As alternative data becomes more deeply embedded in investment research and quantitative models, the quality of a dataset is increasingly about more than its size. Coverage and granularity remain important, but so too does the ability to maintain a consistent, representative signal as underlying data sources and panels evolve.
That challenge is central to the latest expansion of investor data products from Facteus, which has added 15 million US card accounts to its consumer transaction data coverage, alongside new consumer cohort feeds and a significant expansion of its product-level Onyx UPC dataset.
The additional card programme takes Facteus’ coverage to more than 120 million US accounts, with more than eight years of historical data. The new data, delivered with a one-day lag, is designed to broaden geographic and economic representation within the company’s existing transaction datasets.
Facteus has also introduced cohort feeds designed to provide separate views of spending among higher- and lower-income consumers. For investors seeking to understand changes in consumer behaviour, such segmentation can potentially provide a more nuanced picture than aggregate spending data, particularly during periods when inflation, interest rates and other economic pressures affect different income groups in different ways.
Alongside the card expansion, Facteus is increasing the coverage of Onyx, its UPC-level consumer spending dataset. By the end of the third quarter, the company expects the dataset to cover more than 30,000 stores and eight million UPCs, representing over $265 billion in annual spending.
The increased granularity reflects a broader development within alternative data: the ability to move beyond measuring overall company or retailer performance towards analysing individual brands, categories and products. Facteus’ expanded dataset includes online and offline spending, discounts, tender type and brand, category and ticker-level metrics.
But increasing the breadth and depth of alternative datasets also creates another challenge for investors: maintaining continuity of the underlying signal.
Changes in upstream data sources or panel composition can create gaps or shifts in datasets that have already been incorporated into investment processes. For quantitative users in particular, that can affect historical comparisons and potentially undermine models developed and backtested against earlier versions of the data.
Facteus Chief Product Officer Lorn Davis describes this as a potential “model-integrity problem”, arguing that investors need confidence that changes to the underlying data supply will not unexpectedly alter the characteristics of the signal.
Facteus says it seeks to mitigate that risk by sourcing data through diversified direct relationships with banks, credit unions and fintech companies, while benchmarking its datasets against external economic indicators. The objective is not simply to increase the number of transactions captured, but to ensure that the resulting signal remains representative and usable over time.
That distinction is becoming increasingly important as alternative data moves further into mainstream investment workflows. The competitive battleground is no longer simply about who can offer the biggest or fastest dataset. For investors putting alternative data into production, the ability to demonstrate that a signal is representative, granular and sufficiently stable to support models over time may prove equally important.
Subscribe to our newsletter
