Companies that produce or sell perishable goods often rely on estimating day-to-day or weekly sales in order to manage inventory. Accurate estimation of consumer demand
is essential in order to meet demand while keeping the number of products that expire
unsold to a minimum. Our daily estimates of sales service was initially designed
to help a local company maximize revenue and minimize the costs associated with
disposing of their unsold, expired products. The potential to help other local and
nonlocal companies facing similar challenges led to generalization of the service,
and to the creation of our web site.
Each day, we refine a heavily cross-validated, predictive model using recent sales
data and the latest weather forecasts. This model is used to produce a report that
estimates the number of sales that will be made each day over the upcoming seven days. Each night, we compare our estimates to actual sales figures in order
to fine-tune the predictive model and to allow for easy assessment of the model's
(and our own) performance. Estimates are recalculated each evening using the most
recent
regional weather forecasts [
Why?]
so that the next morning's report contains updated estimates that account for both
the previous day's actual weather and the latest weather forecasts.
We are not accepting new clients for this service at this time. If you wish
to be put on the waiting list, please
contact us to discuss
your company's individual needs.
Our pricing is determined by (1) how many models are required to produce the reports
you request, (2) how frequently these models have to be analyzed and updated, and
(3) the number of reports you would like sent out each month. How many models
we are required to build is dependent on the following factors:
- Each different Zipcode generally requires us to develop a model for that weather
region.
- Some customers have requested that sales predictions be broken down by product
category. If the models for different categories vary, we must develop and maintain
model for each category.
There is a minimum two-month subscription period, after which you may cancel at
any time. We automatically collect weather forecast data and fine-tune each model
given the latest available weather forecast and your most recent sales data.
Depending on the method by which you send us your recent sales numbers, we are usually
able to send out prediction reports for the upcoming week by 11pm Pacific Standard
Time each night. We strongly value privacy. Therefore, the names of our clients
and any sales data provided is treated as confidential information and will not
be disclosed to third parties.
If you would like a quote or have additional questions, please
contact us.
Go back to the
Services Overview page.