Operational intelligence: a practical guide

Operational intelligence is knowing what is going wrong in your business today, while you can still fix it today. This article answers one question: what does operational intelligence look like in a company of 10 to 250 people, and how is it different from the monthly reports you already get? The short answer is that a report tells you what happened, and operational intelligence tells a named person what to do next. If your numbers arrive after the month closes, you are reading history. Restarting growth depends on acting on signals within hours, not weeks.

Why monthly reporting stops working once growth stalls

When revenue was climbing, a monthly review was enough. Problems were hidden by volume. A lost lead here or a late invoice there did not change the total, because more were coming in behind it.

Flat revenue removes that cushion. Every enquiry that waits two days for a reply, every proposal that sits unsent, every client who goes quiet before renewal now shows up in the bottom line. A monthly report will show you the damage. It will not show you the twelve individual moments where someone could have stepped in.

That is the gap operational intelligence closes. It moves the question from "how did we do?" to "what needs attention before close of business?"

Operational intelligence versus operational business intelligence

The two terms get used interchangeably, and the confusion costs money. Operational business intelligence is usually a reporting layer: dashboards that pull data from your systems and show trends by team, product or period. It is useful for planning. It is read by managers, often weekly.

Operational intelligence is narrower and faster. It watches events as they happen and turns specific conditions into specific actions. A new lead with no reply after 30 minutes is a condition. A task created for the sales lead is the action. Nobody has to open a dashboard for it to work.

You need both, but in a stalled business the second one usually matters more. Most firms with 10 to 250 staff already have more dashboards than anyone reads. What they lack is the wiring that puts the right exception in front of the right person at the right moment.

What most articles on this topic get wrong

Most writing on operational intelligence assumes you are a logistics company or a bank, streaming millions of events per hour into a specialist platform. The advice that follows is about data pipelines, event processing engines and real-time analytics stacks.

For a 60-person agency, consultancy, mortgage brokerage or estate agency, that framing is wrong. Your event volume is small. You have perhaps a few hundred meaningful events a day: enquiries, bookings, no-shows, proposals sent, invoices due, reviews received. The problem is not processing volume. The problem is that those events live in five different tools and nobody owns the exceptions.

So the first step is not buying an analytics product. It is getting the events that matter into one record, where a rule can watch them. On the levelpartner.ai platform, enquiries, conversations, bookings, opportunities and payments sit in the same CRM, which is what makes a rule like "no reply in 30 minutes" possible in the first place.

The six signals worth watching in a service business

You do not need hundreds of alerts. You need a short list of conditions that, when they fire, almost always mean money is about to be lost. These six cover most service firms.

Unanswered new enquiries

A lead that waits loses interest. Set a threshold that matches your market, such as 15 minutes in office hours, and alert the owner of the pipeline when it is breached. This is the single highest-value signal for most firms, and it is the one most often missing.

Opportunities with no next step

Every open deal should have a dated next action. A deal with no task attached is drifting. A daily list of these, sent to each salesperson, does more for close rates than any training session.

Booked calls that did not happen

No-shows are recoverable if someone reaches out the same day. They are mostly lost if the first follow-up is a week later. Flag every no-show to the person who owned the booking.

Proposals sent and not opened

If a proposal has not been viewed within a few days, the deal is at risk. That is the moment for a call, not a reminder email.

Clients gone quiet

A retained client who has not replied to anything for several weeks is a renewal risk. Track last meaningful contact per account and surface accounts that cross your threshold.

Overdue invoices

Cash flow is an operational signal. An invoice overdue by a week should create a task, not wait for the month-end debtor list.

A worked example

The numbers are illustrative. The pattern is not: the gain comes from catching leaks earlier, not from generating more demand.

How to build it without a data team

Start with one signal and one owner. Add a second only when the first is being acted on every time it fires. The order below works for most firms.

1. Pick the signal that costs you the most. For most firms, it is slow replies to new enquiries.
2. Name the person who acts when it fires. Not a team, a person.
3. Define the condition precisely: which pipeline, which hours, which threshold.
4. Build the rule as a workflow in the platform's Automation area, creating a task or notification for that person.
5. Review weekly for a month: did it fire, did someone act, did the outcome change?
6. Only then add the next signal.

This is slower than switching on twenty alerts at once. It is also the only version that survives. Alerts that nobody acts on train your team to ignore alerts, and that habit is very hard to reverse.

If you want the reporting layer as well, keep it to one page that your management team reads every week. Our guide to customer retention measures shows how to choose a short set of leading and lagging numbers and decide who sees each one, and the same approach works for operational measures.

The trade-off you are accepting

Operational intelligence makes problems visible to specific people, quickly. That is uncomfortable. A salesperson who used to have a quiet backlog of undated deals now gets a list every morning. A manager who used to discover lost renewals at quarter end now sees them coming.

You give up some autonomy and some comfort in exchange for fewer surprises. In teams where trust is low, this can feel like surveillance. The way through is to frame each signal as protecting revenue, not monitoring people, and to let the team help choose the thresholds. If the signal is wrong, change it. If the signal is right, act on it.

You also give up breadth. A short list of signals means some problems will still slip through. That is the right choice. Ten signals that nobody trusts are worth less than three that everyone acts on.

Where it fits with the rest of your operating model

Operational intelligence is the feedback loop for your processes. If your processes are undocumented, the signals will tell you where things break but not what should have happened instead. That is why firms that get value from it usually have their core procedures written down first. See a worked sample SOP if that work is not done.

It also feeds planning. Patterns in the signals, such as no-shows clustering on Mondays or slow replies on Fridays, are exactly the evidence you need when you set next quarter's priorities. Bring the month's signal log to the quarterly planning meeting and use it to choose the one or two processes to fix next. Two procedures that usually sit with the same operations lead are covered in the contract management SOP and the guide to IT operational efficiency.

Frequently asked questions

Do we need a separate operational intelligence tool?

Usually not, at 10 to 250 staff. If your leads, bookings, deals and invoices sit in one CRM with workflow rules, that system can watch the conditions and create the tasks. A separate tool makes sense when your volume or data sources outgrow what a CRM can hold.

How many alerts should each person receive?

As few as possible. A good target is a handful a day, each one requiring a clear action. If someone is receiving dozens, the thresholds are too loose or the signals overlap.

Who should own operational intelligence?

An operations lead or the managing director, with each individual signal owned by the person who acts on it. The overall owner reviews whether signals are firing and being acted on, and retires the ones that are not useful.

How long before we see a difference?

The behaviour change is visible within the first fortnight, because people start acting on items they previously missed. Revenue effects take longer to read, typically one full sales cycle, because you need to see deals through to close.

The journey this guide describes