I want to monitor the SERP of several important keywords for my website, and the 100 free queries from custom search API seemed to fit the bill. However, it requires a programmable search engine to use so I set one up and played around with it.
Based on the custom search API results, a test website ranked #6 for the target keyword. The same keyword / website came up at #9 when using the programmable search engine, and #2 when I searched the keyword on google.com. I tested some others and they were far worse, with a discrepancy of 20-30+ spots.
Why are these results so different? I would much rather use the provided tools and pay for it when necessary, but this is pushing me towards the path of scraping the data to get consistent results.
Serp uses a VPN to do the searches from another country so they can pass the cheap queries onto you.
So you’ll get different rankings
When it comes to SERPS, especially google, there is no “canonical SERP”. Google tries to be really smart to deliver the “right results to the right users”. So the results will vary depending on things like geolocation, user profile, IP, past searches and a lot of other factors. This can be just slight variations or sometimes really big ones.
I once worked at a company and checked our ranking from inside the office building. Always looked good. Until I check it once outside of the office building. Totally different SERP for the same search term. My guess: since people in the office all used google to search for our company website (as one does) google registered that from our IP block our website was the most clicked on result and started ranking it higher, but only for our IP block. 
Best approach imo: use an IP not related to you (home, office, etc, eg a cloud VM), and use a (VPN) geo location where you expect most of your traffic will come from. From there emulate a real browser do the SERP query.
You make a very valid point. While I had assumed this played some part, what I found weird was how drastically different the results were, even when all 3 searches originate from the same (my) IP address. PSE for example returned about 75% Reddit posts, whereas Google searches were more what you would expect and API was kind of a mix between the two.
I just turned on my VPN, then ran the same search in PSE and Google.com from an Incognito browser. To my surprise the results were much closer than before. Some were mixed, like #1 and #2 on Google were swapped on PSE, but otherwise nearly identical.
Yes it’s still Google even though it’s a VPN
