How to Track Top Actions and Response Performance Across Your Stores

Most review dashboards are very good at telling you what happened. Ratings went up. Sentiment is 72% positive. You received 43 reviews.

None of that tells a store manager what to change on Monday. And none of it tells you whether your team is actually keeping up with replies — which, for anyone paying for review management, is the number that justifies the spend.

The honest way to find recurring problems without tooling is to read everything and take notes:

• Export or scroll through the week's reviews per location.
• Tally the complaints that come up more than once.
• Check which negatives never got a reply.
• Work out roughly how long replies are taking.

At one location this is an hour. Across six it is most of a day, every week — and it is the first task to get skipped when the week gets busy.

Sentiment dashboards are a solved problem. Birdeye, Podium and most of the mid-market tools will all show you sentiment trends, rating movement and volume by location, and they will show it well.

Two gaps show up repeatedly:

• Themes are generated fresh each period. When the wording changes week to week — "slow service" one week, "long waits" the next — you cannot tell whether the underlying problem is improving. Comparability requires a fixed vocabulary, and free-form AI output does not have one.
• Your own responsiveness goes unmeasured. Platforms grade customer sentiment thoroughly and their own users barely at all. Response rate, time to reply, and the count of negative reviews still sitting unanswered are the numbers an agency needs to justify a retainer, and they are usually the ones missing.

There is a commercial reason too. Analytics is a common upgrade trigger: on Birdeye the tiers run from roughly $299 to $449 per location per month, and deeper reporting is one of the things that moves you up the ladder. For a franchise group, the reporting you need to manage the estate is priced against the size of the estate.

Inform ships Top Actions against a fixed theme set and Response Performance as standard, on every plan.

The summary pages now carry two sections built to answer those exact questions — Top Actions, which tells you what to fix, and Response Performance, which tells you how well you are replying. Both appear on individual location summaries and on the organisation-wide view.

Top Actions groups review text into recurring themes rather than free-form suggestions. Each card carries the theme, how many times it was mentioned, verbatim customer quotes, and a specific recommended action.

Because the themes come from a fixed set rather than being reinvented each week, a theme means the same thing every time it appears — which is what makes tracking progress possible.

Take the card above for stock and availability, flagged with three mentions. It carries two verbatim quotes — "doors were locked" and "they have stopped taking orders at 11:00 pm" — and then a concrete recommendation: track the items customers found unavailable and adjust par levels or update your listings.

The quotes matter as much as the count. A manager who disputes an abstract metric rarely disputes a direct quote from their own customer.

Underneath the action cards is a What's working row — the themes customers praised, each with a mention count. Product and food quality with twelve mentions, staff friendliness with ten, service speed with six.

It is not filler. Knowing which strengths customers actually notice tells you what to protect when you start changing things, and it is the part of the summary worth forwarding to the team.

The second section measures you rather than your customers. Three numbers, all deterministic — no AI involved:

• Response rate — what share of reviews received a reply, shown as a percentage and as a raw count like "56 of 61 reviews".
• Average reply time — roughly how long customers wait for a response.
• Unanswered negative — how many one and two star reviews are still sitting without a reply.

That last one is the number to watch. An unanswered negative review is the single most expensive thing on a listing, and it is the metric most dashboards never surface.

On the organisation view the same two sections aggregate across every connected location, with two additions that only make sense at scale.

Top Actions cards gain a trend — whether a theme is new compared with the previous period, or down from a higher count — and a location chip showing which stores the theme came from. Response Performance gains a by location breakdown, so you can see at a glance which branch is dragging the average down.

Both sections respond to a period selector next to the platform picker. Weekly is the default and stays the right lens for operations. The longer windows are for a different question: not "what happened this week" but "is this actually a pattern".

A complaint that appears once is noise. The same theme across a month is a process problem.

The organisation summary carries the same selector, and the headline figures relabel themselves to match the window you picked — so a monthly view reports the four-week average rather than silently showing weekly numbers under a monthly heading.

This is the view to use for a franchise review or a client check-in, where one week is too narrow to be representative.

A practical routine that takes about ten minutes:

1. Open the organisation summary and check Unanswered negative first. Clear anything sitting there.
2. Read the Top Actions cards and pick the one with the highest mention count that you can actually influence this week.
3. Check the location chips to see whether it is one branch or all of them — that decides whether it is a coaching conversation or a process change.
4. Switch to a longer period once a month to see which themes persisted and which were noise.
5. Send the relevant location summary to each manager.

The point is not to action everything. It is to action the one thing with the most evidence behind it.

They are drawn from your actual review text and grouped into a fixed set of themes — things like wait times, order accuracy, cleanliness and staff friendliness. Using a fixed set means a theme means the same thing week to week, so counts are comparable over time.

No. Response rate, average reply time and unanswered negative counts are calculated directly from your reviews and replies. They are deterministic figures, not estimates.

A review rated one or two stars that has no reply posted. It is shown separately because it is the highest-impact item on any listing.

Yes. On the organisation summary each Top Actions card is labelled with the locations that theme came from, and Response Performance includes a per-location breakdown.

No. The period selector aggregates summaries Inform has already produced for your locations — switching windows does not trigger new analysis or additional cost.

Analytics should end in a decision, not a chart:

• ✅ Actions, not just sentiment — every theme comes with a recommended next step.
• ✅ Evidence attached — verbatim customer quotes and mention counts, not vague scores.
• ✅ Your responsiveness measured — response rate, reply time and unanswered negatives.
• ✅ Location-level attribution — know which branch a problem belongs to.
• ✅ Weekly, monthly or long-range — separate a bad week from a real trend.

Related guides

• Learn how to set up an AI store insights dashboard.
• Or see how to gain franchise-wide insights.