install. Unfortunately, sensor tech advances render it obsolescent, and large sensors are both expensive to research and prohibitive to refit. Speed isn't as much of an issue as it is in the heart of the fleet. Unsubtle as hell when a 50 HS res 100 sensor PINGS the system. I found it useful to build PDCs with active sensors back in VB6, because there wasn't an issue of tooling required. Sensor advancements would just make the sensor smaller, which isn't that much of a consideration for a PDC, it would still cost about the same.
Scouts generally are not armed, but they could be paired with variants that strip out the sensor and perhaps some range for a weapon system, and then would operate with the scouts to seek out and destroy enemy scouts. They wouldn't need much armament to be effective, just a beam weapon and enough speed and initiative to close.
So start your design process from the mission. Figure out the speed you need, the sensor range vs cross-section, and then figure out how much endurance you can have and still do the mission reliably.