SecProbe.io

CORRESP
1
filename1.htm

March 11, 2008

By EDGAR
and by Facsimile to (202) 772-9202

Securities and Exchange Commission

100 F Street

Washington, D.C.  20549

Attention:      Millwood Hobbs

                                                                        Staff
Accountant

                                                                        Division of
Corporation Finance

Re:                               Overstock.com, Inc.

SEC Comment Letter dated November 15,
2007

Form 10-K for the
fiscal year ended December 31, 2006

Forms 10-Q for the fiscal
quarters ended September 30, 2007, June 30, 2007 and March 31,
2007

                                                                                                File No. 0-49799

Dear Mr. Hobbs:

We write in response to your letter dated March 7,
2008 regarding the above-referenced filings.
The following responses are numbered to correspond to the Staff’s
comments.

SEC Comment — 1.  Where a comment below
requests additional disclosure to be included, please show us in your
supplemental response what the revised disclosures will look like.  These additional disclosures should be
included in your future filings.

See responses below.  We will provide these additional disclosures
in future filings.

SEC Comment — 2. Please tell us
and revise your disclosure to explain what consideration was given to the
impact of the correction of an error relating to your revenue recognition
accounting policy with respect to your conclusion on the effectiveness of
disclosure controls and procedures over financial reporting.

As part of our evaluation of the
effectiveness of internal control over financial reporting, we analyzed the
impact of the error relating to our revenue cut-off and the application of GAAP
to our financial statements. We determined that the control deficiency
underlying the error in our revenue recognition model represented a reasonably
possible likelihood that we did not achieve one or more of our financial
statement assertions. We determined that the magnitude of the relating
reasonable possible misstatements was not material to either the interim or
annual financial statements. We then determined that the deficiency was
important enough to merit the attention of those responsible for oversight of
our financial reporting; however, we concluded that a well-informed, competent
and objective individual would not conclude that the deficiency is a material
weakness.

Accordingly, we conclude that the deficiency
constitutes a significant deficiency as defined by AS5.

As the deficiency did not have a material
impact on the financial statements or related disclosures, we would also
conclude that our disclosure controls and procedures over financial reporting
were effective as of December 31, 2007. As we have concluded that the
deficiency did not cause a material error to our financial statements, we do
not believe it would be appropriate to specifically discuss this deficiency
within our disclosure regarding the effectiveness of disclosure controls and
procedures over financial reporting.

SEC Comment — 3. We have read
your response and proposed revised disclosure to comment 3 of our letter dated January 4,
2008 regarding your revenue recognition accounting policy. Your disclosure on page F-15
states you adopted SAB 108 for the fiscal year ended December 31, 2006 as
required. Accordingly, the cumulative effect provision available in fiscal 2006
at the time of our adoption of SAB 108 is a one-time option and is not
available in connection with your reporting in fiscal 2007 of the correction of
the error in your revenue recognition policy. Please revise your proposed
disclosure to clarify your reporting of the error correction in fiscal 2007.

In response to your comment, we will modify
the disclosure in our Form 10-K for the fiscal year ended December 31,
2007 as follows:

“From our
inception through the third quarter of 2007, we recorded revenue based on
product ship date. In the fourth quarter of 2007, we determined that we should
not have recorded revenue until the delivery date. We performed a detailed
analysis of this error and the impact of recording the cumulative effect of the
error in the fourth quarter of 2007, and have determined that the impact of the
correction is immaterial to the full year and fourth quarter of 2007 and to all
prior periods.

As a result, we recorded the
cumulative effect of this correction in the fourth quarter of 2007. This change
resulted in a deferral of $7.8 million of revenue (including $2.1 million of
direct revenue and $5.7 million of fulfillment partner revenue), and a decrease
in cost of goods sold of $6.6 million ($1.8 million direct and $4.8 million
fulfillment partner), which reduced gross profit and net income by $1.2 million
(see Item 15 of Part IV, “Financial Statements” —Note 2 — “Summary of
Significant Accounting Policies” — “Revenue Recognition”).  Prospectively, we will continue to recognize
revenue based on delivery date.”

