Competitor Pulse
Where you stand against each rival on share of voice, and which way the gap moved since the last run.
SentinOS reads what people said about your brand across Instagram, YouTube, X and Reddit, then files one dated run: what moved, how far, and how sure it is. Not a feed to monitor. A position to act on.
“Positive” and “negative” is two buckets for a conversation that has nine. Every post a run reads is placed on one ordinal scale, from the people who defend you to the people spreading something false. Hue carries the tone group; lightness carries the step inside it.
Positive · step 1 of 9
People defending the brand unprompted, usually to someone who criticised it. The scarcest and most valuable thing a run finds.
A run is not a dashboard you configure. It is a fixed set of readings, taken the same way every day, so this morning’s number means the same thing as last Tuesday’s.
Where you stand against each rival on share of voice, and which way the gap moved since the last run.
Exposure share per platform, with the sentiment and momentum of each. Shares sum to the whole, so a gain is always someone's loss.
What the conversation is actually about, each theme scored for health and watched for drift away from the ground it started on.
The cultural windows in play, whether they were on the calendar or surfaced out of the conversation itself.
What the run thinks you should look at first, split into alerts that need a decision and observations that need a note.
Whether the things you did last week landed, measured against the conversation rather than against the media plan.
A figure on its own is a number. The same figure against the six runs behind it is a direction. Every reading a run files is plotted against its own history, and every chart is one scale with no second axis to make an unrelated line look like it agrees.
Ours against each rival, run over run
Share per platform. The four sum to the whole
Where each theme is actually being discussed
| YouTube | X | |||
|---|---|---|---|---|
| Freshness | 12% | 41% | 28% | 19% |
| Price & value | 8% | 22% | 46% | 24% |
| Delivery | 6% | 14% | 38% | 42% |
| Protein & diet | 31% | 37% | 18% | 14% |
Volume means nothing until you know what window it fell in. Every moment a run files carries a class, and the class sets how to read everything else about it. The same nine classes apply to every brand; only which moments land in them differs.
Half of them are not on any calendar. A derived moment is found in the conversation itself — a regulator acting, a safety incident, a conversation being born — and classified by what it contains rather than by a date someone entered in advance.
Heightened sensitivity. Demand drops and quiet is expected, so amplifying here carries risk.
A public occasion that lifts demand. Quiet here is a missed activation, not a safe week.
A planned high-demand buying period. Spikes signal supply readiness before they signal sentiment.
A campaign, a launch, or a known external event, measured against its own plan.
A regulator acting: a notice, recall, suspension or seizure. Not people discussing one.
A safety or harm incident. One evidenced post is enough; severity rides the evidence, not the volume.
Verifiable good news: a certification, lab report, award or audit passed.
A conversation being born on a theme under 14 days old. Never a storm, whatever its tone.
A real flare matching none of the above. Context more than action.
Every figure on every surface opens a thread. Counsel answers from the run that produced the number, shows the stages it went through to get there, and offers the next three questions worth asking.
It says when it cannot answer. An analyst that always has an answer is an analyst you cannot use.
Which competitor gained the most?
Thought for 19s
Croft & Co gained the most, and it came out of your share. Their advocacy recovered off a low base while your critiques held steady on price. The gap narrowed 1.1 points across the last two runs, all of it on X.
Main contributors
What drove the price critiques?Top examples
Show the posts behind the shift on XAcross
Share of voice trend over timeNoticing is the easy half. Most tools hand you a chart and leave the doing to a spreadsheet and a thread, where it stops being measurable. Here the run opens work: something gets flagged, someone picks it up, and what happened to it stays on the record.
The run ranks what it found. Alerts need a decision; observations need a note. Each carries the evidence it was raised from.
Not advice in a report. A recommendation enters the system with a target, a severity band and the run that produced it.
It becomes work someone holds: pending, active, awaiting approval, executed. Approvals are recorded against a role, not a name in a thread.
Every transition is kept — approvals, rejections, bypasses, SLA breaches. Including the ones that went wrong.
Answer the price critiques on X before the weekend peak.
Every state a step can reach
The last two are the ones that matter. A system that can only record success is not recording anything.
Runs are dated and kept. Every number carries the run it came from, so a claim made in a board deck in March can still be checked in June.
Yesterday's conversation is collected across every connected platform and matched to the themes already being tracked.
Each post is placed on the intent scale, themes are scored, moments are classified, and every figure is compared against the run before it.
Alerts and observations are written with the evidence attached, ranked so the thing that needs a decision is at the top.
Counsel answers against that run and no other, so two people reading the same screen get the same answer.
We connect your platforms, read back thirty days, and walk you through the first verdict with your team. It takes about a week.
Book a walkthrough