obtain value from unknown string key property, using jq

I have data structured like this:

{
  "data": {
    "2017-06-20": {
      "shifts": [
        {
          "id": 24,
          "shift_request_id": 24,
          "created_at": "2017-06-27 15:10:50",
          "updated_at": "2017-06-27 15:10:50"
        },
        {
          "id": 38,
          "shift_request_id": 38,
          "created_at": "2017-06-27 15:10:50",
          "updated_at": "2017-06-27 15:10:50"
        },
        {
          "id": 85,
          "shift_request_id": 85,
          "created_at": "2017-06-27 15:10:51",
          "updated_at": "2017-06-27 15:10:51"
        }
      ]
    },
    ...
    ...
    ...
    ...
    "2017-06-21": {
      "shifts": [
        {
          "id": 26,
          "shift_request_id": 26,
          "created_at": "2017-06-27 15:10:50",
          "updated_at": "2017-06-27 15:10:50"
        },
        {
          "id": 28,
          "shift_request_id": 28,
          "created_at": "2017-06-27 15:10:50",
          "updated_at": "2017-06-27 15:10:50"
        },
        {
          "id": 88,
          "shift_request_id": 88,
          "created_at": "2017-06-27 15:10:51",
          "updated_at": "2017-06-27 15:10:51"
        }
      ]
    }
  }
}

So, with those date-like formatted keys, let's say I want to extract shifts from say, first such a key. I am not interested in extracting by particular date as that would be easy as specifying that date as a key, but rather selecting from the whole set, one of them or range, by numeric index, once obtained those keys.

I tried:

jq '.[] | keys'

which gives me an array of all keys, but then I don't know how would I extract array of shifts for a particular key indexed by number.

jq '.[] | keys[3]'

will give me the 4th key in a row, but then .shifts won't work on that, saying that it Cannot index string with string "shifts"