Skip to main content
AI Search Value estimates the monthly dollar value of a brand’s presence in AI search answers: what it is worth to show up, and to show up prominently, when ChatGPT, Claude, Perplexity, and other AI systems answer questions in your market. The number is always returned as a range, never a single figure, and it is grounded in real Google keyword economics rather than an opaque score.
AI Search Value is rolling out in phases. Until it is enabled for your organization the endpoint returns 403, the same as any other gated resource. Contact your AthenaHQ representative to be added to the rollout.

​
What you get back

The response separates channel value (what the market is worth and how much of it you capture) from contribution (how much measured impact your content drives). These are never summed or conflated.
  • market — the addressable monthly value of your tracked topics (value_usd_mo), how many of those topics are priced (coverage_pct), and your value-weighted AI share of voice (current_sov_pct).
  • captured / headroom — value bands (conservative, expected, upside) for the value you already capture and the value still on the table.
  • measured_responses — how many AI answers (28 days) back the read. A zero here means the market has not been sampled yet, so the site is still “gathering”.
  • is_stale_value — true when the site was priced but has since dropped below the refresh threshold and the value is no longer being maintained (see When a value is stale). The value is still returned; treat it as a last-known figure, not a current one.
  • topics — per-topic detail: category, position-weighted share, mention rate, sample size, and each topic’s own captured / headroom bands. The counted_in_total flag marks which rows roll up into the headline totals (duplicate canonical markets are counted once).
  • merge_suggestions — pairs of near-duplicate topics worth merging.
  • contribution.modeled_attributed — the slice of your captured value attributable to content you have shipped, plus attributed_share and an attribution_unavailable flag that distinguishes “no attributed content” from “the attribution read failed”.
  • forecast — when an active forecast exists, its realized share delta versus the do-nothing baseline.

​
Quickstart

curl "https://api.athenahq.ai/api/v1/ai-search-value?website_id=$WEBSITE_ID" \
  -H "x-api-key: $ATHENA_API_KEY"

{
  "assumptions_version": "2026.06",
  "computed_at": "2026-06-20T07:14:02.000Z",
  "has_computed_valuations": true,
  "is_stale_value": false,
  "market": {
    "value_usd_mo": 105100,
    "topics_priced": 18,
    "topics_total": 20,
    "coverage_pct": 90.0,
    "current_sov_pct": 12.4
  },
  "captured": { "conservative": 45000, "expected": 58500, "upside": 90000 },
  "headroom": { "conservative": 60100, "expected": 46600, "upside": 15100 },
  "measured_responses": 2843,
  "topics": [
    {
      "topic_id": "9f1c…",
      "category": "market_segment",
      "local_intent": false,
      "share_basis": "pw_sov",
      "pw_sov_pct": 11.8,
      "mix_adjusted_share_pct": 12.4,
      "mention_rate_pct": 31.2,
      "drift_pp": 0.4,
      "standard_error_pp": 1.1,
      "responses_28d": 412,
      "models_measured": 4,
      "personas_measured": 3,
      "locations_measured": 2,
      "captured": { "conservative": 3200, "expected": 4100, "upside": 6300 },
      "headroom": { "conservative": 4800, "expected": 3900, "upside": 1700 },
      "counted_in_total": true
    }
  ],
  "merge_suggestions": [],
  "contribution": {
    "modeled_attributed": {
      "conservative": 5400,
      "expected": 7000,
      "upside": 10800
    },
    "attributed_share": 0.12,
    "attribution_unavailable": false
  },
  "forecast": null
}

​
Reading the number

  • It is a range, not a point. Report the band, not a single dollar figure. The spread reflects genuine uncertainty in how AI answer volume converts to value.
  • current_sov_pct reads lower than your Share of Voice chart, on purpose. The Share of Voice metric counts every brand mention equally across all topics, including your own branded queries where you naturally dominate. current_sov_pct is position-weighted, measured over your commercial (non-branded) topics only, and weighted by each topic’s market value, so the expensive topics where you actually compete drive it. Expect it to read lower. That is the metric working as intended, not a discrepancy.
  • Channel value and contribution are separate. market / captured / headroom size the opportunity; contribution measures the impact your shipped content drives. Do not add them together.

​
When a value is not available

A few states return 200 with no priced value:
  • Gathering data — computed_at is null or has_computed_valuations is false. The market has not accumulated enough AI answers for a reliable read yet; this usually fills in within a day or two of scanning.
  • No addressable market — has_computed_valuations is true but market.value_usd_mo is 0. None of your tracked topics map to priced search demand.
Paused websites are treated as deleted across the v1 surface and return 403, not a value.

​
When a value is stale

is_stale_value is true when a site that was previously priced has since dropped below the activity threshold (fewer AI answers in the trailing 28 days than the sweep requires) and its value is older than the refresh window. The daily sweep stops re-pricing the site, so the returned captured range is frozen at its last computed figure. The endpoint still returns the value; use is_stale_value and computed_at to decide whether to surface it as current or as a last-known estimate.

​
Rate limits

Standard per-organization API rate limits apply. Exceeding them returns 429 with Retry-After. See Rate limits.

​
Authentication

Standard AthenaHQ API key authentication. See Authentication.

​
Endpoint reference

See GET /api/v1/ai-search-value for the full query and response schema.