Client only has 18 months of billing data - missing several months in the middle due to their poor record keeping. How do you handle analysis when you don't have complete 24-month dataset? Do you proceed with what you have or insist on getting complete records first?
Handling incomplete data sets
Glen, depends on what months are missing. If you're missing peak summer or winter months, that's a bigger problem than missing shoulder months. I usually try to get at least one complete seasonal cycle before proceeding. Utilities often have better records than the customer - worth asking them directly.
I've had good luck contacting the utility directly for historical bills. Most will provide copies for reasonable fees. Cheaper than doing a partial analysis and missing potential savings. Document the data gaps in your proposal though - sets expectations appropriately.
If you must proceed with incomplete data, be very clear about limitations in your report. I've done analyses with 15-18 months and found significant savings, but always include disclaimers about potential additional opportunities with complete dataset. Under-promise, over-deliver.
Greg's right about documentation. I also adjust my fee structure for incomplete datasets - either lower upfront fee or success-based pricing. Protects both parties when you're working with limited information. Better to be conservative than overpromise based on partial data.