I'd be looking as the cell site data for Sarah's phone. I may have been switched off as she was abducted or the SIM card removed, but it's rough last known position using triangulation could have been identifed. Phones with similar triangulation strengths at the same time could be be identified and then manual checks done to identify subscribers - trace, interview, eliminate to find the possible suspects or do more unobtrusive enquiries first to find key suspects.
If Sarah's phone was not switched off immediately then it could possibly be matched to vehicles on CCTV and ANPR travelling a similar route.
Painstaking work but that's what serious crime investigators are well versed in.
Fortunately the eye-witness and bus CCTV prevented having to do all that legwork.
This is exactly what was done within 48 hours of Sarah being reported as missing - it was reported on 5th March that her phone last pinged the mast on Clarence Avenue and nothing further was heard from it so there was no way to trace the movements after that point.
Even with triangulation, it is a built up area, thousands of phones in peoples houses, on buses etc all would have had a similar triangulation, it is not as accurate as the TV shows lead us to believe, maybe up to 1km radius, and many of those phones would be untraceable anyway. And with 4 different networks, there was a 75% chance the perpetrators phone was connected to a different mast so wouldn't even been comparable triangulation wise.
After the event triangulation isn't as accurate as you may think. A phone doesn't constantly talk to cell phone masts as this would consume a lot of battery power and congest the limited radio bandwidth available. Once it has registered to the nearest mast it simply listens without transmitting anything unless a call or data transfer is taking place. The phone will check periodically by listening to see if the mast it is connected to is still the strongest, if it is, it does nothing, no transmission or 'ping' is required and no records of that event are known by the network operator.
When the received signal from the mast becomes weak or another mast becomes stronger, it will make a decision to request a connection to a different mast which will involve making a roam request transmission which will be picked up by one or more masts. The network provider then tells the phone which mast to connect to. Only at that point is it possible for a triangulation to be approximated based on the strength of the 'ping' signal received by the masts that hear that request. The phone then goes back to listening mode and will not transmit anything to other masts even when moving as long as the connected mast remains the strongest.
Even if the phone is engaged in a call or data transfer, the other masts are not aware of its presence as it will be using a frequency that only the mast it is connected to can hear. It can only be triangulated when the phone makes a specific request to roam between masts on a specific frequency heard by all the masts in the area - that can be very infrequent.
That occasional ping can be totally skewed by buildings or even a passing bus that reduces the signal dramatically in one direction leading to a totally inaccurate triangulation. Phones in cars are particularly difficult to triangulate as well because of the metal box they are contained in hampering the signal strength in some directions but not others.
You can't even determine the exact time a phone was switched off either, you can only say that it last pinged at X time, it could be switched off 10 minutes later without any of the masts knowing about that even because it doesn't sign off.
Live triangulation of a phone is a lot more accurate than after the event, it can be forced to make lots of pings by the network operator which would provide increasingly accurate data - but that is generally the stuff of TV shows and secret services.
Sorry for the long post on this, I just think the expectations of most people in triangulation of phones is much higher than reality due to TV shows.