Google Can Read Flash Now. Your Flash Site Is Still a Problem.

Google announced at the end of June that it can now index Flash. Using technology supplied by Adobe, its crawler extracts text and links from inside SWF files. The announcement has traveled through the industry as absolution.

We have already fielded the celebratory version from two prospects with all-Flash sites: see, the invisibility problem is solved. We would like to gently spoil the party. The fine print matters more than the headline, and the decisions people make on this news will outlive it.

What the announcement actually says

Read carefully, Google’s own description is modest. The crawler can now discover text that appears inside Flash files and follow links embedded in them. Previously a SWF was an opaque box. The crawler saw the file existed and nothing more.

That is a real improvement for a specific failure. The restaurant menu, the program description, the biography locked inside a Flash movie is no longer literally invisible to search. Adobe’s cooperation, providing a specialized player that lets the crawler ‘operate’ Flash content, suggests both companies understood how much of the web had disappeared into the format.

Notice what the announcement does not say. It does not say Flash sites will now rank the way equivalent HTML sites rank. It does not say the extracted text carries the structural signals, like headings and emphasis, that ranking systems weigh. And it says nothing about the problems that were never about indexing at all. Those are where the party trips.

The problems indexing can’t touch

• One URL, many contents. The deep problem with the all-Flash site was never just invisible text. It is that the entire experience lives at a single address. Your programs, your people, your fifty pages of content: one URL. Search ranking happens per page, per URL. When everything is one URL, there is nothing granular to rank, nothing specific to link to, nothing for a visitor to bookmark or email. Google extracting your text does not create the fifty addressable pages that fifty pieces of content deserve.

• Extraction isn’t structure. An HTML page tells the crawler what is a heading, what is navigation, what is emphasized, what the page is about. Text pulled from a SWF arrives as, roughly, words. Present, but stripped of the architecture that makes relevance legible. Being indexed is the qualifying round, not the competition.

• Everything else on the list. Accessibility: screen-reader support for Flash content remains poor in practice, and the extracted-text improvement does nothing for actual visitors using assistive technology. Mobile: the iPhone renders Flash as empty space, a fact widely documented last summer, and Apple shows no sign of revisiting it. Analytics: measuring behavior inside a Flash experience remains a custom project, so the all-Flash site stays a fog on the reports. None of these were indexing problems. None of them moved this month.

What to actually do, by situation

If you own an all-Flash site: this announcement upgrades you from invisible to faintly visible, and that is genuinely better. But the strategic conclusion is unchanged, and we would rather you hear it plainly than discover it in next year’s traffic reports. The renovation path is an HTML site with real pages at real URLs, with Flash retained where it merits its place. Use this news as breathing room, not as a reprieve.

If you are commissioning new work and a vendor proposes Flash-first architecture ‘now that Google indexes it’: that pitch is this month’s test of vendor judgment, and it is failing. The indexing improvement narrows one gap in a format whose other gaps are structural. Ask the vendor how deep links, the iPhone, and your analytics will work. Enjoy the silence.

If you use Flash as a component embedded in an HTML page, carry on with our blessing. A video player, a data visualization, an interactive feature: all fine. This has been our line since the format wars began. Flash earns its keep inside a site, doing the things HTML genuinely cannot yet. The page’s text, structure, and address still live in markup. The component pattern gets all of Flash’s strengths and inherits none of the architectural penalties.

Auditing what Google actually sees of you

Before any strategy conversation, run the ten-minute audit. It converts this from abstract debate to specific findings.

A site: search on your domain shows what Google has indexed. For an all-Flash site, count the pages. A number near one is the architecture problem made visible. Google’s cached view of any result, in its text-only mode, shows roughly what the crawler extracted: your content as the engine reads it, minus the design.

If the cache of your homepage is a sentence and a copyright line, you have just seen your search presence’s raw material. Compare against an HTML competitor’s cache and the gap explains itself to any stakeholder in the room. We run this exercise in meetings precisely because it ends arguments that slides never could. Nothing persuades like watching the engine describe your site back to you.

For the component-pattern sites doing it right, one implementation blessing worth naming. The established embedding approach, the SWFObject pattern the development community has converged on, places real HTML content in the page that the Flash replaces when available. Done properly, search engines and script-less visitors get the genuine text while capable browsers get the rich version. That is progressive enhancement, applied to plugins.

If your vendor embedded your Flash without fallback content, that is an afternoon’s retrofit with outsized indexing return. It is also a fair test of whether they knew the pattern existed.

The pattern behind the announcement

One more reading, for the strategy file. Google keeps investing in reading through formats to the content underneath. First the sitemap protocols that let sites describe themselves, now machinery for opening opaque files. The direction is consistent: the engine wants the world’s content regardless of wrapper, and it will keep building can-openers.

But every can-opener arrives late and partial. The sites that never sealed their content in cans keep collecting the compounding advantage. Fully structured, fully addressable, fully measurable. On every device, for every visitor, indexed completely on day one.

So: nice news. The text in the box is visible now. The box is still the problem.