Ran the CSV through our deduper first. Website URLs were populated for ~80% of rows, far better than the competition. Genuinely useful.