~ / insights / templates

Template · September 2026

Migration Decision Record.

A one-page record that forces the question "what problem does the move solve" before "which platform", and ends in a signed decision with its evidence attached.

Amit TiwariTemplate9 min read

The decision this protects

Fill the sections in this order, and stop the moment one empties the case 1 The symptom, in the owner's words, not the vendor's 2 Evidence the platform causes it, and the test that would prove it 3 What else could cause the same symptom 4 What the platform genuinely cannot do, from its documentation 5 Cost in months, cost in risk 6 The decision, dated and signed, with a review date
Figure 1. Most records end at section 2 or 3, which is the point.

The decision is whether to spend three to six months and a developer’s fee moving a site to a different platform, and the record protects it by refusing to discuss platforms until the problem has been written down and tested. Most migration conversations I am brought into begin with the destination already chosen, usually WordPress, and confirming it is easy and often wrong, because a page not ranking or leads drying up has six or seven common causes and the platform is rarely the one with evidence behind it. Writing the record takes an afternoon. Moving the site takes a quarter.

Before you start

  • The owner’s description of the problem, in their words, taken down before anyone offers an explanation.
  • Search Console access to the current site, and every URL on it with its clicks over the last 12 months.
  • The current platform’s help documentation open in a tab, so every claim about what it cannot do can be checked.
  • The quotes already received for the move, in rupees, with what each includes and excludes.

Stated limits

  • The record does not choose a platform. It decides whether a platform change is the fix for a named symptom, and stops there.
  • It does not replace a migration plan. The redirect map, content freeze and launch checklist belong in a separate document.
  • It does not settle taste. If the owner wants to leave Wix because they dislike the editor, that is a legitimate reason but not a visibility reason, and the record should say so rather than dress it up as one.

1. The symptom in the owner’s words

Field What to record Why it matters
Symptom, verbatim The owner’s sentence as spoken, in quotation marks, uncorrected Paraphrase drifts towards the problem the consultant prefers
Date first noticed Month and year, and what prompted the noticing A start month can be matched against changes made that month
Metric it shows up in Clicks, impressions, position, leads, revenue, or a ranking for one query Each has a different list of causes, and “traffic is down” hides which one moved
Pages affected The URLs, or the section of the site A site-wide drop and a drop on five service pages need different tests

Judgment prompt: If the symptom cannot be tied to a metric and a page set, the honest next step is a month of measurement rather than a migration quote.

2. Evidence that the platform causes it

The four conditions a migration must pass You can name the specific thing you need to do The platform prevents it, per its documentation It cannot be worked around Doing it would move a number you care about ALL All four hold: proceed to sections 5 to 7. Any one fails: record why, close the record, revisit on the review date.
Figure 2. The gate the whole template exists to enforce.

Interactive: the four-condition gate. Needs JavaScript.

Field What to record Why it matters
Proposed mechanism The specific way the platform produces the symptom, in one sentence “Wix is bad for SEO” is a mood, and “Wix serves the service pages with a 1.8 second server response” is a mechanism
Test that would prove it The measurement, such as a Googlebot fetch and word count, a URL Inspection check or Crawl Stats response times against the drop date, with the figure expected either way written before the run A migration argued from a test that was never run is an opinion with a price tag
Actual result and date What the test showed, when, and who ran it The evidence the decision paragraph will cite

Judgment prompt: If nobody can name a test that would clear the platform, the platform is the scapegoat rather than the hypothesis, and section 3 is where the work is.

3. What else could cause the same symptom

Field What to record Why it matters
Alternative cause One row each for thin copy, missing internal links, a competitor’s improvement, a noindex or canonical mistake, a tracking change, seasonality, a Google update near the drop date, and lost backlinks The causes I find behind most “the platform is the problem” conversations
How it was checked The report, crawl or comparison used A cause dismissed without a check is still on the list
Result Ruled out, confirmed or unclear, with the figure The confirmed rows are what the owner should be paying to fix
Survives a migration Yes or no Thin copy moved to WordPress is thin copy on WordPress

Judgment prompt: If two or more alternative causes are confirmed and the section 2 test cleared the platform, the record should propose fixing those causes where the site is, and the migration question waits.

4. What the platform genuinely cannot do

Field What to record Why it matters
Capability One row per thing the platform is claimed to be unable to do, such as editing robots.txt, custom canonicals, custom URL structures for blog or product pages, or server log access The list has to be specific, or it becomes “less control” and cannot be checked
Documentation reference The help centre URL that says whether the capability exists, and the date you read it Platforms add capabilities every quarter, and a limitation remembered from 2021 may be gone
Needed for this symptom Yes or no, tied to the section 2 mechanism A missing capability unrelated to the symptom is a reason to be irritated, not a reason to move
Workaround What is possible on the current platform, and at what cost Some limits cost a fraction of a migration to work around

Judgment prompt: If every capability marked needed also has a documented workaround, the migration case has to rest on something other than capability, and the record should say what.