SEC Comment — 4. We note your
response to comment 3 of our letter dated January 4, 2008 regarding the
revised disclosure included on Exhibit E relating to revenue recognition.
You state in the fourth paragraph that you estimate delivery date based on
transit time information provided by your shipping carriers, and that the
delivery date is typically one to six business days from the date of shipment.
It is not

2

clear from your response why you
believe it is appropriate for you to rely on estimated transit time information
provided by shipping carriers. Accordingly, please explain what method and
approach you used to determine that transit time information provided by
shipping carriers is sufficiently accurate to estimate delivery dates which you
use as the basis for revenue cut-off and the amount of revenue that is
recognized in each period. Tell us what factors and assumptions support your
conclusion that the delivery date is typically one to six business days from
the date of shipment. Also, please specifically address your consideration of
shipping and delivery exceptions as reflected on your website.

As a retailer of general merchandise, we
operate in a high volume/low margin business involving retail sales made to
individual customers. Because our systems are set up to recognize revenue upon
shipment (rather than on delivery), in order to recognize revenue based on
delivery date, we must make a manual adjustment to defer the amount of revenue
and cost of goods sold associated with shipments in transit at the end of each
period. Because we generally ship daily approximately twenty thousand
individual orders with an average order value of $100-$120 from our own
distribution center in Salt Lake City, as well as from our 750 fulfillment
partners’ distribution centers located across the country, it is impractical
for us to track the delivery date of each individual order in a cost-efficient
manner.  As a result, to record revenue
based on delivery date, we must use an estimate.

Therefore, we have developed a methodology
and approach for estimating the delivery date of orders shipped.  Our methodology used both our shipping
carriers’ published transit time tables by the delivery destination state/zone
and actual shipment transit time experience (taken from a sample of total
shipments that occurred during the last ten days of 2007) to calculate an adjusted transit time by carrier and destination. We applied
the adjusted transit times to shipments over the last several days of 2007 and
of prior periods to determine the shipments that should have been deferred at
the end of each period. A more comprehensive description of our methodology is
included below.

As was demonstrated in the materiality
analysis we provided in Part II of our response to comment 3 of your
letter dated January 4, 2008, the adjustment to correct the error in our
revenue recognition was not material to 2007.
In addition, in periods previous to 2007, when including the effect of
the reversal of the prior quarter deferral into the current quarter, the
difference was not material.  We conclude
in our sensitivity analysis (included later in the response to this comment
below) that  changes of 50% to 67% to our
estimated deferral days would not have a significant effect on the materiality
of the revenue deferral to 2007 and prior periods.

A Comprehensive Description of
our Methodology for Determining our Estimate

The following is a comprehensive explanation
of our method and approach for estimating the amount of revenue to defer in
each period. We include the factors and assumptions that support our conclusion
that the delivery date is typically one
to six business days

3

from the date of shipment, and how shipping
and delivery exceptions are factored into our analysis.

To do this, we will describe:

1)              The underlying
data used in the analysis

2)              The
determination of our sample and testing procedures

3)              The development
of our Adjusted Transit Time Schedule

4)              The resulting
calculation of the revenue deferral estimate

5)              Our assertion
that delivery date is typically one to six business days from shipment

6)              The sensitivity
analysis

1.  Underlying Data.

The underlying data used to determine
estimated delivery date included sales orders shipped during the full month
preceding the December 31, 2007 cut-off date. This data was aggregated by
delivery destination and shipping carrier type for each ship date.

Our orders are shipped by two main carrier
types: (1) Medium Sized Orders (up to 150 lbs) which are shipped primarily
via United Parcel Service (“UPS”) and the United States Postal Service (“USPS”);
and (2) Less than Truck Load (“LTL”) orders (over 150 lbs.), which are
shipped primarily through ABF Trucking Service. Therefore, we differentiate our
analysis based on the carrier type.  Note
that approximately 95% of our revenue for the full year 2007 was from Medium
Sized Orders, and 5% from LTL orders.

