Rank Tracking and Index Estimates as Evidence

What a third-party estimate is, and why the word matters

A third-party estimate is a number about a website produced by a company other than the site's owner or the search engine, from data that company collected or modeled on its own. Rank-tracking exports (a tool's record of where a page appeared in search results for a chosen keyword, from a chosen location and device, at the moment the tool checked), backlink indexes (a vendor's crawl-built database of which pages link to which), search-volume figures (an estimate of how many times a keyword is searched per month) and traffic estimates (a modeled figure for how many visits a site receives) all belong to this class. None is a record of what Google did; each records what a vendor's system fetched, crawled or predicted.

The vendors say so themselves, in their help documentation rather than their marketing. Ahrefs states that "Search volume in any SEO tool, including Ahrefs, is always an estimation. Only Google has access to the exact data," and that its organic-traffic figure is "an estimate, not the actual search traffic data of a website." Similarweb describes itself as "an estimations tool." Semrush builds its traffic figure from a clickstream panel (a group of users whose browsing is recorded and extrapolated) of "over 200 million real, anonymized internet users" and a "Neural Network algorithm." The rules of evidence then ask two different questions of a report built on such a figure: whether an SEO practitioner may rely on it at all, and whether it is enough to carry this opinion.

Rule 703 and Rule 702(b) ask different questions

Rule 703 provides that an expert "may base an opinion on facts or data in the case that the expert has been made aware of or personally observed," and that "If experts in the particular field would reasonably rely on those kinds of facts or data in forming an opinion on the subject, they need not be admissible for the opinion to be admitted." SEO practitioners rely on this class of data daily, so Rule 703 is usually satisfied.

Clearing Rule 703 settles less than proponents assume. The 2000 Committee Note to Rule 702 calls the Rule 703 inquiry "relatively narrow" and states that "the question whether the expert is relying on a sufficient basis of information—whether admissible information or not—is governed by the requirements of Rule 702." Rule 702(b) requires that "the testimony is based on sufficient facts or data," and since December 1, 2023, the proponent must show that "it is more likely than not" that it is. The 2023 Note adds that treating sufficiency of basis as a question of weight alone is "an incorrect application of Rules 702 and 104(a)."

The two rules also treat the vendor's dashboard differently. Under Rule 703, when the facts or data "would otherwise be inadmissible, the proponent of the opinion may disclose them to the jury only if their probative value in helping the jury evaluate the opinion substantially outweighs their prejudicial effect," while nothing in the rule "restricts the presentation of underlying expert facts or data when offered by an adverse party." The estimate supports the opinion but does not come in through the expert as proof of what the traffic was, and the opponent remains free to read the vendor's caveats to the jury.

What the vendors say about search volume and traffic

Search volume is the figure most often carried into a damages model, because it converts a ranking position into an assumed number of clicks. Semrush defines it as "the average number of times a specific search query is entered on a search engine per month," a 12-month average built from "sets of data from third-party providers" overlaid with "historical clickstream data" and run through "machine learning algorithms," with no stated error margin. Moz calls its figure a "prediction."

Ahrefs is the only one of the three that publishes a self-assessment. It builds volume from "a combination of Google Keyword Planner (GKP), Google trends data and other third party data sources," warns that GKP "should not be used as a benchmark," and reports that its own estimates "were roughly accurate for 60% of the studied keywords." An expert who multiplies a lost ranking position by a volume figure and a click-through rate (the share of searchers who click a result at a given position) has stacked three estimates, and the vendor has already told the opponent that the middle one is right about six times in ten.

Traffic estimates stack further. Ahrefs sums, across every keyword a site ranks for in its database, a figure derived from "ranking position, monthly search volume, and our estimated CTR for that position." Similarweb displays no data for a site under 5,000 monthly desktop visits. Ahrefs also discloses that when it expanded its keyword database in December 2016 it chose to "give the benefit of a doubt" and drew "a horizontal line" of assumed traffic for the months before, a backfill that an expert who has not read the help center will present as history.

Backlink indexes are crawls, not censuses

A backlink (a hyperlink from one site to a page on another) is counted in these tools only when the vendor's crawler has found it. Moz says its "index is large but it doesn't cover the whole web" and that "Every index uses its own crawler to gather data and will build up a slightly different picture of the web." Ahrefs says less popular domains "will be partially crawled." Majestic keeps a Fresh Index of "backlinks we've seen in the past 120 days" and a Historic Index of "all backlinks we've seen since 2006," and notes that "If a link has not been crawled for some time, then it may drop out of the Fresh Index."