5. Cost in months

Field What to record Why it matters
Build and design Weeks quoted, and by whom The largest line and the one most often underquoted
Content transfer Weeks, and who checks each page Every page has to be read on the new site by someone who knew the old one
Redirect map and QA Weeks, and the number of URLs covered A redirect map for 400 URLs is a week of careful work, not an afternoon
Fees Quote, licences and hosting, in rupees per year The owner should see the running cost next to the build cost
Total elapsed time Weeks from decision to launch, and from launch to the point where rankings can be judged Google’s site move guidance says a medium-sized site can take a few weeks for most pages to move in its index, and larger sites take longer

Judgment prompt: If the total elapsed time is longer than fixing the confirmed causes from section 3 where the site is, the decision paragraph should say so.

6. Cost in risk

Field What to record Why it matters
URLs that will change Count, the share of last year’s clicks they carried, and who maintains the redirects after launch Every changed URL is a redirect that can be missed, and redirect maps rot when their author leaves
Expected settling period Weeks during which rankings may drop The owner signs for a dip rather than discovering one
Pages most at risk The ten URLs with the most clicks, and what protects each The pages to check by hand on launch day
Forms and consent Every form that collects personal data, and how its consent notice will be rebuilt Rebuilding forms is where DPDP obligations get silently dropped
Retraining and upkeep Days of training per editor, and who owns plugin updates, patches and backups A managed platform did this unasked, and a self-hosted one will not

Judgment prompt: If the ten pages most at risk carry more than half the site’s clicks and nobody has written down what protects each, this section is not finished.

7. The decision

Field What to record Why it matters
Decision Stay, move, or fix first and review One of three phrases, so it cannot be read two ways later
Date and signatures The day, and the owner and adviser by name Both names on the paper means both people read the evidence
Evidence attached The section 2 result, the section 3 table, the section 4 links The paragraph makes claims, and the attachments are what the claims point at
Review date Three to six months out, the metric to be checked on that date, and the event that would reopen the record sooner A “stay” decision is a bet that fixing the confirmed causes will move the metric, and the bet needs a settlement date

Judgment prompt: If the owner wants to sign “move” while section 2 shows the test was never run, write that fact into the paragraph above their signature.

The defensible output

What a completed record looks like (illustrative) FIELD FILLED IN Symptom Organic enquiries flat for 18 months while competitors grow Platform-caused? No: content coverage and entity consistency, both fixable in place Cost of migrating 3 to 5 months, redirect risk, team retraining Decision Stay. Fix content and entities. Reopen if a named limit appears. Signed, dated [name], [date], review [date + 2 quarters]
Figure 3. Half a page. It travels with the project and answers the question the next time it is asked.

One signed, dated paragraph with the evidence attached, naming the symptom as the owner said it, the test and its result, the causes confirmed instead, the decision and the review date. The paragraph below is a worked example with made-up figures.

On 12 August 2026, the symptom recorded was “our service pages don’t show up for anything”. The section 2 test fetched the twelve service pages as Googlebot and found all twelve indexed, rendered in full and served in under 600 milliseconds, which clears the platform. Section 3 confirmed two causes, an average of 140 words per service page and no internal links from the blog to any service page. We stay on Wix, rewrite the twelve pages to at least 600 words each by 30 September, add the links, and review clicks to those pages on 12 February 2027 against the August baseline of [VERIFY: figure] clicks per month. Signed, owner and adviser.

Where I could be wrong

The section 2 test can be run badly. A single fetch as Googlebot clears the platform for that page on that day, and a platform that serves slowly under load or fails to render one template in ten would pass and still be the cause. Where the symptom is intermittent, the test has to be repeated across a fortnight.

The alternative causes in section 3 reflect the sites I am asked about, small and mid-sized service businesses in India. A large e-commerce site with faceted navigation has causes I have not listed, and the platform’s crawl behaviour can genuinely be one of them. And a “stay” decision with a review date becomes a permanent stay if nobody checks the metric on the date, because the record cannot make anyone open it again.

Sources

How to cite this template

Amit Tiwari (2026). Migration Decision Record. Template, September 2026. amittiwari.net. https://amittiwari.net/templates/migration-decision-record

Send me the template

A filled example alongside the blank one, so you can see how it is meant to be used.

So I can look before we talk.

Discuss in the community ↗