A competitor analysis is quick to put together, but it dates just as quickly. You put your three biggest competitors side by side, fill a document with strengths and weaknesses, and a quarter later half of it is out of date. What's more, such an analysis usually looks at what competitors say about themselves: you scan their website, their brochures, their price list. But your customer reads that less and less. They simply ask an AI assistant which option is best for them, and get an answer that mentions one supplier and leaves out another.

Below, you will first find a model you can use straight away. Then we work it out for a fictional company, and finally we add the layer most analyses are still missing: how AI assistants describe you and your competitors.

What is a competitor analysis?

A competitor analysis is a structured comparison of an organisation with the parties that want to serve the same customer. It answers three questions: who are the competitors, how do they differ, and what does that mean for choices in positioning, offering and communication? The outcome should not be a report but a decision: where does the organisation invest more, where less, and what does it leave aside?

There are many variants, such as the SWOT analysis or Porter's five forces. The core is always the same. First you decide who counts, then on which points you compare, then how you rate yourself and the others on those points, and finally what you do about it.

The model in five steps

This model works well for a mid-sized organisation. You can work through it in half a day, and repeat it every quarter.

  • Step 1: decide who counts. Distinguish between direct competitors (same offering, same customer), indirect competitors (different offering, same problem) and substitutes (the customer does it themselves, or does nothing). Three to five names are enough.
  • Step 2: choose four to six criteria. Choose what your customer values, not what you think matters: offering, pricing structure, target group, positioning, visibility, reputation.
  • Step 3: gather evidence per criterion. Note for every finding where it comes from. Without a source, it remains an opinion, and you notice that as soon as someone on your team asks a follow-up question.
  • Step 4: rate and compare. Give each party a simple score per criterion, and add one sentence of explanation.
  • Step 5: draw three conclusions. Where are you clearly better, where clearly weaker, and where are you indistinguishable from the rest? That third point is often the most important, because there you give the customer no reason at all to choose you.

A worked example

Take a fictional company: Vlaams Plan, a software company that sells staff scheduling to healthcare institutions. It has two direct competitors, Zorgrooster and PlanPunt, and one substitute: Excel.

  • Step 1. The direct competitors are Zorgrooster and PlanPunt. The substitute is the spreadsheet many institutions already use.
  • Step 2. The criteria are offering, pricing structure, positioning, visibility and reputation.
  • Step 3. The team reads the websites, the pricing pages and the reviews, and notes the source of every finding.
  • Step 4. On a scale of 1 to 5, Vlaams Plan scores high on offering (4) and low on visibility (2). Zorgrooster is the exact opposite: a modest offering (3), but strong visibility (4).
  • Step 5. The conclusions: the product is stronger than its brand awareness suggests, the pricing structure barely differs from PlanPunt's, and Excel as a substitute is not addressed anywhere in the communication.

With that analysis, the team can make a choice on Monday: improve visibility first, and make the pricing structure stand out more clearly. Still, there is a gap. In this example, the visibility criterion was filled in on gut feeling. So how do you measure it?

The layer that comes on top: what AI assistants say

Many buyers no longer start their search with a search engine, but with an AI assistant. They ask, for example, which software is suitable for staff scheduling in healthcare. The answer names a few suppliers, describes each of them in a couple of sentences and sometimes gives sources. In effect, that answer is a competitor analysis someone else has made for your customer. And you don't automatically see what is in it.

That is why it pays to add a sixth criterion to the model: how are we and our competitors described in AI answers? From that, you can read four things:

  • whether you are mentioned for the questions your customer asks;
  • how often you are mentioned compared with your competitors;
  • how you are described: as first choice, as an alternative or not at all;
  • which sources the assistant cites to back that up.

In GripIQ, this comes together in two places. Under Brands, you record your own brand and your competitors. In Prompts, you write the questions a buyer asks an AI assistant, and with every scan GripIQ records what ChatGPT, Gemini, Claude and Google AI Mode answer. From those answers, the analysis extracts which brands are mentioned, with what sentiment and tone. That gives you Share of Voice: your brand's share of all mentions of tracked brands, in the answers for the selected period.

For Vlaams Plan, it would work like this. The team adds the three brands under Brands and adds Prompts such as "which software is suitable for staff scheduling in a healthcare institution?" and "what is the difference between Zorgrooster and PlanPunt?". After a scan, the team sees how often each brand is mentioned and what share that adds up to. That figure replaces the gut feeling behind the 2 for visibility. Whether Vlaams Plan comes out well or badly doesn't matter here. What matters is that the criterion now has a source.

GripIQ also shows the sources an answer cites. So you see which pages the assistant uses to describe your competitor, and whether your pages are among them. To have the same question answered per country, see How to run a market analysis per country with regions per prompt.

What this changes in your analysis

The classic model still stands. The new layer does change three things.

  • Your evidence stays current. An AI answer can read differently tomorrow than today. A quarterly analysis is enough for your offering, but not for your visibility. By scanning regularly, you see how that visibility develops, instead of a single snapshot.
  • You discover competitors you didn't expect. An assistant sometimes mentions parties that are not on your list. GripIQ shows them as discovered brands under Brands, so you can start tracking them and add them to your model.
  • The comparison becomes fairer. You no longer compare what parties say about themselves, but what the outside world, via the assistant, says about them.

That said, a competitor analysis is only one part of market insight. How to build up broader market information is covered in What is market intelligence, and how does it differ from business intelligence?.

What you can do now

  • Work through the five-step model today for three competitors, and note the source for every criterion.
  • Write down ten questions a buyer would ask an AI assistant, and put them to two assistants yourself. Note which brands are mentioned.
  • Add AI answers as a sixth criterion and schedule a repeat in three months.

Want to see this without tracking everything by hand? Book a demo and we'll show you Brands, Prompts and Share of Voice for your own market.