Template-based analysis: when to use pre-built vs. custom

Every BI tool now ships with templates: ready-made dashboards and reports for common questions. They save a lot of time when they fit and waste time when they do not. The skill is recognizing which situation you are in before you start.

This guide explains what a good template contains, when a template is the right starting point, when you need a custom analysis instead, and how to move between the two.

What an analysis template is

A template is a reusable design for a business question. It usually specifies:

• Metrics: what to calculate, such as revenue, churn rate or conversion rate
• Dimensions: how to slice them, such as by month, channel, product or region
• Visualizations: which chart suits each metric
• Data requirements: which kind of source and which columns the metrics need
• Structure: which questions come first and how the answer is laid out

A good template captures how experienced analysts approach a common question. A poor one is a pretty layout with vague metrics.

When templates work well

Standard metrics with agreed definitions

Every subscription business tracks MRR, churn and customer acquisition cost. Every online shop watches conversion rate, average order value and repeat purchase rate. These have well-known definitions and presentations, so a template gets you a working view without debating what to include.

Recurring reports

A weekly sales summary, a monthly marketing review or a quarterly business review has the same shape every time with new data. A template removes the rebuild.

Getting a new team started

A template shows a team that is new to analytics which metrics matter for their function and what a useful layout looks like. It is a teaching tool as much as a shortcut.

Consistency across teams

When two teams report the same metric, building both from the same template keeps the definitions identical.

When templates fall short

Unusual business models

A marketplace that earns from both buyers and sellers, or a company that mixes subscriptions with one-off projects, will not find a template that models its revenue correctly. Forcing one produces numbers that look right and are not.

Investigations

"Why did churn among annual customers double last quarter?" is not a template question. You do not know in advance which dimension explains it, so you need to explore: split by plan, then by signup month, then by support volume, following what the data shows.

Your own segments and scores

If you segment customers by signals specific to your business, or use your own scoring, no generic template knows about them.

Questions that cross functions

"How does the marketing channel mix affect customer lifetime value by product line?" needs data from several systems combined in a way few templates anticipate.

Metrics still being defined

If the team is still deciding what to measure, a template locks in choices too early. Explore first; standardize later.

A quick decision guide

Ask four questions:

1. Would any analyst in your industry recognize the question? "What is our monthly churn rate?" or "What does our pipeline look like by stage?" Use a template.
2. Is it recurring or one-off? One-off investigations rarely justify building a template. If the same kind of investigation keeps coming back, turn your approach into one.
3. Who will use it? Templates help most when the person who needs the answer is not an analyst.
4. How standard is your data? A CRM with contacts, companies and deals, or a shop with products, orders and customers, fits templates well. Heavily customized data needs adjustment.

The practical approach: start from a template, then adjust

For most questions the fastest route is neither pure template nor blank page:

1. Start with the template closest to your question.
2. Check the fit: which metrics are relevant, which are missing, which your audience will ignore.
3. Adjust: remove what does not matter, add your own metrics, set the periods and filters.
4. When the result is something you will run again, save it as your own template.

You get the speed of the template and the fit of a custom analysis.

How this works in clariBI

clariBI has 378 templates, 320 for reports and 58 for dashboards, across 27 categories, from marketing and finance to industry sets such as healthcare, manufacturing and professional services. Templates are included in the trial and on the Starter, Professional and Enterprise plans. Our guide to using the template marketplace walks through finding one.

Each metric in a template carries hints about the columns it needs, and clariBI matches those to your connected sources. On the Reports page, templates that fit your sources show Recommended Data Sources and why. If your data does not contain what a template needs, the metric is shown as empty, or a report says Data Mismatch Detected and lists the fields it found, rather than filling the gap with a guess. The report templates article covers running one.

For custom analysis there are two routes:

• Ask clariBI for investigations: type the question in plain English, follow up, and save the conversation as a report when it is worth keeping. The numbers are computed from your data, and narrated figures that the chart cannot back are removed. The first question guide shows how sessions work, and 40 example questions help with phrasing.
• Custom metrics in the dashboard builder for recurring views: describe the metric in plain language, pick a chart type and let clariBI work out which columns to use. See the dashboard builder reference.

When a custom analysis becomes a regular report, turn it into a template with Create Custom Template on the Reports page's Templates tab. It stays within your organization unless you choose to publish it to the marketplace.

Building templates your team will reuse

• State the question the template answers and who it is for.
• List the data it needs: which kind of source and which columns, in which format.
• Add a line of guidance on how to read each metric.
• Test it on a second source of the same kind. A pipeline template should work on HubSpot data and on Pipedrive data.

Summary

Use a template when the question is standard, the metric is well defined and speed matters. Use custom analysis when the question is new, the investigation is open-ended or your data model is unusual. Most good analysis starts from a template and ends up slightly custom, and the best custom analyses end up as templates. The templates page lists the categories if you want to see what already exists for your field.

Try it

Pick the question your team asks most often and search the template library for it. If a template exists, run it on your data and note which metrics you would drop or add. If none exists, ask the question in Ask clariBI instead and save the answer as a report; when you have asked it three times, make it a template.