I would use
WHERE columnName LIKE '%[%]%'
SQL Server stores string summary statistics for use in estimating the number of rows that will match a LIKE clause. The cardinality estimates can be better and lead to a more appropriate plan when the square bracket syntax is used.
The response to this Connect Item states
We do not have support for precise cardinality estimation in the presence of user defined escape characters. So we probably get a poor estimate and a poor plan. We'll consider addressing this issue in a future release.
An example
CREATE TABLE T
(
X VARCHAR(50),
Y CHAR(2000) NULL
)
CREATE NONCLUSTERED INDEX IX ON T(X)
INSERT INTO T (X)
SELECT TOP (5) '10% off'
FROM master..spt_values
UNION ALL
SELECT TOP (100000) 'blah'
FROM master..spt_values v1, master..spt_values v2
SET STATISTICS IO ON;
SELECT *
FROM T
WHERE X LIKE '%[%]%'
SELECT *
FROM T
WHERE X LIKE '%\%%' ESCAPE '\'
Shows 457 logical reads for the first query and 33,335 for the second.