Skip to main content

Why tiramisu is like data integration

Derek Munro 4 minute read Data quality

The hours of planning and preparation had been worth it, we had all enjoyed a delicious meal. I joined my wife in the kitchen to fetch the dessert, home-made tiramisu.

“I don’t think I prepared it early enough, the coffee doesn’t seem to have dissolved yet. The recipe asked for six tablespoons of strong coffee”, my wife said.

It was delicious, but gritty. She had used six tablespoons of ground coffee beans (coffee), instead of making some strong coffee and using six spoons of the resulting liquid (also coffee).

From my wife’s point of view the recipe was clear. However English is not her first language, and in her family the only time “coffee” and “tablespoons” are referred to in the same sentence is when someone is filling up the espresso machine (with ground coffee).

The instructions were as clear to me as they were to my wife, but we had understood different things. We had different points of view and spoke different languages, so the recipe wording was indeed ambiguous.

It dawned on me, long afterwards of course because I was under too much pressure to think in such a sensible way at the time, that this is exactly what happens when people with various experiences, skill sets, backgrounds and responsibilities work on data integration projects.


  • Investigate to choose a suitable desert
  • Find and print off a concise recipe
  • Prepare  – choose & assemble the correct ingredients – work hard and follow instructions  carefully – taste samples occasionally
  • Final tasting finds a big problem just before serving

Data Integration

  • Analysis
  • Design specification
  • Build – select the technical architecture, a skilled team and tools – good team  members work conscientiously – test samples occasionally
  • Final testing finds a big problem just before the deadline


If only the recipe had come with a sample of the food so that my wife could have compared her tiramisu against a “perfect result” early enough to recognise and correct any mistakes.

Better still, if only the people writing the recipe could have done so with my wife and me to validate that we all understood the instructions in the same way and were able to arrive at the same perfect result through preparing tiramisu prototypes.

Experian Pandora can’t help you in the kitchen, but it can help you collaborate to create complete, correct and unambiguous data Integration specifications through building a prototype. It can also provide you with a “perfect result” to compare with and validate the results of your build phase.

So, back to the comparison. I had already set expectations with our dinner guests by announcing what we were about to serve. The pressure was on, now what?


  • Bin the first attempt and make another desert
  • Scrape off the coffee grounds
  • Eat it anyway, at least it tastes good!
  • Blame the chef
  • Serve everyone a knife and an apple
  • Pour salt over it and blame a small child
  • Send everyone home

Data Integration

  • Start again, ask for more time and budget
  • Reduce the scope, making it unrecognisable
  • Force everyone to accept the compromise
  • Blame the Systems Integrator build team
  • Think up a manual solution
  • Blame someone who left the project earlier
  • Abandon the project

Data Integration

Of course we chose to do what most data integration projects do, we served what we had made and through the crunching of teeth we aggressively claimed it was exactly how true tiramisu should be.

Experian Pandora data management software enables organisations to understand, transform and manage the quality of their data more easily and quickly than any other product or manual approach. Experian Pandora is used on projects such as data integration, compliance, DWH, fraud detection and data governance.