2.  Determining our Sample and Testing Procedure.

To determine the most accurate transit time
assumptions to be applied to our shipments, we initially obtained the standard
transit time schedules for our primary carriers. We then took a
non-statistical, risk-based approach to selecting a large sample of shipments
(i.e., 300 shipments) to test the published transit time assumptions provided
by the carriers for reasonableness. Although ours is not a statistical sample,
we based our sample on the recommended size for a statistical sample to reach a
5% error rate and 95% confidence level for the population of shipments tested.
The sample was taken from the shipments that occurred during the last ten days
of 2007, as this was the population of shipments from which the deferral would
be calculated. The selection was judgmentally weighted towards fulfillment
partner shipments for which we have less visibility of shipping origin on a
real-time basis and which account for approximately 70% of our sales. We also
placed more emphasis on making selections from days where there would be more
question of deliverability by the December 31, 2007 cut-off date (days 3-6
days from the end of the period).

3.  The Development of our Adjusted Transit Time
Schedule.

The primary factor we used in determining our
revenue deferral estimate was our adjusted transit time schedule by state and
by carrier (“Adjusted Transit Time

4

Schedule”).
The Adjusted Transit Time Schedule was initially based on our shipping
carriers’ published transit time tables by the delivery destination state/zone
for LTL and UPS.  We then adjusted these
transit times based on the actual transit
times we obtained from our sample of 300 shipments.

We took our sample of 300 shipments from the
total population of shipments that occurred during the last 10 days of the
period (i.e., December 22-31, 2007), and from all shipping carrier types.
Using tracking numbers assigned to these shipments, we obtained delivery
confirmations from the respective carriers’ to determine the actual delivery
date of each shipment sampled.  This
enabled us to calculate the actual transit time for the shipments in our
sample.  We then compared the actual
transit time to the published transit time tables provided by our shipping
carriers and calculated the difference between the actual transit times and the
carrier published transit times.

We then summed up the differences between
actual and published transit times and calculated an overall average
difference, grouped by type of shipping carrier. We then adjusted the published
transit times by the overall average difference to more accurately reflect the
actual delivery times in order to create our Adjusted Transit Time Schedule, by
destination state and by carrier (for example, if the carriers’ published
transit time to California from Utah was two days, and the overall average
transit time determined by our sample was an additional one day, we adjusted
the transit time from Utah to California to three days).

Note that for 12 of the 300 shipments
sampled, we were not able to calculate the actual transit time due to the
tracking information being unavailable or invalid, and therefore the actual
delivery dates were not available.  Due
to the large size and non-statistical nature of our sample, we did not replace
these with additional selections. However, the lack or invalidity of tracking
information did not necessarily indicate transit times that would differ from
the remainder of the sample.

4.  The Resulting Calculation of the Revenue
Deferral Estimate.

We then applied our Adjusted Transit Time
Schedule to all individual sales orders shipped during the ten days preceding
our reporting cut-off period to determine an updated estimated transit time and
delivery date for each state and shipping carrier for these sales orders.

Using this data, we determined by day, the
orders shipped but not delivered by the end of the reporting cut-off
period.  The calculation does not result
in a 100% deferral of the last several days’ shipped sales, as it defers sales
in relation to the actual shipping destination and the average transit time to
each destination according to our Adjusted Transit Time Schedule. The total
revenue deferral estimate by day at December 31, 2007 is shown in the
table below:

5

5. Delivery Date is Typically One to Six Business Days from Shipment.

To support our disclosure that “the delivery
date is typically one to six business days from the date of shipment,” we took
our sample shipments and calculated the actual days in transit for each
shipment.  The results below show that
approximately 90% of shipments were delivered between one to six business days:

6. Sensitivity Analysis.

Below is a sensitivity analysis that shows
what the result would be if our estimated number of days in transit were off by
up to plus or minus two days (note that our base average deferral is between
3-4 days). Based on this analysis, we conclude that as much as a 50%-67% error
in our overall transit time estimate (up to two days different than our base
average deferral) would not result in a material adjustment to our financial statements.
The highlighted row in our analysis below comes directly from our materiality
analysis provided in Part II of our response to comment 3 of your letter
dated January 4, 2008.

In the future, the reversing revenue deferral
from one period will offset the revenue deferral for the following period.
Accordingly, the net effect of this adjustment to revenue and net income will
be less significant in future periods than the cumulative percentages shown
above.

6

SEC Comment — 5. Please tell us
and revise your disclosure to clarify whether you estimate the delivery date
for each order and if the transit time information used to estimate the
delivery date is based on the type of product shipped in addition to the