Stress is a factor to be considered; both humans and pets have some days which are more stressful than others. Diet can be another factor in variability, certainly for people and especially for pets who are not on one of the many diabetes prescription diet foods.
No matter how a person with diabetes may micromanage his or her blood glucose, he/she will never be successful at making today's readings exact "photocopies" of yesterday's. The variability factor in persons with diabetes can be anywhere from 15-50% from one day to the next. 
Intervet states some fine reasons for variability at the link below.  As you see, there are many different factors involved.  When there's too much variance in them, though--marked differences in highs and lows--it's a signal that things need to be re-evaluated.  More/less insulin, bad/weak insulin, changing insulins, etc.--all need to be considered if and when this occurs.
What needs to be realized is that insulins can be variable in two ways--in different patients (interpatient) and also in the individual patient (intrapatient).  Variability is not solely an "insulin problem"; it can occur with ANY drugs or treatments.
The study at the link below  shows how different insulins (in this case NPH, Lantus, and Levemir) can give different variabilities.
The 2002 German study linked below  re: variability of insulins in humans gives some idea of the sort of problem it is for all those with diabetes. (It is unclear whether Levemir was part of this study; Lantus was available at the time and therefore can be considered to be part of it.) All insulins, have some degree of variability from patient to patient and for the same patient. The variability problem is not solved by the analogs, as they have the same variability problems as all of their non-analog counterparts.
It was believed that by changing the way insulin was administered (from injection to inhalation) might elminate some of the variability factors; this proved to be false. Further complicating the variability matter when looking at human studies is the fact that there are no applicable studies of the variability of injected insulins in persons with diabetes.
What is known on variability is that R/Neutral non-analog insulin varies less in "healthy subjects" (those without diabetes), than Intermediate-acting (NPH and mixes, Lente) and Long-acting (Ultralente). The study does not give head-to-head data regarding the analog insulin variability, but it does indicate that the analog insulins--both rapid-acting and long-acting, have the same variability issues. There is not yet a method to significantly reduce variability. 
Absorption and the insulin depotEdit
The size of the Insulin Depot influences variability, with smaller unit doses varying less than larger ones.  Dr. Richard K. Bernstein,  father of the modern Glucometer,  also makes a valid case for this,  as does Dr. Hanas 
Intervet also addresses the issue when stating that the dose per injection on a twice-daily regimen is less (then the large once-daily injection), which results in less hypoglycemia and better glucose control. 
Variability and needle lengthEdit
Some caregivers report difference in insulin absorption with different length needles. BD Diabetes  explains that you should consult with your health care professional before using a short needle, and carefully monitor blood glucose when changing to a shorter needle.
People have found their blood glucose may not be well-controlled when switching to the shorter needles; this also has been the case with some dogs. Switching back to a longer needle solved the problem. You should consider experimenting with the different length needles as insulin absorption may vary. 
Part of diabetic lifeEdit
It was once thought that when regulated, dogs who were fed the same meals on the same food and insulin schedule had very little day-to-day variability of their blood glucose. Dr. Fleeman and Rand investigated, and found that dogs,  like humans, can vary quite a bit.
It's possible for some dogs to vary by 100 mg/dl (5.555 mmol/L) on consecutive days even though the same dose of insulin has been given. Don't just "live by the numbers" but consider other factors such as polydipsia, polyuria, etc.
- ↑ Schermerhorn, Thomas (2008). Strategies for monitoring diabetes mellitus in dogs. DVM 360.
- ↑ Caninsulin-Serial Blood Glucose Curves (Page 8). Intervet.
- ↑ Type 1 Diabetes Mellitus & Use of Flexible Insulin Regimens. American Family Physician (1999).
- ↑ Blood Glucose Averages. BD Diabetes.
- ↑ Insulin Treatment-Individual Factors. Patients Up To Date.
- ↑ Heise, Tim, et. al. (June 2004). Comparison of Levemir (insulin detemir), NPH/Isophane Insulin & Lantus (insulin glargine). Diabetes.
- ↑ Heinemann L. (2002). Variability of Insulin Absorption and Insulin Action. Diabetes technology and therapeutics.
- ↑ Heinemann, Lutz (January 2008). Variability of Insulin Action:Does It Matter?-page 40 (4 of 9). Insulin Journal.
- ↑ Ravis WR, Comerci C, Ganjam VK. (1986). Pharmacokinetics of Insulin Following Intravenous & Subcutaneous Administration in Dogs-Biopharmaceutics & Drug Disposition. Biopharmaceutics and drug disposal.
- ↑ Goeders LA, Esposito LA, Peterson ME. (1987). Absorption & Kinetics of Regular (Neutral) & NPH (Isophane) Insulin in the Normal Dog. Domestic Animal Endocrinology.
- ↑ Higgins, Thomas. Insulin. Boulder Medical Center.
- ↑ 12.0 12.1 Bernstein, Richard. The Laws of Small Numbers-Part 2. Diabetes in Control.
- ↑ Mendosa, David. David Mendosa's Page on Dr. Richard K. Bernstein's Accomplishments. Mendosa, David.
- ↑ . Hanas, Ragnar (1999). Insulin-Dependent Diabetes (Pages 10 & 11). Children With Diabetes.
- ↑ Caninsulin-Starting Insulin Therapy in the Healthy Diabetic Patient (Page 4). Intervet.
- ↑ Using Short Needles. BDdiabetes.com.
- ↑ Position Statement. Diabetes Care-ADA (2004).
- ↑ Fleeman, Linda, Rand, Jacqueline (2003). Evaluation of Day-to-Day Variability of Blood Glucose Curves in Diabetic Dogs. JAVMA.
- ↑ Nichols, Rhett (2008). Diabetes Mellitus In the Dog and Cat With Special Emphasis on Feline Diabetes. Michigan State Veterinary Association.