Two consequences follow for a negative-SEO or lost-links opinion. A "lost" link in a 120-day Fresh Index may have been removed, or the crawler may simply not have revisited the page; an expert who testifies that a site "lost N backlinks" has to say which. And a count that differs between two vendors is the expected result of two crawlers following different paths, not a contradiction, which is why the opinion should state which index, which date, and why.

Rank-tracking exports: location, device, timing and retention

A ranking position is not a fact about a page. It is a fact about one fetch of a results page, from one location, on one device type, at one moment. Semrush explains that a Position Tracking campaign fixes its location and device at creation and "updates your data every 24-48 hours," while "search results can change many times over a given day." Ahrefs says positions between ten and twenty "can easily differ by 5 positions each update." SE Ranking stores "a cached copy of the search engine results received while checking the rankings," the closest thing in this category to a contemporaneous capture and the exhibit to request in place of a bare position number.

Retention is where most rank-tracking evidence dies. Semrush keeps Position Tracking data daily "for 60 days, then weekly for up to 140 weeks" from the start of the campaign. After two months the daily series that would show the day a ranking fell no longer exists; after about three years the campaign history ends. A campaign that was not running before the events in dispute cannot be reconstructed later from the tool. In my experience the rank-tracking project that matters was almost never set up with litigation in mind, and the first question to ask of any export is when the campaign was created and in what configuration. That configuration is itself a fact to produce with the exhibit.

What courts have done with third-party data

No reported decision was located, as of November 4, 2026, in which Semrush, Ahrefs or Similarweb data was offered through an expert and challenged under Rule 702, Rule 703 or the hearsay rule. The older Alexa rankings were credited without objection in a default-judgment order and two precedential Trademark Trial and Appeal Board decisions; none evaluates the methodology.

The governing law is therefore the general law on experts who rely on someone else's figures. In TK-7 Corp. v. Estate of Barbouti, 993 F.2d 722 (10th Cir. 1993), a damages expert "clearly adopted the projections" of a distributor's market study without familiarity with the methods behind them; the Tenth Circuit held the reliance did not satisfy Rule 703. In Montgomery County v. Microvote Corp., 320 F.3d 440 (3d Cir. 2003), the opinion rested on a document of unknown authorship and on tapes a party's attorney had selected; the Third Circuit affirmed exclusion, because a well-qualified witness "still may be barred if it is not based on sound data." In Factory Mutual Insurance Co. v. Alon USA (5th Cir. 2013), by contrast, an appraiser's reliance on plant employees' estimates was upheld because he had interviewed the sources and applied his own expertise; the line the court drew was at an expert who becomes "the mouthpiece of the witnesses on whose statements or opinions the expert purports to base his opinion."

These cases describe an SEO expert's exposure precisely. One who pastes an Ahrefs traffic chart into a report and cannot say how the curve is built or why it diverges from the site's own analytics is the expert in TK-7. One who works from rank-tracking exports that counsel selected has walked into Microvote. One who reconciles the estimate against first-party data and states which number the opinion rests on is doing what Factory Mutual permits.

What an opinion built on an estimate has to show on its face

The 2000 Note to Rule 702 says the gatekeeping function "requires more than simply 'taking the expert's word for it,'" and the 2023 Note says the judge "should (where possible) receive an estimate of the known or potential rate of error of the methodology employed." A report that uses third-party estimates meets that standard only by being candid about what the estimate is. It names the vendor, the product and the export date. It reproduces the vendor's description of the method and any stated accuracy figure. It states the campaign configuration behind any rank-tracking data and the index and window behind any link count. It reconciles the estimate against first-party sources such as Google Search Console and the site's own analytics wherever they exist, and explains the difference rather than choosing the more favorable figure.

The weakest opinion in this discipline is a damages number whose only input is a vendor's traffic estimate: the vendor has disclaimed it, the first-party data was available and not used, and the opponent can read the help center aloud. The strongest use of the same estimate is as a yardstick, a trend line that corroborates a loss the site's own records already show, offered for direction rather than magnitude. Rule 703 lets the expert consider it either way. Rule 702(b) decides whether the opinion survives, and that decision belongs to the judge, by a preponderance, before the jury hears a word.

Frequently Asked Questions

Published 2026-11-04