A new, data-savvy chief constable took over the Cleveland Police force in 2014, raising awareness of the impact of good data and its importance to different areas of the force. Maria Hopper, data protection manager at Cleveland Police, explained to DataIQ that the force was seeing its funding from the Home Office falling rapidly and that poor data quality had even become a strategic risk.
"We had 600,000 residents and 1.8 million records. Something was wrong."
She and her colleagues knew from Census data that Cleveland had approximately 600,000 people, but the police service had 1.8 million police records on its system. The records are in a central repository with all the force’s crime data, categorised by individual victims, witnesses and perpetrators.
With three times more records than the size of the population it was clear that there were many duplicates. “We know that not everybody who lives in our policing area contact the police. Some never need our assistance, so that was a clear indication that there was something quite wrong,” said Hopper.
A data matching tool had been brought in many years prior, but t was not prperly utilised until the new chief constable came in. However, it was only able to match 16,000 of the 1.8 million records. The service decided to switch its focus to the front end. “It was about improving the data that comes - that was the crux of it, not just preventing duplicate records from being created. Where there is an existing record, we identified it as quickly as possible and did not create a duplicate,” she explained.
Cleveland Police worked with Experian to enable its call handlers, equipped with a call management system, to cleanse the data with the golden nominal’programme, which sits on the Experian data quality platform as a data set with 1.8 million records.
Existing records were identified from five dimensions, such as name, DOB, address and phone number
The telephone number was the first piece of information to be identified within the call management system. The call handlers would then go to the golden nominal and pull out all the records linked to a telephone number and select the individual from a drop-down menu. “That would be the first search,” said Hopper. “If nobody is presented, the next search is the inputting of the name.”
What they considered as a pre-existing record was based on five entities: first name, surname, date of birth, address and telephone number. A variation of those entities at different percentages would identify whether a pre-existing record actually existed or not.
To merge a record, the call taker would put in three of those five items of data and pull that record back. “If the person is in the golden nominal with those tools, they will be able to identify with that level of data. If nobody is produced, they create a new record.”
Call handlers have gone from 1.5 hours to just 15 mins on data quality
Bobbies on the beat need not worry about being the source of data or correcting anything erroneous as the interface is only in the control room. Hopper stated that only 10% of all crime data in the central repository comes from out in the field.
This is the second year that the golden nominal has been in place and the results have been impressive. A year before, Cleveland Police call handlers were creating 3,500 new records a week. Last year, that number plummeted to 1,300, though it has risen slightly to 1,400 this year.
Hopper questioned one team about the implications of this new system. They said that one-and-a-half hours a day were spent dealing with duplicate data and incomplete records. This has now gone down to just 15 minutes. “That’s a massive resource saving,” said Hopper.