This got me interested in the site-selection process, so I reread Chapter 10 of the NASA history LJK linked. It's really quite a story. The best data they had before the Viking Orbiters arrived was Mariner 9 imagery with a resolution of about 100 meters, and people like Carl Sagan rightfully fretted about the roughness of the landscape at a scale of about 10 centimeters--the scale of the lander footpads and the lander legs footprint. Their plan was to quickly use new imagery from the Viking Orbiters to refine the earlier site selections, and indeed what they found was shocking--areas that were smooth in Mariner 9 images showed up as chaotic and broken terrain at the Viking scale. They had to work through their prioritized list of sites, command the Orbiters to take pictures of each site, and when the images came in a few days later, frantically examine them to see if the next site would be acceptable. And all the while the missions were already at Mars and the clock was ticking.

Nothing like today. We've got it easy compared to those guys at JPL in the 70s, with the 1M MGS-MOC images and soon centimeter-scale imagery. Yet still we lose lander missions. Upon reflection, the follow-on missions in 1979 probably had a high-probability of failure due to insufficient data about the surface. Even if the landers made it to the ground safely, the data wasn't good enough to predict whether the rovers could get around and whether they'd find anything interesting. Perhaps we had to wait into the 1990s to get good high-resolution orbital imagery, before mounting a comprehensive campaign of surface exploration.