The process

Two flows: getting the bordereau in, and getting the data right

Ingestion moves a submission from the inbox to the correct folder and the market system. Cleansing turns whatever the coverholder sent into a return that meets the standard. Both are the same every month, and both leave a record.

BDX ingestion process

Receipt to upload, with the contract identified before anyone opens the file.

  1. 1

    Receive

    A submission lands in the monitored mailbox or a watched repository. The message, its metadata and every spreadsheet attachment are recorded against the sender.

  2. 2

    Ingest

    Attachments are opened, every sheet is detected and previewed, and compatible sheets or files are merged into one working dataset.

  3. 3

    Capture

    UMR, coverholder, year of account, reporting period and identifiers are extracted with a confidence score and matched to the contract register.

  4. 4

    Name

    The file is renamed to the contract convention: UMR, coverholder, year of account, bordereau type, period and identifier, honouring monthly or quarterly frequency.

  5. 5

    Save

    The named file is written to the SharePoint or OneDrive path resolved from the contract, alongside the original as received.

  6. 6

    Upload

    The cleansed return is released for upload to the market system, with the VIPR reference tracked against the submission.

BDX cleansing process

Nothing destructive happens without being opted into, and the original upload is never touched.

  1. 1

    Ingest and prepare

    Headers are located, blank and total rows removed, and multiple years of account split into their own tabs.

  2. 2

    Standardise

    Columns map to the Lloyd's template for the class of business. Dates, currencies, transaction types and premium types are normalised, with the original value kept in an adjacent column wherever a value was changed.

  3. 3

    Validate

    Mandatory fields, sign conventions, period coverage against risk inception and identifier consistency with the binder are all checked. Contentious rows are written to a validation log rather than guessed.

  4. 4

    Enrich and match

    Premium chain figures are derived where they can be calculated, commission percentages filled from amounts, addresses scored and geocoded, and names screened against sanctions lists.

  5. 5

    Report

    A cleanse report, a mapping and data quality report and a cell-level activity log are produced so every transformation can be evidenced.

  6. 6

    Ready

    Contract columns are appended, the identifier split is applied, and the output is delivered by download or email once you confirm.

Cleansing is optional, stage by stage

Each stage can be enabled, automated or left to a reviewer, per portal and per role group. Some books want everything automated; others want a human to confirm the mapping before a single value changes.