A Digital Reality Check, start to finish

This is a made-up company, but the pattern is one almost every repositioned business will recognise. If you want to run this yourself, start with the guide to running a Digital Reality Check and the Canvas.

Meet Northwind

Northwind started life as a data-pipeline tool: it moved data from source systems into warehouses, quickly and cheaply. Last year the team made a real bet and rebuilt the product around data observability, catching data-quality incidents before they reach dashboards and executives. New product, new pricing, new buyer: they now sell to data-platform leads, not just data engineers.

Six months in, sales kept hearing the same thing on calls: "I thought you were just a pipeline tool." Deals were being compared against the wrong competitors. The team suspected the internet was still telling the old story, so they ran a Digital Reality Check to find out.

Step 0: what they wrote on the Canvas

Before touching AI or search, the team spent thirty minutes filling the Canvas. This is what they believed was true, on paper:

With the Canvas committed to paper, they had something concrete to compare the internet against. Then they ran the Check.

Try one question first

That single unbranded question was enough to make the room go quiet. Now for the full picture.

Check Step 1: Ask AI (no brand)

They asked an AI assistant the questions their buyers ask, without naming themselves:

This is an AI Representation problem. Whatever the model understood Northwind to be, it did not consider it a candidate when a buyer asked for observability options.

Check Step 2: Ask AI (with brand)

Now they put their name in, to test whether AI even understood what they were:

This is a Machine Understanding problem, and it explains the first one. The model was not being unfair; it was reading a company that still looked, on the public internet, like a pipeline tool. You cannot be recommended as the thing you are if the internet believes you are something else.

Check Step 3: Search like a buyer

They ran the same intent through a search engine and noted what surfaced:

• "Northwind": the homepage, and a two-year-old G2 profile filed under Data Integration.
• "Northwind alternatives": a comparison page that listed pipeline tools, not observability platforms.
• "best data observability tools": listicles featuring the three competitors. Northwind appeared on none of them.

Search confirmed where the AI answers were coming from: the public web still categorised Northwind by its old job.

Check Step 4: Read your public evidence (website, reviews, 3rd party sites)

Finally they read their own evidence coldly, the way a stranger or a model would:

• Homepage: the hero still led with "move data faster" — a pipeline message.
• Proof: case studies described faster loads and lower costs, not incidents caught.
• Third-party profiles: G2, Crunchbase, and two directories all listed the old category.
• Reviews: older reviews praised the pipeline speed; none mentioned observability.

Every source a model could read was telling the old story. The new positioning existed almost entirely inside the company.

Check Step 5: Compare to find the Gap

They put the two columns side by side. This is the moment the gap stops being a feeling and becomes a list:

The Canvas Readout

The team wrote a one-page Readout: what they found, and what to do. No score, just a shared, specific picture.

The 90-day plan

Two named gaps made the plan almost obvious. Northwind picked the three moves most likely to change what the internet says:

• Rebuild the homepage and product pages around observability. Lead with catching incidents, not moving data. Make the new category unmissable to a first-time reader.
• Correct the category everywhere third parties describe you. Update the G2 and Crunchbase profiles, the directory listings, and the comparison pages that still file Northwind under pipelines.
• Publish proof a model can cite. Ship two case studies and a short benchmark about incidents caught and time saved, written so an AI can quote them.

Notice that none of these are prompts or clever tricks. They are changes to the public evidence the models are reading. That is the only durable way to close the gap.

90 days later

After the changes had time to be crawled and re-indexed, Northwind ran the same Check again. The branded question now returned "a data observability platform," and the unbranded "best data observability platforms" began including Northwind in the longer lists. Still behind the incumbents, but no longer invisible. The gap had narrowed because the evidence had changed.

This is also where the free Check reaches its limit. Northwind could feel the improvement, but not size it, benchmark it against the three competitors, or watch it drift as the models updated. That is the job of measurement.

Run your own

You can run everything above yourself, for free, this week. Fill the Canvas, ask the five questions, and write your own Readout.

When you want to turn a directional read into a defensible number, a benchmark, and ongoing monitoring, that is where PageRadar comes in, with the Digital Reality Score and Radar.

← Back to the guide